Pyridoxamine-5-phosphate enzyme-linked immune mass spectrometric assay substrate for linear absolute quantification of alkaline phosphatase to the yoctomole range applied to prostate specific antigen

Article abstract of DOI:10.1021/ac502572a

There is a need to measure proteins that are present in concentrations below the detection limits of existing colorimetric approaches with enzyme-linked immunoabsorbent assays (ELISA). The powerful enzyme alkaline phosphatase conjugated to the highly specific bacterial protein streptavidin binds to biotinylated macromolecules like proteins, antibodies, or other ligands and receptors with a high affinity. The binding of the biotinylated detection antibody, with resulting amplification of the signal by the catalytic production of reporter molecules, is key to the sensitivity of ELISA. The specificity and amplification of the signal by the enzyme alkaline phosphatase in ELISA together with the sensitivity of liquid chromatography electrospray ionization and mass spectrometry (LC-ESI-MS) to detect femtomole to picomole amounts of reporter molecules results in an ultrasensitive enzyme-linked immune mass spectrometric assay (ELIMSA). The novel ELIMSA substrate pyridoxamine-5-phosphate (PA5P) is cleaved by the enzyme alkaline phosphatase to yield the basic and hydrophilic product pyridoxamine (PA) that elutes rapidly with symmetrical peaks and a flat baseline. Pyridoxamine (PA) and ¹³C PA were both observed to show a linear relationship between log ion intensity and quantity from picomole to femtomole amounts by liquid chromatography-electrospray ionization and mass spectrometry. Four independent methods, (i) internal ¹³C isotope PA dilution curves, (ii) internal ¹³C isotope one-point calibration, (iii) external PA standard curve, and (iv) external ¹³C PA standard curve, all agreed within 1 digit in the same order of magnitude on the linear quantification of PA. Hence, a mass spectrometer can be used to robustly detect 526 ymol of the alkaline phosphatase streptavidin probe and accurately quantify zeptomole amounts of PSA against log linear absolute standard by micro electrospray on a simple ion trap.

Full text of DOI:10.1021/ac502572a

Article
pubs.acs.org/ac
Pyridoxamine-5-phosphate Enzyme-Linked Immune Mass
Spectrometric Assay Substrate for Linear Absolute Quantification of
Alkaline Phosphatase to the Yoctomole Range Applied to Prostate
Specific Antigen
Angelique Florentinus-Mefailoski and John G. Marshall*
Department of Chemistry and Biology, Ryerson University, 350 Victoria Street, Toronto, Ontario, Canada, M5B 2K3
S
* Supporting Information
ABSTRACT: There is a need to measure proteins that are present in concentrations
below the detection limits of existing colorimetric approaches with enzyme-linked
immunoabsorbent assays (ELISA). The powerful enzyme alkaline phosphatase
conjugated to the highly specific bacterial protein streptavidin binds to biotinylated
macromolecules like proteins, antibodies, or other ligands and receptors with a high
affinity. The binding of the biotinylated detection antibody, with resulting amplification of
the signal by the catalytic production of reporter molecules, is key to the sensitivity of
ELISA. The specificity and amplification of the signal by the enzyme alkaline phosphatase
in ELISA together with the sensitivity of liquid chromatography electrospray ionization
and mass spectrometry (LC−ESI-MS) to detect femtomole to picomole amounts of
reporter molecules results in an ultrasensitive enzyme-linked immune mass spectrometric
assay (ELIMSA). The novel ELIMSA substrate pyridoxamine-5-phosphate (PA5P) is
cleaved by the enzyme alkaline phosphatase to yield the basic and hydrophilic product
pyridoxamine (PA) that elutes rapidly with symmetrical peaks and a flat baseline.
Pyridoxamine (PA) and 13C PA were both observed to show a linear relationship between
log ion intensity and quantity from picomole to femtomole amounts by liquid
chromatography−electrospray ionization and mass spectrometry. Four independent
methods, (i) internal 13C isotope PA dilution curves, (ii) internal 13C isotope one-point calibration, (iii) external PA standard
curve, and (iv) external 13C PA standard curve, all agreed within 1 digit in the same order of magnitude on the linear
quantification of PA. Hence, a mass spectrometer can be used to robustly detect 526 ymol of the alkaline phosphatase
streptavidin probe and accurately quantify zeptomole amounts of PSA against log linear absolute standard by micro electrospray
on a simple ion trap.
ionizable product by alkaline phosphatase for ion mobility
spectrometry.8 Moreover, mass spectrometry has been used to
confirm the nature of the products from an ELISA reaction.9
Hence, there is good reason to believe it should be possible to
implement ELIMSA using the substrate PA5P that releases
pyridoxamine.10
Prostate Specific Antigen (PSA) Model System. PSA is
not selective for malignant disease but is a tissue-specific
kallikrein protease that indicates benign and cancerous
proliferation of prostate tissue and cells.11 The prostate specific
antigen (PSA) has a concentration distribution from picogram
to nanogram amounts in 100 μL of normal human plasma
(NHP) from men, and so like many cytokines and regulatory
factors it is often near or below UV−vis detection in normal
patients.12 The quantification of PSA over the complete range
of NHP is challenging by ELISA12 or mass spectral methods,2
INTRODUCTION
■
Enzyme-Linked Immune Mass Spectrometric Assay
(ELIMSA) for Low-Abundance Analytes. Human blood
contains low-abundance proteins of biomedical interest at
concentrations often below the detection limits of either
enzyme-linked immunoabsorbent assays (ELISA)1 or mass
spectrometry2 when used separately. Combining the enzymatic
amplification of ELISA with the detection of the enzymatic
products by liquid chromatography, electrospray ionization and
mass spectrometry (LC−ESI-MS) results in the ultrasensitive
quantification method termed ELIMSA,3 potentially allowing
the routine analysis of femtogram amounts of low-abundance
proteins in solution. The enzymatic amplification is obtained
from alkaline phosphatase (AP), a near perfect enzyme4 that
may be attached to the bacterial protein streptavidin (SA).
Streptavidin conjugated5 to the reporter enzyme (AP-SA) may
be used to detect biotinylated antibodies, ligands, or probes
with a high affinity.6 Enzymatic reactions have been previously
monitored by mass spectrometry,7 and the substrate pyridox-
amine-5-phosphate (PA5P) has been used to generate an
Received: July 12, 2014
Accepted: September 26, 2014
Published: September 26, 2014
© 2014 American Chemical Society
10684
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691

Analytical Chemistry
Article
and many normal samples are below the detection limit of the
assay. Detection of an increase in baseline PSA over time in
normal or prostatectomy patients may be of clinical importance
in some cases. Many other tissue or cell specific proteins,
regulatory factors, and cytokines remain to reliably assay in
blood.13 There are other good experimental reasons to consider
PSA as a test protein for developing more sensitive assays such
as ELIMSA: the analyte PSA and the enzyme AP-SA have been
used to benchmark the sensitivities of ELISA,14 direct mass
spectral detection,2 and electrochemical detection15 as well as a
previous ELIMSA substrate.3 The PSA ELISA using the high
affinity reagents described here is an excellent model system for
demonstrating the principles of ELIMSA and testing substrates.
Linearity and Accuracy of LC−ESI-MS. It has been
previously shown that an ELISA with the colorimetric enzyme
substrate amplex red, and naphthol AS-MX phosphate, can be
measured by LC−ESI-MS with linearity similar to that of UV−
vis based detection.3 Here, the same specific PSA calibrant,
capture, and detection reagents used previously were tested
with the novel ELIMSA substrate PA5P.3 It remains to be
determined if the enzymatic product PA shows a predictable
relationship between intensity and quantity in the context of
the immune assay. A linear relationship between log ion
intensity and quantity was observed for some blood peptides or
fragment ions, peptide standards, resorufin, naphthol AS-MX,
and in the present experiment pyridoxamine (PA) that were all
shown to be linear after simple log transformation.3,16 Internal
isotopic labeled standards may be used to test the assumptions
about the linearity and accuracy of the LC−ESI-MS assay.
Advantages of PA5P Substrate. In this paper, we show
that the substrate pyridoxamine-5-phosphate (PA5P) has
several important practical advantages over the previous
ELIMSA substrate naphthol AS-MX phosphate in that the
enzyme product pyridoxamine (PA) is basic, ionizes readily but
is also hydrophilic and elutes rapidly with a more symmetrical
peak. Most importantly, PA5P shows a more predictable
chromatography, is not strongly retained, and does not show a
creeping increase in baseline over the course of measurement.
The use of PA5P has an important additional advantage in that
the enzyme product PA, sometimes termed vitamin B6, is a
common metabolite that is widely available in 13C isotope
labeled form that can be used as an internal standard to confirm
the absolute quantification of PA for ELIMSA. Thus, to address
the issue of transformation, linearity, and absolute quantifica-
tion in ELIMSA by LC−ESI-MS, we controlled the ELIMSA
experiments with internal, and external, 13C labeled pyridox-
amine (13C PA) standards to confirm the quantification of the
enzyme product. If mass spectrometry is log linear and normal
for PA, then instead of using UV−vis spectroscopy to read the
96 well ELISA plate, the production of the colorless PA could
be measured using LC−ESI-MS with a sensitive ion trap.3 The
capacity to measure zeptomole amounts of blood proteins may
have great application to biomedical research and permits the
common analysis of many ultra low abundance analytes.
The alkaline phosphatase-streptavidin (AP-SA) enzyme-probe
conjugate was obtained from Jackson ImmunoResearch
Laboratories (West Grove, PA). The liquid chromatography−
mass spectrometry grade solvents were obtained from Caledon
Laboratories (Georgetown, Ontario, Canada).
Immunoassay. The PSA ELIMSA was performed as
previously described but with the substitution of the substrate
PA5P instead of naphthol AS-MX phosphate.3 The PA5P
substrate, PSA standards and AP-SA probe were dissolved in
water on ice, aliquoted, and freeze-dried for day-to-day
consistency in the assay. The plates were coated with the
capture antibody in bicarbonate coating buffer prior to blocking
the plates in 1% albumin and 1% goat serum in 1× PBS pH 7.4
followed by incubating the plasma in 1× PBS, 0.3 M NaCl pH
7.4 for 2 h. After the wells were washed again three times with
washing buffer, the substrate was reacted with the streptavidin−
AP in substrate buffer (20 mM Tris pH 8.8). The experimental
assay was first optimized and tested using large amounts of PSA
with colorimetric reagents to ensure there is no background
binding and to confirm the antibodies and calibrants are
working and linear in color assays before switching to ELIMSA
to measure the lower standards and concentrations for normal
human plasma (NHP).
Liquid Chromatography−Electrospray Ionization-
Mass Spectrometry. The LC-ESI-MS analysis was performed
by manual injection (Rheodyne 1755) of 2 μL of the standard
or sample diluted 20-fold in 0.1% formic acid. The sample was
analyzed via an Agilent 1100 high-pressure liquid chromatog-
raphy (HPLC) pump set at 70% acetonitrile in 0.1% acetic acid
for isocratic separation over a 15 cm normal phase column (300
μm i.d., 5 μm particle diameter, 300 Å pore) at 20 μL/min. We
compared normal phase and C18 reversed phase columns.
Although a C18 gradient provided sharper peaks, it was not as
convenient as the rapid isocratic normal phase column. The
normal phase chromatogram produced symmetrical peaks that
were resolved from the buffer components with low back
pressure and high flow rates allowing rapid analysis with
baseline to baseline separation over a stable background. The
HPLC was coupled via an electrospray source fitted with a
metal needle17 mounted on a linear quadruple ion trap (LTQ
XL, Thermo Electron Corporation)18 as previously described.3
The PA5P [M + H] product ion was monitored by intensity at
169 m/z as previously described3 with the injection of 70% IPA
via the sample loop to rapidly clean the column between
injections. The instrument was tuned and the correct
calibration of the product m/z assured using the manufacturer’s
tuning solution and the PA product. The intensity of the
pyridoxamine (PA) was monitored by selected ion monitoring
(SIM) of the window 168−170 m/z where the intensity of the
[M + H] ion at 169 m/z was extracted and plotted. The
intensity of the 13C PA standard was monitored by the SIM
window 171−173 m/z where the intensity of the [M + H] ion
at 172 m/z was extracted and plotted. The results were log-
transformed, and the ion intensity results were analyzed for
normality and linearity16b using the open source R statistical
analysis system.
MATERIALS AND METHODS
■
Materials. The NHS-biotin coupling reagent was obtained
from Pierce. The PSA capture and detection antibodies were
purchased from Medix Biochemica (Kauniainen, Finland). The
PSA calibration antigen was obtained from the Scripps
Laboratory (San Diego, CA). The PA5P, pyridoxamine, PA,
and 13C (PA), and all other buffers and salts were obtained
from Sigma-Aldrich and were of the highest quality available.
RESULTS
■
Pyridoxamine Has Excellent Chromatography Char-
acteristics. Previously, the alkaline phosphatase (AP) enzyme
system was more sensitive than the HRP system that requires
the strong oxidizing agent H2O2 as a cofactor,3 and so the
development of the AP based ELIMSA was continued by
10685
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691

Analytical Chemistry
Article
Figure 1. Comparison of the PA and 13C PA external standards by
LC−ESI-MS. Panels: (A) the isocratic chromatogram showing the
dilution series of PA [M + H] monitored at SIM 169 m/z (inset shows
linearity after log transformation); (B) the isocratic chromatogram
showing dilution series of 13C PA monitored at SIM 172 [M + H] m/z
(inset shows agreement between natural and isotope-labeled 13C-
pyridoxamine). The isocratic chromatograms show the injection of 2
μL of the standard in nanomolar as indicated by the arrows in 0.1% FA
on normal phase in 70% acetonitrile connected to an electrospray
source.
Figure 2. Enzymatic production of pyridoxamine (PA) from the
substrate pyridoxamine-5-phosphate (PA5P) by alkaline phosphatase
conjugated to streptavidin (AP-SA). Panels: (A) normal phase
isocratic chromatogram of the time course of PA production by the
AP-SA conjugate (5 ng) added to 1 mL of 1 mM PA5P in 20 mM tris
pH 8.8 as measured using LC−ESI-MS with SIM of pyridoxamine
(PA) at 169 m/z [M + H] where the arrow shows blank (b), no
enzyme N.E., or the enzyme reaction time in minutes (inset shows the
plot of log 169 [M + H] intensity versus time); (B) AP-SA conjugate
amount shown by the arrow in nanograms was added to 1 mL of 1
mM PA5P in 20 mM tris pH 8.8 (inset shows the log linear
relationship between log PA ion intensity when the amounts of AP-SA
in log ng/mL were added to the reaction). The production of PA was
sampled by removing 10 μL of the reaction that was mixed with 190
μL of 0.1% formic acid and 2 μL of the diluted sample was manually
injected. Blank injection (b) and no enzyme (N.E.) reaction injections
are shown where indicated.
looking for more effective substrates. The use of the alkaline
phosphatase substrate PA5P was found to be a strong
improvement on the previous substrate naphthol AS-MX
phosphate in that the latter’s hydrophobic naphthol substituent
causes distinct tailing in the chromatographic peaks with the
result that baseline to baseline separation is not always obtained
at convenient injection intervals (≤5 min.) significantly
increasing analysis time. In contrast, the basic and hydrophilic
product pyridoxamine (PA) ionizes readily, shows much more
symmetrical peaks and easily achieves baseline to baseline
separation even at close injection intervals (see the Supporting
Information).
Pyridoxamine Ion Intensity Is Log Linear with
Quantity. The ELIMSA assay for PSA depends on the
production of PA that is measured by LC−ESI-MS and so it is
necessary to demonstrate that the relationship between PA
quantity and signal intensity is log linear and predictable. A
simple dilution curve of pyridoxamine (PA) in ionization buffer
reveals a log linear relationship between intensity and quantity
when injected into the isocratic LC−ESI-MS system. The PA
[M + H] product ion showed a log linear relationship between
ion intensity versus absolute quantity as measured using LC−
ESI-MS with single ion monitoring (SIM) at 169 m/z (Figure
1A). Similarly, the 13C PA standard parent ion [M + H] at SIM
172 m/z also showed a log linear relationship between ion
intensity and quantity (Figure 1B). However, since electrospray
ionization is a competitive process, the linearity of PA in pure
ionization buffer may not be the same as the ionization of PA in
the ELIMSA reaction where there may be some complex
confounding effects of the tris buffer, blocking agents, nonionic
detergents, or other buffer components.
Enzyme Dependent Release of Pyridoxamine by
Alkaline Phosphatase-Streptavidin. The ELIMSA assay
depends on the amplification of the signal from the AP-SA
probe by the enzymatic activity of alkaline phosphatase. If the
PA signal is generated by the AP-SA probe then it should be
dependent on enzyme incubation time and the enzyme
concentration. Adding the biotin-specific molecular probe
streptavidin, conjugated to the enzyme AP-SA, to the substrate
10686
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691

Analytical Chemistry
Article
Figure 3. Illustration of the experimental setup to measure ELIMSA with absolute external PSA standards by the pyridoxamine (PA) production
from the AP-SA probe alongside internal isotope dilution, internal one-point calibration, and external PA and 13C PA standard curves. Panels: (A)
internal 13C isotope dilution curve for 12 NHP samples (left axis) that showed an average of ∼2.01 0.0377 pmol of PA injected from each replicate
well (right axis); (B) isocratic chromatogram showing the ELIMSA with an external PSA standard curve that was linear from 0.1 to 10 ng per well
with 1.5 pmol 13C internal standards, followed by 10 NHP samples with 1.5 pmol 13C internal standards, and finally PA and 13C PA external
standards as shown; (C) main panel shows the average PSA standard intensity values over 3 separate days with an R2 of 0.99 (insets: upper left, the
external pyridoxamine (PA) standard curve with SIM 169 [m/z] and the 13C external PA curve SIM 172 m/z; lower right, the intensity match
between the NHP PSA ELIMSA at SIM 169 and 1.5 pmol of the 13C pyridoxamine internal standard that showed a nearly equal intensity match
between the NHP PSA ELISA at SIM 169 and 1.5 pmol of 13C PA internal one-point calibration standard).
PA5P in tris reaction buffer released the product PA and the
biochemical reaction was essentially complete by 3 h (Figure
2A). The time course of the AP-SA signal closely matched an
expected enzyme activity with an initial nearly linear period of
rapid production followed by an asymptotic decrease in slope as
the reaction approached completion. The AP-streptavidin
conjugate (AP-SA ∼190 000 g per mole) was diluted down
to 0.1 ng per mL and then the enzyme was reacted with the
substrate PA5P for 3 h to produce PA. After diluting the
reaction 20-fold and injecting 2 μL, i.e., the equivalent of 0.1 μL
of the reaction, about 50 zmol of AP-SA on column was
detected. The enzyme dependent production of PA was clearly
detectable with a high signal-to-noise ratio and low error in the
baseline (Figure 2B). The pattern with time and concentration
both seemed to indicate that the signal measured at 169 m/z
was PA [M + H] produced by the enzymatic dephosphor-
ylation of the substrate PA5P by the AP-SA probe.
Estimate of PA Production from ELIMSA by Internal
13C PA Dilution. The intensity values of the LC−ESI-MS
assay for PA remained linear in the context of the ELIMSA
biochemical reaction conditions (Figure 3), and the internal
isotope dilution method was used to estimate the quantity of
internal 13C labeled PA that matched the average production of
PA from the ELIMSA. The internal isotope dilution method
was used to estimate the average production of PA from the
same replicate NHP sample (Table 1) and was found to be
about 2.01 pmol of PA measured against the internal 13C PA
internal isotope dilution curve (Figure 3A, Table 1).
10687
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691

Analytical Chemistry
Article
Table 1. Agreement between Internal 13C PA Isotope
Dilution, Internal One-Point 13C PA Calibration, External
PA Standard Curve and External 13C PA Standard Curve,
Alongside the Results of the External PSA Standard Curve
ion [M + H] monitored at SIM 169 m/z against the external
standard 13C PA monitored at 172 m/z showed 1.58 and 2.19
of pmol of PA on column in agreement with internal isotope
dilution and internal one-point calibration (Figure 3C, Table
1). Thus, when on the order of 1 amol of PSA standard was
analyzed, the ELIMSA reaction produced on the order of 1
pmol amounts of PA and thus apparently achieved nearly a
million-fold amplification (Table 1).
a
Applied to Selected NHP Samples
method
result
13C PA isotope dilution
2.01 pmol PA
2.19 pmol PA
1.58 pmol PA
1.28 pmol PA
57.78 pg PSA/well
PA external curve
13C external curve
13C PA one-point internal calibration
ELIMSA for Prostate Specific Antigen from Normal
Human Plasma. ELIMSA was applied to measure PSA in
NHP that was preselected to be below the detection limit of the
colorimetric assay as a test of the system under authentic
conditions. The external PSA standard showed a log linear
relationship between PA [M + H] ion intensity versus PSA
quantity and good technical reproducibility as measured by
replicate analysis of test samples using LC−ESI-MS with single
ion monitoring (SIM) at 169 m/z (Figure 3C). A total of 10
randomly selected normal plasma (NHP) samples were
analyzed on 3 independent days in a row with good agreement
and fitted to the calculated normal distribution. The average
amount of PSA in the NHP samples was about 67.8 pg with a
standard error of about 4.3 pg from a total of 30 NHP samples
(50 μL) (Table 2). The estimated PSA values showed a
Guassian, i.e., normal distribution, by the Shapiro-Wilk test (N
= 30, W 0.968499 Prob < W 0.3222) and the plotted
distribution showed little deviation from the normal curve as
illustrated with the quantile plot produced with the statistical
analysis system R (see the Supporting Information). In the
application of ELIMSA to measure PSA in the preselected
NHP samples that were undetectable by color assays, the
results were all 33 pg per well or greater and were within the
limits of the known standards for detection and quantification
by ELIMSA. Since all of the normal male plasma samples were
within the limit of quantification of the assay, it was not
possible to establish the lower limit of ELIMSA for PSA in
patient samples.
PSA external curve
a
The normal human plasma (NHP) samples were pre-selected from
samples that failed to yield colorimetric values. The levels of prostate
specific antigen (PSA) in selected normal human plasma samples
(NHP) was estimated from an external standard curve of absolute PSA
amounts using an LC−ESI-MS assay for pyridoxamine produced by
the enzyme conjugated AP-SA probe that was bound to the
biotinylated PSA detection antibody. The NHP sample was selected
from those that had less than 0.1 ng of PSA per 100 μL and so could
not be measured by colorimetric methods and was aliquoted for
replicate experimentation. The experimental design of the analytical
chromatography experiments used to address absolute quantification
of PSA by an LC−ESI-MS assay to provide these values of PA is
shown in Figure 3. The results indicate that about 2 pmol of PA is
produced when about 3 amol of PSA is provided and so shows about a
million fold amplification of the signal.
Table 2. Replication of the Analysis in Normal Human
Plasma That Was Selected As Too Low to Estimate PSA by
a
the Colorimetric Assay Is Shown in pg PS/well)
day 1
day 2
day 3
93.04
58.30
61.28
83.35
41.06
63.34
65.46
50.10
67.64
84.02
105.33
58.13
42.26
41.88
35.33
33.75
89.46
75.67
85.32
73.26
58.36
61.30
68.37
77.90
48.96
61.30
82.59
102.53
88.51
75.20
Ultrasensitive Detection. In order to estimate the lowest
possible detection of the AP-SA enzyme conjugate, a dilution
series of AP-SA was made that showed a clear detection of 1
pg/mL as quantified by the injection of 0.1 μL (∼526 ymol of
AP-SA on column) at 20 μL per minute with robust
electrospray and a simple ion trap (Figure 4). The
contamination of the SIM 169 m/z channel by some minor
component in the injection buffer is especially obvious at low
concentrations. Hence, it was possible to clearly detect the
presence of ∼1 fg of AP-SA on column or less at 20 μL per
minute through an electrospray source with a simple ion trap.
all 3 days
67.76
mean
66.67
5.07
64.04
8.01
72.5
5.06
Std Err.
4.32
a
The within day mean and standard error (Std Err.) is shown vertically
below and the summary over all 3 days to the right.
DISCUSSION
The amount of PA produced by the ELIMSA reactions was
also estimated by the one-point calibration against the addition
of 1.5 pmol of 13C PA added to each NHP ELIMSA reaction
sample. Each replicate ELIMSA reaction was monitored for PA
[M + H] at SIM 169 m/z alongside 1.50 pmol of the 13C PA
[M + H] internal standard monitored at SIM 172 m/z (Figure
3B, Table 1). There was good agreement between the
independent internal isotope dilution curve and the one-point
ratio estimate from a pooled normal sample that showed about
1.28 pmol of PA. On the same day, and on the same
chromatogram as the one-point calibration experiment, we also
measured PA production in terms of external 13C PA and PA
standard curves with the analysis of 0.1 μL analyzed from each
100 μL well. Reading the amount of pyridoxamine (PA)
produced by the ELIMSA of NHP plasma from the molecular
■
Sensitivity. Direct analysis of the generic detection reagent
AP-SA by LC−ESI-MS permits the quantification of the AP-SA
probe to at least 1 pg per mL range from the injection of 0.1 μL
from the sample (i.e., yoctomole amounts of the molecular
probe on column) and was sufficiently sensitive to detect all the
low-abundance PSA samples in this study. The sensitivity was
such that the PSA values of all NHP were well within the
standards at concentrations far below the robust detection limit
of color, fluorescence, or ECL methods.3 The ELIMSA assay
seems to detect the presence of the probe to levels comparable
to or beyond that of radio immunometric assays but without
the administrative burden of handling radioisotopes. The
sensitivity of the assay is limited at least in part by confounding
substances in the injection buffer and the background binding
10688
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691

Analytical Chemistry
Article
Figure 4. Detection of AP-SA from 1 to 100 pg per mL with analysis of 0.1 μL that shows sensitivity on the order of 5.26 × 10−22 mol, i.e., 526 ymol
or ∼1 zmol, of AP-SA on column.
of the biotinylated enzyme probe to the ELISA plate that must
be accounted for in the calculations. Sensitivity will vary on a
case by case basis depending on the quality of available
reagents. However, the capacity of ELIMSA to quantify
molecules at low concentrations is so great that the sensitivity
of the detection method will no longer be the limiting factor in
most experiments.
PA internal isotope dilution curve. Together, these several
controls unambiguously demonstrate that ELIMSA of PSA is
based on LC−ESI-MS assays that are log linear and accurate.
The PA and 13C PA ion intensity values were linear after log
transformation, and the calculated PSA results were normal and
so suitable for analysis of variance (ANOVA) statistical analysis
between experimental treatments. However, since the PSA
standard data may have to be log transformed and so cannot
pass through zero, it is traditional to calculate the intensity or
quantity of the data as log (i + 1). After log transformation, the
data apparently shows acceptable linearity over at least 3 orders
of magnitude (R2 of ∼0.99). Moreover, the results of PSA in
the low-abundance normal human plasma by ELIMSA here
closely match the correct order of magnitude for PSA reported
in the literature. Thus, multiple lines of evidence in agreement
indicate that the ELIMSA, standardized against known amounts
of the PSA calibrant, provides the accurate absolute
quantification of PSA. It was previously demonstrated that
ELISA and ELIMSA gave results that were linear and consistent
in the nanogram per well range by direct comparison to
colorimetric assays.3 It seems that ELIMSA may be used to
absolutely quantify PSA proteins at picograms per well. Since
all the normal samples analyzed here were no less than 33 pg
per well, all samples were well within the quantitative range of
the assay as measured against absolute PSA standards.
Agreement between Independent Methods. The result
with the internal 13C isotope dilution curve was in close
agreement with the external PA curve and the external 13C PA
curve. The one-point calibration method showed the greatest
divergence, but four independent methods were within 1 digit
in the same order of magnitude and so in general showed good
agreement. The results of ELIMSA to quantify PSA in the
colorimetric range provided results that closely matched the
results of well-established quantification using the amplex red
and naphthol AS-MX phosphate substrates. Hence, there are
multiple lines of evidence in agreement that LC−ESI-MS may
provide for the accurate quantification of molecules such as PA
after log transformation. The internal isotope assays here
required the assumption of log linearity and agreed with the
external curves that were also suitably linear after log
transformation. No complex mathematics of any kind were
required to solve for the values presented, and the log linear
regression of the signals and quantification was robust.
Absolute Quantification of PSA by ELIMSA. The
ELIMSA employs an external absolute PSA standard dilution
curve monitored from LC−ESI-MS of the PA released by the
AP-SA probe. The linear results with the authentic PSA protein
used as an external, absolute standard were consistent with the
Log linearity of LC−ESI-MS. In this study, the external
PSA curve, the internal 13C PA isotope dilution curve, the
external PA curve, and the external 13C PA isotope standard
curves all agreed that mass spectrometry showed a linear
relationship with quantity after simple log transformation. The
log linear results with PA were in agreement with log linear
linearity of PA and 13C PA external standard curves and the 13
C
10689
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691

Analytical Chemistry
Article
Author Contributions
results from some standard peptides or peptides from blood
proteins,16a resorufin, and naphthol AS-MX,3 many peptides
and proteins from cells,16c or proteins secreted from cells into
the experimental media.16b Taken together, the results with
small molecules, isotopic labels, and many peptides all indicate
that LC−ESI-MS is apparently log linear for some pure
compounds and thus may approach linearity for some
molecules after preseparation by chromatography. Linear signal
intensities that are only a small fraction of the total ion current
are consistent with the competitive nature of the electrospray
ionization process19 in which the presence of other analytes
may affect observed intensities in unpredictable ways. The
potentially confounding effects of competition for ionization
were apparently not a factor in this case in which relatively pure
analyte PA dissolved in a 1 mM tris buffer is efficiently resolved
by high-pressure liquid chromatography prior to ionization with
results closely approaching log linearity.
A.F.-M. screened the substrates, optimized the detection
conditions, performed most experiments and numerical
analysis, created the charts and graphs, and proofed the
manuscript. J.G.M. performed preliminary experiments, con-
ceived and designed the ELIMSA experiment, wrote the
proposal, performed the statistical analysis, created the final
figures, and wrote the paper.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by a Discovery Grant from the
Natural Science and Engineering Research Council of Canada
to J.G.M.
REFERENCES
■
(1) Engvall, E.; Perlman, P. Immunochemistry 1971, 8 (9), 871−874.
(b) Van Weemen, B. K.; Schuurs, A. H. FEBS Lett. 1971, 15 (3), 232−
236.
CONCLUSION
■
The use of LC−ESI-MS to measure an enzyme activity with a
conjugated molecular probe provides a universal system for the
quantification of macromolecules including proteins, ligands,
and/or receptors at femtogram to picogram amounts or less
with good signal-to-noise values. The detection of yoctomoles
of the AP-SA probe and measurement of zeptomoles of PSA is
comparable to, or more sensitive than, other methods claimed
to date2,14,15,20 but requires only commonly available micro
electrospray and a simple ion trap. ELIMSA with the substrate
PA5P that releases pyridoxamine showed more symmetrical
peaks, baseline to baseline separation even at close injection
intervals, and a much flatter baseline compared to the previous
substrate naphthol AS-MX phosphate and was a major
improvement to the method. The experiments here with
external and internal standards and isotope dilutions clearly
demonstrate that LC−ESI-MS may provide a linear relationship
between intensity and quantity after log transformation. All of
the PSA samples in this study were within the robust
quantification range of the ELIMSA in the zeptomole range
on column and could be estimated by interpolation between
known amounts of PSA standard solutions that were log linear
with a high signal-to-noise ratio. Here pyridoxamine, AP-SA,
and PSA molecules were absolutely quantified by LC−ESI-MS
against external standards and the linearity and accuracy of the
PA was supported by the use of internal isotopic standards. The
log linearity and normality of mass spectral intensity after log
transformation opens up broad vistas for the direct
quantification of many molecules by classical linear models
and ANOVA statistical analysis.3,16 The experiments here
unambiguously demonstrate that enzyme labeled probes can
sensitively quantify proteins or receptor−ligand complexes to
zeptomole amounts by direct mass spectrometric analysis and
will have a broad application and utility for measuring
molecules at low concentrations that greatly extends the
practical sensitivity of ELISA biomolecular analysis.
(2) Kulasingam, V.; Smith, C. R.; Batruch, I.; Buckler, A.; Jeffery, D.
A.; Diamandis, E. P. J. Proteome Res. 2008, 7 (2), 640−647.
(3) Florentinus-Mefailoski, A.; Safi, F.; Marshall, J. G. J. Proteomics
2014, 96, 343−352.
(4) Simopoulos, T. T.; Jencks, W. P. Biochemistry 1994, 33 (34),
10375−10380.
(5) Staros, J. V. Biochemistry 1982, 21 (17), 3950−3955.
(6) Chaiet, L.; Wolf, F. J. Arch. Biochem. Biophys. 1964, 106, 1−5.
(7) Bothner, B.; Chavez, R.; Wei, J.; Strupp, C.; Phung, Q.;
Schneemann, A.; Siuzdak, G. J. Biol. Chem. 2000, 275 (18), 13455−
13459.
(8) Pris, A. D.; Mondello, F. J.; Wroczynski, R. J.; Murray, A. J.;
Boudries, H.; Surman, C. M.; Paxon, T. L. Anal. Chem. 2009, 81 (24),
9948−9954.
(9) Hempen, C.; van Leeuwen, S. M.; Luftmann, H.; Karst, U. Anal.
Bioanal. Chem. 2005, 382 (1), 234−238.
(10) Banks, J. F., Jr.; Shen, S.; Whitehouse, C. M.; Fenn, J. B. Anal.
Chem. 1994, 66 (3), 406−414.
(11) Orsin, F.; Shulman, S. J. Exp Med. 1971, 134 (1), 120−140.
(12) Black, M. H.; Grass, C. L.; Leinonen, J.; Stenman, U. H.;
Diamandis, E. P. Clin. Chem. 1999, 45 (3), 347−354.
(13) Marshall, J.; Schmit, J. C.; Betsou, F. Clin. Proteomics 2014, 11
(1), 3.
(14) Diamandis, E. P. Clin. Biochem. 1988, 21 (3), 139−50.
(15) Chikkaveeraiah, B. V.; Bhirde, A. A.; Morgan, N. Y.; Eden, H. S.;
Chen, X. ACS Nano 2012, 6 (8), 6546−6561.
(16) (a) Bowden, P.; Thavarajah, T.; Zhu, P.; McDonell, M.; Thiele,
H.; Marshall, J. G. J. Proteome Res. 2012, 11 (4), 2032−2047.
(b) Florentinus, A. K.; Bowden, P.; Sardana, G.; Diamandis, E. P.;
Marshall, J. G. J. Proteomics 2012, 75 (4), 1303−1317. (c) Florentinus,
A. K.; Jankowski, A.; Petrenko, V.; Bowden, P.; Marshall, J. G. J.
Proteomics 2011, 75 (2), 450−468.
(17) Chelius, D.; Huhmer, A. F.; Shieh, C. H.; Lehmberg, E.; Traina,
J. A.; Slattery, T. K.; Pungor, E., Jr. J. Proteome Res. 2002, 1 (6), 501−
513.
(18) (a) Schwartz, J. C.; Senko, M. W.; Syka, J. E. J. Am. Soc. Mass
Spectrom. 2002, 13 (6), 659−669. (b) Stafford, G., Jr. J. Am. Soc. Mass
Spectrom. 2002, 13 (6), 589−596.
(19) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse,
C. M. Science 1989, 246 (4926), 64−71.
(20) (a) Salehpour, M.; Possnert, G.; Bryhni, H. Anal. Chem. 2008,
80 (10), 3515−3521. (b) Liu, G.; Wang, J.; Kim, J.; Jan, M. R.; Collins,
G. E. Anal. Chem. 2004, 76 (23), 7126−7130. (c) Lohmann, W.;
Hayen, H.; Karst, U. Anal. Chem. 2008, 80, 9714−9719. (d) Munge,
B.; Liu, G.; Collins, G.; Wang, J. Anal. Chem. 2005, 77 (14), 4662−
4666. (e) Rozet, E.; Morello, R.; Lecomte, F.; Martin, G. B.; Chiap, P.;
Crommen, J.; Boos, K. S.; Hubert, P. J. Chromatogr., B: Anal. Technol.
ASSOCIATED CONTENT
* Supporting Information
This material is available free of charge via the Internet at
http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*E-mail: 4marshal@ryerson.ca.
■
10690
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691

Analytical Chemistry
Article
Biomed. Life Sci. 2006, 844 (2), 251−260. (f) Takatsy, A.; Boddi, K.;
Nagy, L.; Nagy, G.; Szabo, S.; Marko, L.; Wittmann, I.; Ohmacht, R.;
Ringer, T.; Bonn, G. K.; Gjerde, D.; Szabo, Z. Anal. Biochem. 2009,
393 (1), 8−22. (g) Tang, C. K.; Vaze, A.; Rusling, J. F. Lab Chip 2012,
12 (2), 281−286. (h) Valentini, F.; Compagnone, D.; Gentili, A.;
Palleschi, G. Analyst 2002, 127 (10), 1333−1337. (i) Zhang, S.; Yang,
J.; Lin, J. Bioelectrochemistry 2008, 72 (1), 47−52. (j) Shi, T.; Sun, X.;
Gao, Y.; Fillmore, T. L.; Schepmoes, A. A.; Zhao, R.; He, J.; Moore, R.
J.; Kagan, J.; Rodland, K. D.; Liu, T.; Liu, A. Y.; Smith, R. D.; Tang, K.;
Camp, D. G., 2nd; Qian, W. J. J. Proteome Res. 2013, 12 (7), 3353−
3361. (k) Cook, D. B.; Self, C. H. Clin. Chem. 1993, 39 (6), 965−971.
(l) Kricka, L. J. Clin. Biochem. 1993, 26 (5), 325−331. (m) Zhang, H.;
Li, X. F.; Le, X. C. Anal. Chem. 2012, 84 (2), 877−884.
10691
dx.doi.org/10.1021/ac502572a | Anal. Chem. 2014, 86, 10684−10691