ARTICLE

pubs.acs.org/ac

Strategy and Its Implications of Protein Bioanalysis Utilizing

High-Resolution Mass Spectrometric Detection of Intact Protein
Qian Ruan,† Qin C. Ji,*,‡ Mark E. Arnold,‡ W. Griffith Humphreys,† and Mingshe Zhu†
†
Biotransformation and ‡Bioanalytical Sciences, Bristol-Myers Squibb Comapny, Route 206 and Province Line Road, Princeton,
New Jersey 08543, United States
bS Supporting Information
ABSTRACT: Currently, mass spectrometry-based protein bioanalysis is
primarily achieved through monitoring the representative peptide(s) resulting
from analyte protein digestion. However, this approach is often incapable of
differentiating the measurement of protein analyte from its post-translational
modifications (PTMs) and/or potential biotransformation (BTX) products.
This disadvantage can be overcome by direct measurement of the intact
protein analytes. Selected reaction monitoring (SRM) on triple quadrupole
mass spectrometers has been used for the direct measurement of intact protein.
However, the fragmentation efficiency though the SRM process could be
limited in many cases, especially for high molecular weight proteins. In this study, we present a new strategy of intact protein
bioanalysis by high-resolution (HR) full scan mass spectrometry using human lysozyme as a model protein. An HR linear ion-trap/
Orbitrap mass spectrometer was used for detection. A composite of isotopic peaks from one or multiple charge states can be isolated
from the background and used to improve the signal-to-noise ratio. The acquired data were processed by summing extracted ion
chromatograms (EIC) of the 10 most intense isotopic ions of octuply protonated lysozyme. Quantitation of the plasma lysozyme
was conducted by utilizing high resolving power and an EIC window fitting to the protein molecular weight. An assay with a linear
dynamic range from 0.5 to 500 μg/mL was developed with good accuracy and precision. The assay was successfully employed for
monitoring the level of endogenous lysozyme and a potential PTM in human plasma. The current instrumentation limitations and
potential advantages of this approach for the bioanalysis of large proteins are discussed.

C oncentration determination of analytes such as biomarker

molecules and drug substances and their related compounds
in biological matrixes, termed as “bioanalysis”, is a critical part of
drug discovery and development. Although ligand-binding assays
are still the main platform for the bioanalysis of protein and
peptide analytes, liquid chromatography
mass spectrometry
(LC
MS) assays play increasingly more important roles as a
complementary platform.1
3 Two primary strategies used in the
LC
MS bioanalysis of proteins in biological matrices are (1)
protein quantitation through mass spectrometric detection of a
representative peptide segment(s) generated from enzymatic
digestion or chemical cleavage4
7 and (2) protein quantification
through direct mass spectrometric detection of intact proteins.8,9
As shown in Figure 1, while the strategy through measurement of
a representative peptide segment(s) has gained increasing ac-
ceptance, this “bottom-up” approach has a major limitation that
the peptide(s) may not accurately represent the protein analyte.
For example, it cannot differentiate a protein analyte and its
products derived from post-translational modification (PTM)
and/or in vivo biotransformation (BTX) processes if unchanged
peptide segments are monitored. This disadvantage could be
overcome by the second strategy by direct measurement of the
intact protein analyte.
To apply this “top-down” approach, selected reaction mon-
itoring (SRM) using triple-quadrupole mass spectrometers has
been developed.8,9 Although SRM detection provides excellent
selectivity and duty-cycle, it also has some inherent disadvantages
in the fragmentation-dependent MS detection. As the molecular
weight of a protein increases, the fragmentation efficiency of the
protein through the SRM process often decreases.10 Even when
the protein is fragmented, the charges are more evenly dis-
tributed into many fragments rather than concentrated onto a
few major fragments like small molecules. Furthermore, the
electrospray ion source commonly used on triple-quadrupole
MS tends to ionize a protein into different charge states. SRM
can only detect the one particular fragment transition from
a precursor ion at a specific mass to charge ratio, while the
majority of the ions in all the other charge states cannot be
utilized at the same moment.
Full scan MS detection is independent of the fragmentation
efficiency of a protein. In addition, if protein ions in several
charge states have similar intensities, full scan MS with a nonion-
filtering type mass spectrometer allows summation of these ion
signals, thus improving the analyte signal intensity. In addition,
full scan MS can provide additional qualitative and quantitative
information of protein analyte related components in each
individual sample, allowing the flexibility to reprocess acquired
data for additional investigations, such as quantitation of
Received: June 18, 2011
Accepted: October 4, 2011

r XXXX American Chemical Society A dx.doi.org/10.1021/ac201540t | Anal. Chem. XXXX, XXX, 000–000
Analytical Chemistry
Figure 1. Protein bioanalysis strategy.
unpredicted PTMs in patient samples. One common drawback
of the full scan MS approach is the lower selectivity caused by
higher background or interferences as compared to that seen in
the SRM approach. Recently, there have been significant ad-
vances in the capabilities of high-resolution mass spectrometry
(HR-MS) instrumentation.11,12 HR-MS can provide additional
advantages in resolving each isotopic ion from the background so
that higher selectivity can be achieved. Furthermore, when used
for identification of protein modifications present in various
in vivo samples, the high accuracy offered by HR-MS can be
critical to the mass assignment, e.g., differentiating oxidation
(+15.9949 Da) and methylation (+14.0156 Da) as well as
determining oxidation/reduction states of disulfide bonds.
Here, we report our first exploratory use of full scan HR-MS
for bioanalysis of an intact protein. Human lysozyme, a 15 kDa
glycoside hydrolase13,14 circulating in plasma, was used as a model
protein to demonstrate the feasibility and potential advantages of
this approach. Lysozyme has four disulfide bonds, and their
integrity is a critical indication of normal functionality.15 The
HR-MS capability in isotopic resolution and accurate mass
determination of a protein analyte is indispensable for the
identification of the normal structure of the protein based on
exact mass deviations, such as lack or presence of expected PTMs
or amino acid substitutions. The assay development for direct
LC
MS analysis of the intact human lysozyme, such as sample
preparation and optimization of chromatographic separation, are
also discussed, and the ability of this assay to quantify the protein
in human plasma is demonstrated.
’ EXPERIMENTAL DETAILS
Chemical, Reagents, Materials, and Apparatus. Chemicals
and Reagents. Human milk lysozyme, chicken egg yolk lysozyme,
acetic acid, trifluoroacetic acid (TFA), and ACS grade ammo-
nium acetate were purchased from Sigma-Aldrich (St. Louis,
MO). HPLC grade acetonitrile and methanol were obtained
from Thermo Fisher Scientific Co. (Waltham, MA). Human and
monkey K2EDTA plasma were obtained from Bioreclamation
(Hicksville, NY).
Materials and Apparatus. Plates used for sample collection
and injection were Costar Brand 96-well assay blocks from VWR
B dx.doi.org/10.1021/ac201540t |Anal. Chem. XXXX, XXX, 000–000
ARTICLE
International (Bridgewater, NJ). Oasis HLB 30 mg solid phase
extraction (SPE) plates were purchased from Waters (Milford,
MA). The SPE manifold for a Tomtec Quadra 96 model 320
robotic liquid handler (Tomtec, Hamden, CT) was used for
SPE processing. An SPE Dry microplate sample concentrator
(Biotage AB, Uppsala, Sweden) was used for solvent evaporation
of SPE eluents.
LC
MS Equipment. All sample analyses were performed on
an LC
MS system that consisted of Shimadzu LC-20AD HPLC
pumps (Shimadzu Corporation, Kyoto, Japan), a LEAP auto-
sampler (LEAP Technologies, Carrboro, NC), an Xbridge BEH300
C4, 50 mm  2.0 mm, 3.5 μm column (Waters, Milford, MA)
with a Sidewinder column heater (Restek, Bellefonte, PA) and an
LTQ/Orbitrap Classic mass spectrometer (Thermo Fisher Scien-
tific, Waltham, MA). HPLC was performed at a constant flow rate
of 400 μL/min using a binary solvent system. Mobile phase A was
composed of 0.1% TFA and 0.5% acetic acid in HPLC grade
H2O; mobile phase B was 0.1% TFA and 0.5% acetic acid in
acetonitrile. The HPLC gradient started with 5% B, linearly
increased to 40% B in 4 min, continued to 50% B in 4 min,
increased to 100% B in 0.5 min, and maintained at 100% B for
0.4 min prior to column re-equilibration. The LTQ/Orbitrap
settings were as follows: sheath gas at 60, auxiliary gas at 30, sweep
gas at 5, spray voltage at 5 kV, tube lens at 100 V, automatic gain
control (AGC) at 5  105, max injection time at 300 ms, and
resolving power of 100 000 at 400 m/z for all analyses. During
method development, MS full scans were carried out using a
mass range of 200
4000 m/z. For protein quantitation of the
human plasma samples, the MS full scan range was narrowed
down to 1580
1850 m/z.
Sample Preparation. Preparation of Calibration Standards
and Quality Control (QC) Samples. Stock solutions containing
100 mg/mL human lysozyme (analyte) and 100 μg/mL chicken
lysozyme (internal standard (IS)) were prepared in water. An
aliquot of 20 μL human lysozyme stock solution was added into
3.98 mL of control monkey plasma to make a plasma working
solution (500 μg/mL). Two sets of eight standard levels
(0.5, 1.00, 2.50, 10.0, 50.0, 100, 200, and 500 μg/mL) were prepared
by successively diluting the working solution into control mon-
key plasma. Four replicates of five QC levels (0.5, 1.5, 12.0, 100,
and 160 μg/mL) were prepared by adding an appropriate volume
of the working solution into control monkey plasma.
Solid Phase Extraction. Aliquots of 400 μL of QCs, standards,
or human plasma samples were mixed with 80 μL of chicken
lysozyme (IS, 100 μg/mL) and 400 μL of buffer (50 mM
NH4OAc, 0.2% TFA in H2O). The mixture was vortexed and
then centrifuged at 1500g for 10 min. The supernatant was
separated and subjected to SPE cleanup.
Each well in a SPE plate was conditioned by 1 mL of methanol
followed by 1 mL of 0.1% TFA in H2O. After plate conditioning,
800 μL of each sample mixture was loaded into the plate. Each
well on the plate was washed with 1 mL of 1% TFA in ACN/H2O
(15/85 v/v) and eluted with 1 mL of 1% TFA in ACN/H2O
(70/30 v/v). Samples were dried under a stream of N2 for 2 h at
room temperature and reconstituted in 150 μL of 0.1% TFA and
0.5% acetic acid in 5% ACN/H2O. After vortexing and centrifu-
gation, the supernatants were transferred to injection vials for
LC
MS analysis. Each injection contained 30 μL of the recon-
stituted samples.
Data Processing. The acquired data were processed by
summing the extracted ion chromatogram (EIC) of the most
intensive isotopic ions of octuply protonated lysozymes. For
Analytical Chemistry
Figure 2. Mass spectra of human and chicken lysozymes: (A) full scan spectrum of human and chicken lysozymes in multiple charge states, (B) full scan
spectrum of human lysozyme and PTM carrying eight charges in plasma, (C) enlarged spectrum of human lysozyme carrying eight charges, and (D)
enlarged spectrum of human dehydrated lysozyme carrying eight charges.
chicken lysozyme (IS), the most intensive isotopic ions were
m/z 1788.6280, 1788.7540, 1788.8810, 1789.0060, 1789.1300,
1789.2550, 1789.3800, 1789.5060, 1789.6300, and 1789.7580;
for human lysozyme (analyte), m/z 1836.9140, 1837.0400,
1837.1590, 1837.2890, 1837.4160, 1837.5560, 1837.6670,
1837.7920, 1837.9150, and 1838.0410; for dehydrated human
lyzozyme, m/z 1834.9910, 1835.0320, 1835.1610, 1835.2850,
1835.4040, 1835.5350, and 1835.6580. The EIC window for
quantitation was set at (10 ppm. Peak areas were calculated using
QualBrowser software (Thermo Fisher Scientific, Waltham, MA).
The standard curve was fitted to a linear regression model with a
1/x2 weighting factor using in-house software. All statistic para-
meters were calculated by Microsoft Excel. The exact molecular
weight of the most abundant isotope was determined by Protein
Prospector (University of California, San Fransisco).
’ RESULTS AND DISCUSSION
High-Resolution Mass Spectrometry Detection. When a
strategy of MS detection is used for protein bioanalysis, HR-MS
has unique advantages over low-resolution approaches. In a com-
monly used ESI source, proteins can be protonated into multiple
charge states (Figure 2A). The ion at each charge state consists
of a cluster of isotopic peaks (Figure 2B,C), which cannot be
resolved in low-resolution mass spectrometry (Figure 3A,B).
C dx.doi.org/10.1021/ac201540t |Anal. Chem. XXXX, XXX, 000–000
ARTICLE
If the resolving power increases enough to resolve these peaks, only
the MS response of the resolved peaks will be used as the signal
for quantitation while the MS response from background inter-
ference between these peaks will be excluded (Figure 3D). As a
result, the assay sensitivity (signal-to-noise ratio) for quantitation
will be enhanced. This is especially important for most biological
samples, such as plasma or serum samples, which contain large
amounts of interfering components. As shown in Figure 4, the
interference of background ions in plasma spreads over the entire
mass range of the analyte protein’s isotopic peaks of interest.
Utilization of HR-MS in the analysis of biological samples
provides an opportunity for minimizing such interference.
In order to obtain effective improvement in selectivity and
sensitivity using this approach, three critical interrelated factors
need to be considered: mass spectrometer resolving power,
molecular weight of the protein analyte, and the EIC window
for quantitation. We propose here that the value of the resolving
power needs to be at least 4 times the molecular weight of the
protein analyte, and the EIC window should be set around the
peak width of each isotopic ion. The following is the rationale of
this consideration using the 15 kDa human lysozyme as an
example. The full-scan mass spectrum of an ionized lysozyme
should present every two adjacent isotopic peaks one Dalton
apart, which is ∼60 ppm (1 Da/15 kDa) apart regardless of
the charge states. When the resolving power of 10k (full width at
Analytical Chemistry
half-maximum (FWHM): 1/10k = 100 ppm, baseline peak width
= 200 ppm) was employed on the low-resolving power linear ion
trap MS (LTQ) using the zoom/ultrazoom scan function, the
mass spectrometer was unable to resolve the isotopic peaks from
the baseline (Figure 3B). When the resolving power was increased
on an HR-MS to 30k (twice the protein’s molecular weight),
baseline separation of the isotopic peaks was just barely achieved
(FWHM, 1/30k ≈ 30 ppm, baseline peak width ≈ 60 ppm
(Figure 3C)). However, if all isotopic peak signals were used
for quantitation, EIC windows of these peaks would all be joined
Figure 3. Effect of resolving power on separation of isotopic peaks of oct-
uply charged lysozymes: (A) resolving power of 2k, (B) resolving power
of 10k, (C) resolving power of 30k, and (D) resolving power of 60k.
Figure 4. LC
MS spectra comparison of lysozyme and matrix interference in human plasma: (A and B) full scan MS of lysozyme and (C and D) full
scan MS and enlarged full scan MS of interference peaks eluted after lysozyme.
D dx.doi.org/10.1021/ac201540t |Anal. Chem. XXXX, XXX, 000–000
ARTICLE
together, suggesting no advantage compared to using one EIC
window to cover the whole cluster of isotopic ions in the low-
resolution approach. When the resolving power further increased
to 4 times of the protein’s molecular weight (60k), a gap of one
peak width in the background emerged between adjacent iso-
topic peaks (FWHM, 1/60k ≈ 15 ppm, baseline peak width ≈
peak gap width ≈ 30 ppm (Figure 3D)) and enabled clear
separation of the background from the signal of lysozyme. If the
EIC window was narrowed down to one peak width accordingly,
interference signal present in the gaps could be removed without
sacrificing the analyte’s peak signal.
Taken together, it is recommended to use the following
equation to obtain resolving power that allows adequate separa-
tion of isotopic peaks for HR-MS quantitation of an intact
protein:
R ¼ kM
where M represents the molecular weight (Da) of the protein of
interest; k represents the factor of isotopic peak separation (k = 2
indicates baseline separation; k g 4 is recommended for selective
quantitation of an intact protein); and R represents the resolving
power that needs to be obtained for the ions used for quantita-
tion. When the practical R is determined, the EIC window can be
set to around the base peak width of the isotopic ion (2/R) to
allow maximum signal from the analyte with minimal signal from
the background interference.
With the application of this equation to the human lysozyme
(M = 15 kDa), the expected practical resolving power (R) for
sensitive quantitation (k = 4) is 15 000  4 = 60 000. Since the
resolving power of the Orbitrap declines proportional to m/z
0.5,
the highest resolving power setting of 100 000 at 400 m/z was
utilized to ensure the practical solution of ∼60 000 for the oct-
uply charged lysozymes at m/z 1800
1850 (analytes and IS
(Figure 3D)). The FWHM of the isotopic ions was thus calculated
as 1/60 000 ≈ 15 ppm (baseline peak width ≈ 30 ppm). As
demonstrated in Figure 5A
C, when the EIC window for human
Analytical Chemistry
Figure 5. Effect of EIC window on assay selectivity and sensitivity:
(A) EIC window, (0.5 m/z; (B) EIC window, (30 ppm; (C) EIC
window, (10 ppm; (D) EIC window, (1 ppm.
lysozyme was narrowed down from (0.5 m/z to (10 ppm, the
peak intensity ratio of the analyte over the interference increased
5 times. Further reduction of the extraction window from
(10 ppm to (1 ppm caused reductions in the peak intensity
of both analyte and interference and thus did not significantly
improve the analyte/interference ratio (Figure 5C,D). Com-
bined with consideration of minimizing the matrix interference in
plasma samples, the EIC window was set to a little less than the
baseline peak width (at (10 ppm).
The example of lysozyme quantitation demonstrates improve-
ment of the signal-to-noise ratio using the isotopic cluster peaks
from one charge state (z = 8). Full scan mass spectrometry
detection enables the capability in summing signals of isotopic
cluster ions from multiple charged states. The ion selection for
quantitation depends on the distribution of signal intensity in
different charged states. In this case, the protonated lysozyme
was concentrated at the z = 8 charge state in the optimized
ionization conditions. The isotopic peaks from the other charged
states were <30% of the intensity of the highest z = 8 peak and
thus excluded for quantitation. For analyte proteins evenly ionized
into different charge states, all ions in the charge states with
significant signal intensity should be included in the combined
EIC to achieve the best sensitivity.
For quantitation of proteins with higher molecular weights, a
mass spectrometer with higher resolving power is required. The
higher the resolving power of a mass spectrometer, the better its
selectivity; and therefore, better sensitivity as a narrower EIC
window is utilized to remove the background interference. For
example, to fully utilize this approach to quantify an intact
antibody with 150 kDa molecular weight, it requires a 600 000
practical resolving power based on the above model equation. It
is still a big challenge to achieve such high resolving power with
high scan speed. When ultra high-resolution mass spectrometers
become available and are compatible with HPLC chromato-
graphy, they will provide a powerful tool to quantitate a range
of intact proteins in biological matrixes with good sensitivity,
E dx.doi.org/10.1021/ac201540t |Anal. Chem. XXXX, XXX, 000–000
ARTICLE
Figure 6. Effect of HPLC column temperatures (T) on the chromato-
graphy: (A) T = 30 °C, (B) T = 40 °C, (C) T = 50 °C, (D) T = 60 °C, and
(E) T = 70 °C.
selectivity, and the capability to simultaneously differentiate
proteins from their products of potential PTMs and BTX as well
as other homologues.
Assay Development of Sample Preparation and Chroma-
tographic Method. An LC method with good chromatographic
peak shape is often difficult to obtain for protein samples;
however, it is essential for achieving quality bioanalysis results.
Sample preparation is another important part of assay develop-
ment. Although the objective of this research was not to emphasize
the development of sample preparation and chromatographic
conditions for the model analyte, assay development was con-
ducted to have a workable assay to demonstrate the concept of
HR LC
MS bioanalysis of proteins. For the sample preparation,
protein precipitation and SPE were tested. The results suggested
that SPE provided better sample cleanup (results not shown).
Different combinations of organic and aqueous phases were used
to obtain the optimal conditions for wash and elution steps of
SPE sample preparation. The result indicated that using 40
80%
of ACN as elution solvent had the highest recovery of the analyte,
while there was minimal elution of the analyte when less than 20%
ACN was used. Therefore, 15% of ACN in washing solution and 70%
of ACN in elution solvent were used to remove polar interference
components and collect cleaned up samples, respectively.
For the HPLC chromatography development, one chromato-
graphic factor we tested was column temperature. As shown in
Figure 6A, room temperature chromatography exhibited broad
peak width and poor peak shape. Even at elevated column
temperature up to 50 °C (Figure 6C), the peak shape was still
undesirable, with multiple overlapping wide peaks. When the
column temperature increased to 70 °C (Figure 6D,E), the peak
narrowed down to ∼20 s wide and became symmetric. This
phenomenon suggested full equilibrium of the different isoforms
of the lysozyme at this temperature. In addition, both isocratic
and gradient chromatography were tested. The results showed
that isocratic chromatography caused significant peak tailing for
lysozyme, while a slow gradient chromatography could achieve
Analytical Chemistry ARTICLE

1.92 ( 0.60
6.0 ( 1.3
average
human plasma human plasma human plasma human plasma human plasma human plasma human plasma human plasma
lot 8

1.1
7.2
lot 7

1.4
6.7
lot 6

3.0
5.1
lot 5

1.9
7.4
Figure 7. EIC of human plasma sample no. 1 from Table 1 with (10 ppm

lot 4

1.9
7.3
window: (A) EIC of IS (chicken lysozyme); (B) EIC of the analyte
(human lysozyme), inserted panel, EIC of the analyte with the EIC
window of (0.5 m/z; and (C) EIC of the dehydrated lysozyme.

lot 3

2.0
4.2
good peak shape and acceptable separation from the major
interference peaks. Thus, by combination of high-temperature
chromatography and mild gradient conditions, satisfactory LC
conditions for the quantitation were achieved (Figure 7B).
Assay Evaluation. Because of the existence of endogenous
lot 2

2.2
human lysozyme in human plasma, the assay evaluation was 4.7
performed using standards and QCs prepared by spiking human
Table 1. Quantitation of Native Lysozyme and Dehydrated Lysozyme

lysozyme into monkey plasma. No detectable signal response at

the retention time of either the human or chicken lysozyme (IS)
lot 1

2.7
5.3

was observed in blank monkey plasma samples (“Double Blank”,

Table 1). Similarly, only the internal standard response was
observed for blank plasma sample with chicken lysozme (“QC0”,
Table 1). Duplicate calibration curves with eight calibration
(monkey plasma) (monkey plasma)
double blank

standard concentration levels were used. As shown in the

supplementary figure, the coefficient of determination (R2)
NA

was 0.9717 with linear regression (1/x2 weighing factor). When

QC samples at five concentration levels were tested in quad-
ruplicate, the overall accuracy (% bias) ranged from
26% to
+29% of the theoretical concentrations, and the overall % CV
<LLOQ

ranged from 7% to 38%. The higher variation of measurement

QC0

results were mainly from the low QC level. Excluding the low QC
level, the assay performance was from 4% to 28% for the accuracy
and from 7% to 13% for % CV.
Endogenous lysozyme concentrations in eight individual lots
ratio of dehydrated lysozyme
lysozyme concn (μg/mL)

of plasma were measured and reported in Table 1. The advantage

of using HR-MS detection was further demonstrated using one
sample name

to lysozyme (%)

sample from Table 1, as shown in Figure 7B. While the EIC

window is (10 ppm, the signal-to-nose ratio of EIC peak is 40. If
same data is reprocessed with an EIC window of (0.5 m/z, the
signal-to-noise ratio for the same peak is reduced to 12. As
demonstrated in Figure 7, the assay sensitivity is sufficient to
measure the endogenous lysozyme concentrations. The average
F dx.doi.org/10.1021/ac201540t |Anal. Chem. XXXX, XXX, 000–000
Analytical Chemistry
concentration obtained was 1.92 ( 0.60 μg/mL, while the
literature value obtained using a radioimmunoassay was 4.6 (
0.8 μg/mL.16 The difference could be explained by the fact that
the radioimmunoassay might include the responses of nonspe-
cific binding or of other protein homologues with different
sequence contexts (cross-reactivity). In order to ensure monkey
plasma as a proper surrogate for human plasma for the prepara-
tion of standard curves and QC samples, a human plasma sample
was diluted 4-fold by monkey plasma and measured again for
lysozyme concentration. Calculated concentrations from the
diluted and undiluted samples differed by 11% (2.44 μg/mL vs
2.73 μg/mL, respectively). Although further improvement of
assay performance may be needed for drug development applica-
tions, the accuracy and precision obtained here can be regarded
as sufficient for biomarker discovery and pharmacokinetic anal-
ysis of therapeutic proteins in the drug discovery stage.
Investigation of the Protein PTMs and Its Implications.
The proposed approach has significant advantages in obtaining
information on PTMs as well. One advantage is in detection of the
formation of disulfide bridges within the analytes. Compared with
the common “bottom-up” approach by breaking disulfide bonds
before protein digestion, the quantitation of intact proteins can
maintain their disulfide bridges during analysis. In this study of
human lysozyme, no reducing agent was used in the sample pre-
paration. The spectrum of octuply charged human lysozyme
showed m/z 1837.5404 as the most abundant isotopic ion
(Figure 2B,C). The molecular weight of the most abundant iso-
tope was, therefore, calculated as 14 692.2648 Da ([1837.5404

1.0073]  8), which is 7.9086 Da less than the theoretical
molecular weight of the unfolded lysozyme (14 700.1734 Da).
This observation demonstrated that the eight cysteine residues in
the natural human lysozyme lost eight hydrogens and formed
four disulfide bonds during post-translational folding. As forma-
tion of a disulfide bond (
2.0156 Da) only causes a marginal
change to the large molecular weight of a protein, high resolution
and high accuracy are crucial to identify the most abundant
isotope and to assign its exact mass for determination of the
accurate number of disulfide bridges. In this regard, the HR-MS
quantitation method is also capable of detecting changes of the
number disulfide bridges caused by mutations, e.g., one less
disulfide bond formed in a mutant human lysozyme with the
Cys65 replaced by an Ala.17
Another advantage of this approach is the capability of
detecting and semiquantifying the other components chemically
derived from the analytes by PTM, e.g., chemical modifications of
the backbone amino acids and potential BTX products. When a
protein containing hundreds of amino acids is denatured during
sample processing, PTMs on one or two residues normally cause
minimal changes to the overall properties of the whole molecule
in LC
MS. It would be expected that any of these components
extracted during sample preparation would elute at a very similar
retention time as the analyte. HR LC
MS facilitates the identi-
fication of PTM products without prior knowledge by comparing
multiple characteristics, such as chromatographic peak shape,
distribution of isotopic isomers, and charge states with those of
the analyte protein. This approach can be readily used for
semiquantitative analysis of specific PTMs. In the study of human
lysozyme, a component with similar LC
MS properties of
lysozyme was detected in human plasma samples but not in
the blank monkey plasma (Figure 7C). This component exhib-
ited the same chromatographic retention time and similar peak
shape to the human lysozyme. The HR-MS spectrum indicated
G dx.doi.org/10.1021/ac201540t |Anal. Chem. XXXX, XXX, 000–000
ARTICLE
that it had the same charge state and isotopic distribution as the
human lysozyme (Figure 2C,D). These observations suggested
that this component was related to the human lysozyme. The
mass shift of the most abundant isotopic isomer was calculated as
17.9648 Da ((1837.5404
1835.2948)  8) less than the native
human lysozyme, indicating loss of a molecule of H2O (dehydration,
loss of 18.0105 Da). This dehydrated product was also present in
the purified human lysozyme used for standard curve and QC
preparation. The EIC peak area ratio of the dehydration product
versus the native lysozyme in human plasma was 6.0% ( 1.3%,
close to the ratio for the low QC samples (5.2 ( 1.1%). The
dehydration could result from PTM on serine, threonine, or
tyrosine residues. Gas phase fragmentation in the ion source or
the ion trap may also cause dehydration of a protein or a peptide,
although it most commonly happens during MS/MS analysis
when the collision energy is applied.18
Taken together, along with the quantitation of the target intact
protein, investigation of the extra information that could be lost
in the “bottom-up” approach could enable a new way to identify
biomarkers related to disease, populations, achievement of
efficacious dose, or other factors that could be critical for clinical
study design and data interpretation.
’ CONCLUSIONS
We report here the first attempt for protein bioanalysis using
HR LC
MS detection of the intact protein. The major advan-
tages of this strategy are specific detection of the intact protein,
independent of fragmentation efficiency, fully utilizing a high
resolving power to achieve good selectivity, and quantifying
proteins with PTM characterization. In this study, only one
charge state of the ionized protein was used for the example. As
the molecular weight of the protein increases and multiply
charged ions are more evenly distributed in intensities, feasibility
of utilizing the most abundant ion peaks by HR-MS detection will
result in further improvement of the sensitivity of the assay. Along
with more advanced instrument development for higher resolu-
tion, faster scan speed, higher charging efficiency, and the ad-
vancement of computer algorithms for signal summation, EIC
window optimization, and calibration of standard regression, this
analytical approach could provide further enhanced selectivity
and sensitivity and therefore become a more valuable approach
with wider applications in protein bioanalysis for drug
development.
’ ASSOCIATED CONTENT
bS Supporting Information. Standard curves and quality
control samples. This material is available free of charge via the
Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
*E-mail: Qin.Ji@BMS.com.
’ ACKNOWLEDGMENT
The authors would like to thank Mr. Huidong Gu for writing
an Excel Macro for the regression model and weighting selection
and Heidi Snapp for the assistance in manuscript preparation.
Analytical Chemistry ARTICLE

’ REFERENCES
(1) Ezan, E.; Bitsch, F. Bioanalysis 2009, 1 (8), 1375–1388.
(2) Ji, Q. C.; Rodila, R.; Gage, E. M.; El-Shourbagy, T. A. Am. Pharm.
Rev. 2004, 7, 84–89.
(3) Carr, S. A.; Anderson, L. Clin. Chem. 2008, 54, 1749–1752.
(4) Heudi, O.; Barteau, S.; Zimmer, D.; Schmidt, J.; Bill, K.;
Lehmann, N.; Bauer, C.; Kretz, O. Anal. Chem. 2008, 80, 4200–4207.
(5) Ji, C.; Sadagopan, N.; Zhang, Y.; Lepsy, C. Anal. Chem. 2009,
81, 9321–9328.
(6) Wu, S. T.; Ouyang, Z.; Olah, T. V.; Jemal, M. Rapid Commun.
Mass Spectrom. 2011, 30 (25), 281–290.
(7) Xu, Y.; Mehl, J. T.; Bakhtiar, R.; Woolf, E. J. Anal. Chem. 2010,
82, 6877–6886.
(8) Ji, Q. C.; Rodila, R.; Gage, E. M.; El-Shourbagy, T. A. Anal. Chem.
2003, 75, 7008–7014.
(9) Ji, Q. C.; Rodila, R.; Morgan, S. J.; Humerickhouse, R. A.;
El-Shourbagy, T. A. Anal. Chem. 2005, 77, 5529–5533.
(10) Ji, Q. C.; Gage, E. M.; Rodila, R.; El-Shourbagy, T. A. Rapid
Commun. Mass Spectrom. 2003, 17, 794–799.
(11) Kellie, J. F.; Tran, J. C.; Lee, J. E.; Ahlf, D. R.; Thomas, H. M.;
Ntai, I.; Catherman, A. D.; Durbin, K. R.; Zamdborg, L.; Vellaichamy, A.;
Thomas, P. M.; Kelleher, N. L. Mol. Biosyst. 2010, 6, 1532–9.
(12) Arnaud, C. H. Chem. Eng. News 2010, 88 (25), 10–15.
(13) Yoshimura, K.; Toibana, A.; Nakahama, K. Biochem. Biophys.
Res. Commun. 1988, 150, 794–801.
(14) Peters, C. W.; Kruse, U.; Pollwein, R.; Grzeschik, K. H.; Sippel,
A. E. Eur. J. Biochem. 1989, 182, 507–516.
(15) Taniyama, Y.; Yamamoto, Y.; Nakao, M.; Kikuchi, M.; Ikehara,
M. Res. Commun. 1988, 152, 962–967.
(16) Peeters, T. L.; Depraetere, Y. R.; Vantrappen, G. R. Clin. Chem.
1978, 24, 2155–2157.
(17) Taniyama, Y.; Yamamoto, Y.; Kuroki, R.; Kikuchi, M. J. Biol.
Chem. 1990, 265, 7570–7575.
(18) Wysocki, V. H.; Resing, K. A.; Zhang, Q.; Cheng, G. Methods
2005, 35, 211–222.

H dx.doi.org/10.1021/ac201540t |Anal. Chem. XXXX, XXX, 000–000