AI-64310-CRFT

526 / 658

2 Data Analysis
All data were processed usin (revision
1.4). HRAM MS data extraction was used for
quantification. To provide additional levels of qualitative
analysis, the three most abundant precursor charge states
per insulin variant were used, as well as the six most
abundant isotopes per charge state. A mass tolerance of
±7 ppm was used to extract all data. Qualitative scoring
was based on mass error, precursor charge state
distribution, isotopic overlap, and corresponding LC
elution peak profiles measured for each sample. Product
ion data were used for sequence verification. The
measured area-under-curve (AUC) values for porcine
insulin were used as the internal standard for all samples.
Results and Discussion
To assess the workflow, the insulin variants were spiked
into two different matrices and processed. The effects of
the matrix, competitive binding/extraction of all insulin
variants, and automated data extraction, verification, and
quantitation were evaluated. The HRAM data acquisition
capability of the Q Exactive mass spectrometer enabled
downstream automated qualitative and quantitative data
processing using Pinpoint software. By acquiring data in a
nontargeted manner, post-acquisition methods can be used
to process the data for any insulin variant sequence or
modification. To increase the qualitative information
obtained, multiple target-specific attributes per insulin
variant were chosen for analysis by the software.
Qualitative Data Processing Strategy
Figure 1 shows the base peak chromatogram for the
human plasma sample spiked with 960 pM of Lantus and
Apidra insulin analogs and 50 pM of porcine insulin
extracted using MSIA. The data acquisition time range
used was 3.5 to 5.5 minutes. The chromatographic trace
shows two peaks eluting, with the peak at 4.52 minutes
attributed to the Lantus insulin analog and the peak at
4.72 minutes attributed to porcine, human, and Apidra
insulin analogs. The inset shows the averaged HRAM
mass spectrum around the +5 precursor charge states for
the insulin analogs. The observed relative abundance of
Apidra to porcine (ca. 6%) was in close agreement with
the spiked amounts of 960 to 50 pM, respectively. The
observed relative abundance of endogenous human insulin
was equivalent to that of porcine. The remaining peaks in
the mass spectrum were attributed to adduct formation
during ionization. Despite the large difference in the
amounts present in the plasma samples, there was little
interference observed when detecting all insulin variants.
The resolution of the Q Exactive mass spectrometer was
more than sufficient to baseline resolve the isotopic
profiles for the +5 charge state across the dynamic range.
Sample Preparation
Two sets of samples were prepared. First, a dilution series
of Humulin S, Apidra, Lantus, NovoRapid, and bovine
insulin, prepared in the presence of porcine insulin
(50 pM) and covering an analytical concentration of
1.5 to 960 pM, were spiked into a phosphate-buffered
saline-bovine serum albumin (PBS/BSA) matrix. The
second set consisted of Apidra, Lantus, and NovoRapid
spiked individually into human plasma at the same
concentration range (1.5 to 960 pM). For quantitation
curve development, both Apidra and Lantus were spiked
into plasma across the same concentration range
(1.5 to 960 pM). Porcine insulin was again spiked into
each sample at 50 pM as an internal standard.
Mass Spectrometric Immunoassay
The affinity capture of insulin was achieved using insulin-
specific MSIA D.A.R.T.’S mounted onto the 96-channel
pipetting head of the
fter rinsing the insulin MSIA
D.A.R.T.’S with 15 cycles of a single aspiration and
dispensing 150 µL 10 mM PBS, the insulin MSIA
D.A.R.T.’S were immersed into the samples and
100 aspiration and dispense cycles of 250 µL were
performed. The multiple cycles allowed simultaneous
affinity enrichment of all of the insulin analogues as well
the internal standard. The MSIA D.A.R.T.’S were then
rinsed with PBS (15 cycles) from another microplate,
followed twice by water (15 cycles) from two additional
microplates (150 µL aspirations and dispenses, from
200 µL in each well).
The affinity-captured insulin analogs were eluted to a
microplate by aspirating and dispensing 80 µL of
15 μg/mL ACTH 1-24 in 33% acetonitrile/0.4% (v/v)
trifluoroacetic acid (TFA) 100 times from a total of
100 µL volume in each well. The eluates were dried down
in a ntil dry
and then resuspended in 100 µL reconstitution buffer
25% acetonitrile/0.2% formic acid (v/v)). The microplate
was sealed and vortexed for 30 seconds to ensure proper
reconstitution, and then spun-down prior to loading
samples onto the LC.
LC/MS Method
Samples were analyzed using a generic LC/MS method.
A
system was used for all LC/MS experiments. To begin,
100 µL of each sample was separated on a 1 x 250 mm
using a
linear gradient (10–50% in 10 minutes) comprised of
A) 0.1% formic acid in water and B) 0.1% formic acid in
acetonitrile. The column was heated to 50 ºC.
All data were acquired using a
perated in data-dependent/dynamic
exclusion mode. A resolution setting of 70,000 (FWHM)
at m/z 200 was used for full-scan MS and 17,500 for
MS/MS events. Full-scan MS data were acquired using a
mass range of 800–2000 Da. A targeted inclusion list was
used to trigger all data-dependent events.