Quantification of peptides in human synovial fluid using liquid chromatography–tandem mass spectrometry.
Araceli Garcia-Ac
1, Sung Vo Duy
2, · Sébastien Sauvé
2· Florina Moldovan
3· V Gaëlle Roullin
4· and Xavier Banquy
1 *1
Canada Research Chair in Bio-inspired Materials and Interfaces, Faculty of Pharmacy, Université de Montreal, Montreal, QC
H3C 3J7, Canada.
2
Department of Chemistry, Université de Montréal, Montreal QC, H3C 3J7.
3
Centre Hospitalier Universitaire St. Justine, Université de Montréal, Montréal, Québec,
4
Pharmaceutical Nanotechnology Laboratory, Faculty of Pharmacy, Université de Montréal, Montreal QC, H3C 3J7, Canada.
* Corresponding author:
xavier.banquy@umontreal.ca
.
Abstract A method to explore the stability of two anti-inflammatory peptides in human synovial fluid (HSF) has been
developed and validated using liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). The two peptides
are BQ123 Cyclo(-D-Trp-D-Asp-L-Pro-D-Val-L-Leu, Mw = 610.7) and R-954 (AcOrn[Oic
2,( αMe)Phe
5, DβNal
7,
Ile
8]desArg
9-bradykinin, Mw = 1194.4). Human synovial fluid samples were analyzed after a protein precipitation step with
acetonitrile and dilution with mobile phase. DMSO was used as anti-adsorptive agent. We used an octyl silane column with
formic acid (0.1%, v/v) in water as the aqueous mobile phase and acetonitrile isopropanol-formic acid (20:80, 0.1 v/v) as the
organic mobile phase and 0.7 mL/min flow rate. The peptides CY-771 and pepstatin A were used as internal standards.
Selective detection was performed by tandem mass spectrometry with a heated electrospray source (HESI), operated in
positive ionization mode and in selected reaction monitoring acquisition (SRM). The method limit of quantification (injection
volume = 10 µL) was 0.17 ng and 1.2 ng, corresponding to 28 and 102 nmol L
-1for BQ123 and R-954 respectively in human
synovial fluid. Calibration curves obtained using matrix-matched calibration standards and internal standard were linear from
20 to 1000 nmol L
-1. Precision values (%R.S.D.) were ≤14% in the entire linear range. Accuracy measured at a low and a high
concentration level ranged from 93.1 % to 102 %. The recoveries (at 800 nmol L
-1) were 96.4 % for BQ123 and 102.0 % for
R-954. The method was successfully applied to follow the degradation kinetics of both peptides in human synovial fluid from
arthritic patients during 72 hours.
Keywords Quantitative bioanalysis · Tandem mass spectrometry ·Peptide quantification ·B1 receptor· Endothelin receptor
antagonist · Synovial fluid · Knee osteoarthritis · Distribution coefficient
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Introduction
Due to increasing insights in the diverse roles of peptides in physiological processes, peptide-based pharmaceuticals
are expanding as a relatively novel class of strategies for treatment. Indeed, peptides represent an attractive point for the design
of novel therapeutics. Their specificity has been observed to translate into excellent safety, tolerability, and efficacy profiles in
humans. The biological functions of peptides are multiple, including carrying signals between glands and cells, structural and
mechanical roles, controlling body functions, acting as chemical transporters, process regulators and many others [1]. They are
also used as drugs for the treatment of diseases, typical examples being insulin, growth hormones, and therapeutic antibodies
[2]. The unprecedented number of marketing approvals in 2012 for peptide therapeutics could be a harbinger for the innovative
peptide-based drugs [3]. New synthetic strategies for limiting metabolism and alternative routes of administration have
emerged in recent years and resulted in a large number of peptide-based drugs that are now being marketed [4]. In parallel, the
increasing importance of biopharmaceuticals necessitates performance improvements in bioanalytical techniques. In particular,
their quantification in complex biological samples such as human synovial fluid (HSF) requires acute sensitivity and
selectivity, as all biological matrices contain a countless number of proteins, all made of the same 20 amino acids (AA)
precursors. In the present study, we focused on two peptides of interest for the treatment of osteoarthritis, namely; BQ123
(Cyclo(-D-Trp-D-Asp-L-Pro-DVal- L-Leu) and R-954 (AcOrn[Oic
2,( αMe)Phe
5, DβNal
7, Ile
8]desArg
9-bradykinin).
Osteoarthritis is the most common form of arthritis, affecting millions of people in the United States and around the world. It is
a complex disease whose etiology bridges biomechanics and biochemistry [5]. No curative treatment exists currently and only
antiinflammatory drugs are routinely prescribed for patient relief. Recently, the therapeutic effects of peptides on arthritis have
been associated with reduction of the autoimmune and inflammatory responses [6, 7]. In that context, both BQ123 and R-954
have shown promising activities.
BQ123 is an endothelin receptor A (ETA) antagonist [8] which blocks the ETA receptor,
resulting in a decrease of lipid peroxidation products [9], increase of the reduced glutathione (GSH) level, and enhanced super
oxidase dismutase (SOD) activity [10]. Moreover, the blockage of the ETA receptor alleviates the lipopolysaccharide
(LPS)-induced oxidative stress [11]. LPS is the main causative agent inducing sepsis, stimulating macrophages to excrete large
amounts of inflammatory biomarkers.
On the other hand, R-954 is a potent, selective and stable peptide antagonist of the
inducible G protein-coupled receptors (GPCR) kinin B1. The kallikrein–kinin system is composed of two major proteolytic
systems that are responsible for the release of pro-inflammatory substances, particularly bradykinin. R-954 shows potential
applications for the treatment of several diseases, including cancer, neurological disturbances of diabetes [12] and
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
osteoarthritis[13]. Previous work showed that both endotelin-1 (ET-1) and bradykinin play a crucial role in vivo in OA
pathophysiology, through their catabolic functions in cartilage degradation and inflammation involving metalloproteases
MMPs (MMP-1-and MMP-13), nitric oxide (NO), and prostaglandins (PGE
2)[13, 14]. From a structural standpoint, both
peptides display significant chromophore groups for UV detection with conventional HPLC methods. Intrinsic peptide
fluorescence, originated with the aromatic amino acids tryptophan (TRP) and phenylalanine (PHE) [15] from BQ123 and
R-954 respectively, precludes analyses with spectrofluorometric methods. Nevertheless, those techniques are not appropriate
when only small sample volumes (≤ 200 µL) are available, such as the ones retrieved during in vivo studies in small rodents.
Therefore, the lack of a significant amount of synovial fluid led us to use LC-MS/MS. Determination of dietary supplements
and prescription drugs have been studied in HSF in a single class of compounds[16] [17, 18] or lipid study profiles[19].
However, to the best of our knowledge, no bioanalytical method for the determination of the BQ123 and R-954 peptides in
synovial fluid has ever been developed. One of the major difficulties faced when developing this analytical method was the
HSF composition. The presence of hyaluronic acid (HA) in HSF and the low concentration of proteins (10-30 mg/mL) [20],
which, unlike plasma (66-89 mg/mL) [20-22], can cause adsorption and loss of the studied analytes. The most abundant
macromolecules in synovial fluid are the sodium salt of hyaluronic acid (3 mg.mL
-1) and blood plasma proteins (albumin, ca.
11 mg.mL
-1and globulins, ca. 7 mg.mL
-1) [23] [24]. In this article, the nonspecific binding of the analytes to the wall of the
peptide containers was assessed. Studies on peptides have indicated considerable adsorption to solid surfaces even to
siliconized glass and polypropylene materials [25]. Such losses may be due to the hydrophobicity of the peptide, the
physicochemical properties of the plastic, and laboratory manipulations [26-28]. This article describes the development of a
sensitive, accurate, and reliable label-free procedure, based on liquid chromatography coupled to tandem mass spectrometry
(LC-HESI-MS/MS) is d for the selective and simultaneous quantification of two peptides in human synovial fluid. The overall
goal of our research was to monitor the stability of the R-954 and BQ123 peptides in a new intra-articular nanoscale drug
delivery system for the treatment of knee osteoarthritis (OA). In this context, a sensitive analytical method capable of detecting
concentrations in the low nanogram–per-mililiter range is essential for the monitoring of the target peptides in human synovial
fluid.
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
Experimental methods
Chemicals
BQ123, Cyclo(-D-Trp-D-Asp-L-Pro-D-Val-L-Leu), 95.51% pure was purchased from China peptides Co. R-954
(AcOrn[Oic
2,( αMe)Phe
5, DβNal
7, Ile
8]desArg
9-bradykinin) (purity >99% by HPLC) was kindly supplied by IPS Pharma,
Sherbrooke Canada. Two internal standards were used, named pepstatin A (PEP A), Isovaleryl-Val-Val-Sta-Ala-Sta purity ≥
90% obtained from Sigma–Aldrich Canada Ltd (
Oakville, ON,
Canada ) and CY-771 (CY),
Val-Trp-Glu-Leu-Leu-Ala-Pro-Pro-Phe-Asn-Glu-Leu-Leu-Pro (purity 98.60%) supplied by China Peptides Co (Shanghai, China). Formic acid (F.A., 98%
pure) and dimethyl sulfoxide (DMSO) were purchased from Sigma–Aldrich Canada Ltd. LC–MS grade acetonitrile (ACN),
H
2O and 0.1% F.A. in H
2O were purchased from J.T. Baker (Phillipsburg, NJ, USA). Human synovial fluid was obtained from
the Musculo-Skeletal tissues Bank, Centre Hospitalier Universitaire CHU Sainte Justine (project No 2252). The synovial fluid
was collected during aseptic arthrocentesis, arthroscopy or total knee replacement surgery of arthritic patients. For this study
we used five synovial fluids, collected and preserved in distinct tubes, anonymized and pooled before we used it. Samples were
centrifuged at 2,000×g for 10 min to remove any dispersed solid matter, and then frozen at −20 ◦C in polypropylene tubes as
five separate 800 µL aliquots.
Standard solution preparation
Stock solutions with a nominal concentration of 1 mg mL
-1of R-954 in methanol-water-DMSO (30/69/1 v/v %), BQ123 in
DMSO and PEP A in methanol-water-F.A. (30/69/1 v/v %) were prepared. A stock solution of the internal standard CY was
prepared at a nominal concentration of 0.7 mg mL
-1in methanol. To obtain the calibration curve and quality control (QC)
samples in neat solution, six intermediate solution standards of BQ123 and R-954 ranging from 0.02-10 µmol L
-1were
prepared in acetonitrile containing 1.7% DMSO. The internal standard (IS) working solution contained 5 µmol L
-1of CY and
PEP A in acetonitrile. These solutions were used to create a seven-point matrix-free calibration curve ranging from 0.02 to 1
µmol L
-1, by mixing 100 µL of intermediate solutions of BQ123 and R-954 (or 100 µL of acetonitrile for zero point) with 100
µL of internal standard solution (CY and PEP A) at 5 µmol L
-1with 800 µL of formic acid 0.1% in water.
The second calibration curve matrix-based was prepared with BQ123 and R-954 at concentrations ranging from 0.1 to 5 µmol
L
-1in HSF containing 1.7% DMSO and spiked with a volume of stock standard solution lower than 3% of the total HSF
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
volume. Working standards in HSF at 0.5 and 4 µmol L
-1were used to prepare QC samples.
The calibration curve and QC
samples were prepared by mixing spiked HSF with acetonitrile to precipitate proteins. To that extent, 40 µL of synovial fluid
containing a known quantity of peptide were precipitated with a total of 120 µL of acetonitrile including 20 µL of internal
standard. The sample was vortexed and centrifuged at 1,400 rpm for 10 min in a microcentrifuge Spectrafuge 16M from
LabNet international, New Jersey. Then, 80 µL of the supernatant were transferred into a disposable autosampler glass vial and
20 µL of 0.1% F.A. in water were added prior LC-HESI-MS/MS analysis. Final concentrations injected into LC-MS/MS were
0.02 to 1 µmol L
-1and 0.1 and 0.8 µmol L
-1for the calibration curve and QC samples, respectively. HESI source optimization
solutions of 5 µmol L
-1of each compound were prepared in F. A 0.1%- acetonitrile (50/50 v/v %). All stock and working
solutions in acetonitrile were stored at −20 ◦C in glass vials for no longer than 6 weeks, whereas the working solutions in HSF
and standard calibration curves were freshly prepared each day.
Stability of BQ123 and R-954 in human synovial fluid and buffer solutions
Stability studies of BQ123 and R-954 were carried out in human synovial fluid at 37°C. To investigate the stability, the
reaction mixture was treated with 1.7% v/v of DMSO and spiked with aliquots of 2 and 4 µL of stock solutions (1 mg mL
-1) of
BQ123 and R-954 to reach a final concentration of 8.5 µmol L
-1, respectively. The mixture was incubated at 37°C in a water
bath (Isotemp 210, Fisher Scientific). At specific time intervals (0, 5, 24, 29, 48, 53 and 72 h)
,
a 20.0-µL aliquot was removed
and processed for BQ123 and R-954 quantification as follows. 80 µL of IS working solution in acetonitrile were added to the
removed aliquot to precipitate proteins. The sample was then vortexed and centrifuged at 1,400 rpm for 10 min in the above
microcentrifuge. Thereafter, 30 µL of the supernatant were mixed with 70 µL of H
2O- 0.1% F.A. and transferred into
disposable 1.5-mL autosampler vials (fitted with an insert) to achieve a 0.5 µmol L
-1final concentration of the corresponding
peptide and IS for HPLC–ES-MS/MS analysis.
A similar study was carried out in three buffer solutions, 0.15 M sodium citrate buffer, pH 5 and 0.15 M phosphate buffers, pH
6 and 7.4 [29] at 37 °C. Reaction solutions were prepared with the described buffer solutions by adding 1.7% v/v of DMSO
and diluting 1 mg mL
-1stock solutions of BQ123 and R-954 with appropriate volumes of 0.15 M sodium citrate and 0.15 M
phosphate solutions to a final concentration of 8.5 µmol L
-1. The resulting spiked buffer solutions with BQ123 and R-954 were
incubated separately in a 37°C water bath as previously described. At 0, 5, 24, 29, 48 and 53 h following incubation start, a
10.0-µL aliquot was removed, diluted 20 times with acetonitrile-H
2O-FA (30:69.9:0.1 v/v), containing 0.5 µmol L
-1of internal
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
standard and achieving a 0.4 µmol L
-1final concentration for the peptides. Three calibration curves in 0.15 M sodium citrate
buffer, pH 5, 0.15 M phosphate buffers, pH 6 and 7.4 were also performed at the same time.
Instrumentation, LC–HESI-MS/MS analysis
Analysis was performed using a liquid chromatography-tandem mass spectrometry (LC–MS/MS) from Thermo Fisher
Scientific (Waltham, MA). It consisted of a LC Accela system (HTC Thermopal autosampler and Accela 1250 quaternary
pump) coupled with a heated electrospray ionization (HESI) source and a TSQ Quantiva™ Triple Quadrupole Mass
Spectrometer. The analytical column was a Luna C8 100 Å (5 µm particle size, 30 mm × 2.0 mm i.d.) protected by a
Phenomenex security guard cartridge C8 (4 mm × 2.0 mm i.d.); both columns were from Phenomenex (Torrance, CA). Mobile
phases were F.A. (0.1%, v/v) in water (solvent A) and acetonitrile-isopropanol (20:80 v/v) containing 0.1% F.A. (solvent B).
Separation was achieved at a 0.7 mL min
-1flow rate under gradient elution conditions with 5% of eluent B during 1.0 min and
rising from 5 to 98% in 1.5 min, holding for 1 min and then returning to 5% for re-equilibration during 1.4 min. Total run time
was 5 min for each injection. Injected sample volume was 10 µL in a full-loop mode. The analytical column was maintained at
28⁰C to improve reproducibility and retention time values. A typical chromatogram of BQ123, R-954 and their internal
standards showing both MRM transitions can be found in Fig. 1. Measure of the exact mass was performed using a Q-Exactive
Orbitrap mass spectrometer controlled by the Xcallibur 2.3 software (Thermo Fischer Scientific, Waltham , MA, USA).
Orbitrap parameters were set as follows: AGC (maximum capacity in C-trap) was set at 5x10
6, maximum injection time at 50
ms and resolution at 70, 000 FWHM at 200 m/z. The mass scan range was set at 150-1000 m/z (full scan MS mode).
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
Figure 1. (A) Chromatogram of a blank solvent injected immediately after the 1 µmol L
-1BQ123 and R-954 standard and their
respective internal standards at a concentration of 0.5 µmol L
-1showing the absence of carry-over after the injection of the
highest calibration curve level. (B) Chromatogram showing both MRM transitions of a solution standard in mobile phase
containing 0.25 µmol L
-1BQ123 and R-954 and their respective internal standards at a concentration of 0.5 µmol L
-1eluted
with the proposed gradient and mobile phase. (C) Chromatogram showing both MRM transitions of a standard solution in HSF
spiked with 0.25 µmol L
-1BQ123 and R-954, after the proposed extraction method by protein precipitation.
Mass spectrometry parameters
Heated electrospray ionization (HESI) was performed in positive mode using a spray voltage of 2.7 kV. The sheath and
auxiliary gas (N
2) were set to 35 and 5 arbitrary units, respectively. Ion transfer capillary temperature was set to 350 °C,
vaporizer temperature 400 °C, and ion sweep gas pressure to 2 arbitrary units. The tandem mass spectrometer was operated in
selected reaction monitoring (SRM) mode.
The first and third
quadrupoles
of the triple quadrupole MS/MS instrument were
operated at 0.7 FWHM
. In order to select precursor ions for each peptide, chromatograms were initially recorded in full scan
mode. In all cases, the pseudo-molecular [M+H]
+or [M+2H]
2+ion was selected. Once precursors were selected, full product
ion scans were recorded, and collision energies were optimized to get the maximum intensity of the obtained fragment ions.
The pressure of the collision gas in the collision cell was set to 1.5 mTorr. The most intense SRM transition (SRM#1) was
selected for quantification and the second most intense (SRM#2) was used for confirmation. SRM transitions, collision energy
(CE) and RF lens were compound-specific and are provided in Table 1
.
Table 1. MS/MS transitions parameters for the different peptide analytes.
Estimation of compound adsorption and lipophilicity constants
To investigate the loss of the target peptides during the solution preparation, the extent of adsorption and lipophilicity constants
(Log D) were calculated. First, adsorption was assessed in HSF, plasma, phosphate buffer (PBS), and mixes of organic
solvent-water (10% v/v) such as acetonitrile, acetone, and DMSO. Adsorption was measured at a low and a high concentration
level (0.05 and 1 µmol L
-1) of peptide by a simple quantitative experiment consisting of seven consecutive steps, transferring
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
the test solution from one polypropylene vial (0.5 mL) to another followed by LC–MS quantification. Three 20-µL aliquots of
each solution were pipetted from each test tube, diluted with mobile phase and followed by LC-MS/MS quantitation. HSF and
plasma samples were processed with acetonitrile precipitation and their corresponding dilution with mobile phase, as described
in the stability study section.
Theoretical calculation for the prediction of physical chemical properties for organic compounds can be used to estimate
lipophilicity constants. Nevertheless peptides widely vary, with large ranges of physical size and formal charge. In the case of
peptides, their physicochemical properties are determined by the analogous characteristics of the amino acids which constitute
them. The huge combination possibilities of these building blocks and their structural modifications cause hard-to-predict
behaviors with regard to physical properties, including solubility, pKa and adsorption
.
In a recent study comparing the
prediction capacity of different modelling tools, no statistically significant relationship between the length of peptide and their
respective lipophilicity descriptor was evidenced. The affinity constant values for cyclic peptides were under-predicted and
those of unblocked peptides were over-predicted [30]. Since R-954 and BQ123 may coexist in different forms (ionized or not)
according to the pH, the values of their molecular physical property log D were experimentally assessed. Direct measurement
by the shake flask method [31] in octanol/water was performed to measure solubility. Briefly, octanol and an appropriate
phosphate buffer in water were the standard solvents for the equilibrium experiments. Three stock bottle of high purity
analytical grade n-octanol with a sufficient quantity of the each tested buffers (pH 1.2, 4.5 and 6.8) were saturated. Both
solvents were shaken for 24 hours on a mechanical shaker. Solvents were then allowed to rest for two days to facilitate phase
separation. The active ingredients at a concentration of 25 mg mL
-1were perfectly solubilized in DMSO, 20 μL of this solution
were dissolved in 480 μL of n-octanol in 1.5 mL eppendorf tubes. They were stirred for 5 min on a Vortemp at 500 RPM,
followed by centrifugation for 5 min at 2000 RPM. A volume of 500 μL of buffer (pH 1.2, 4.5 and 6.8) was added and placed
on a mechanical shaker (Vortemp) for 72h at 25°C ± 1°C. The separation of the two phases was achieved by centrifugation
without temperature control. The resulting liquids were left for equilibration at 23°C ± 1°C for at least one hour before
analysis. The aqueous phase was sampled by minimizing the risk of including traces of n-octanol. To sample the aqueous
phase and avoid cross contamination with the organic phase a syringe initially filled with air was used. The concentrations of
the test substances in both phases were determined by HPLC-UV, using substance-specific methods. The total quantity of the
substance present in both phases was calculated and compared with the quantity originally introduced.
Sample preparation for the matrix effect study
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
Despite the use of internal standards to compensate the unexpected interferences from HSF with the analyte response, the
impact of matrix effects was specifically assessed. Suppression or enhancement of the signal was investigated using the same
mass spectrometric method, for each peptide. Six standards in neat solution (30 v/v acetonitrile in formic acid 0.1%) containing
an increasing concentration of BQ123 and R-954 (0.02, 0.2, 0.4, 0.6, 0.8 and 1 µmol L
-1) were injected in parallel with six
matrix standards spiked with the same increasing amounts of BQ123 and R-954 in HSF and following the proposed method
using acetonitrile to effect the deproteinization. A final concentration of 0.5 µmol L
-1of the internal standards was fixed in all
standards (in neat solution and matrix). Both calibration curves were injected in triplicate, two at the beginning and one at the
end of the sample list. The matrix effect was then determined by comparing the slopes of the matrix-matched calibration
curves to those in the neat solution[32].
Method validation
The validation was performed according to the EMA 2012 guideline on bioanalytical method validation [33]. During method
validation and real sample analysis, a blank of HSF was spiked with the two peptides of interest using them as references to
prepare calibration standards and quality control samples. In addition, internal standards for each of the peptides were added
during sample processing of the chromatographic runs. Selectivity, carry-over, method limit of detection (LOD), method limit
of quantification (LOQ), calibration range (calibration curve performance), accuracy, precision, and matrix effects were
evaluated. Selectivity of the method was proved by preparing and evaluating six different blank synovial fluid samples to
investigate potential interferences of endogenous compounds. The chromatogram of blank synovial fluid sample was compared
with those of HSF samples spiked with BQ123, R954 and their internal standards. The error tolerated on the relative retention
time of each of the tested peptides was ±0.5 min. Carry-over was addressed during method development; it was assessed by
injecting blank samples after a calibration standard at the upper limit of quantification. LOD and LOQ were determined as
respectively 3 and 10 times the standard error of the y-intercept divided by the slope of the calibration curve.
The calibration standards were prepared in HSF as the matrix of the study samples by spiking the blank matrix with known
concentrations of R-954 and BQ123 (range tested: 30-1194 ng mL
-1and 15-610 ng mL
-1). Six calibration concentration levels
were used, in addition to the blank sample (processed matrix sample without analyte and without IS) and a zero sample
(processed matrix with IS). Each calibration standard was prepared and analyzed in triplicate. The blank and zero samples
were not taken into consideration to calculate the calibration curve parameters. For each analysis, the calibration curve was
prepared using freshly-spiked standards. The accuracy was assessed on samples spiked with known amounts of the analyte, a
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
low level concentration 0.1 µmol L
-1and a high level concentration equal to 80% the maximal concentration of the calibration
curve, they were also used as QC samples. These QC samples were spiked independently from the calibration standards, using
separately prepared stock solutions. They were analysed against the calibration curve, and the obtained concentrations were
compared with the expected value. The precision, expressed as the coefficient of variation (CV), described the reproducibility
of individual measures of the analytes. As well, intra-day and inter-day precisions were evaluated for a low QC sample
concentration (0.1 µmol L
-1). For the validation of the within-run precision, three QC samples at a low level concentration were
prepared and injected the same day in a single run. For the validation of the between-run precision, the same process was
followed by performing evaluations during three different days. These samples were measured in triplicates on 3 different days
to assess linearity, accuracy, precision (intra-day and inter-day), LOD, LOQ. Each replicate was injected once in each series of
injections together with a six-point based calibration curve ranging from 0.02 to 1 µmol L
-1. The BQ123 and R-954
concentration was calculated based on the peak area
ratio (analyte/IS)
of each SRM transition and using the mentioned
calibration curve. Resulting concentrations were computed using the embedded Xcalibur software 3.0 (Thermo Corp., USA).
3. Results and discussion
Method Development
Sample pre-treatment
Endogenous proteins were successfully eliminated through protein precipitation resulting in good recoveries,
meaning that the therapeutic peptides were not totally eliminated in the precipitated protein pellet, with recoveries
more than 96% of the initial mass. HSF was mixed with different parts of a water-miscible solvent, acetonitrile or
methanol, followed by sample centrifugation. The best results were obtained using three parts of acetonitrile to one
part of sample, producing a 4 and 12 fold higher response than using methanol for BQ123 and R-954, respectively.
Protein precipitation with acetonitrile, the best choice according to the composition and protein content of synovial
fluid, was additionally easy to perform, and required inexpensive chemicals and reagents.
MS/MS optimization
The full scan MS and ion product spectra of BQ123 and R-954 are shown in Fig. 2. Addition of 0.1% (v/v) formic
acid to the mobile phase improved the response of the four peptides compared to other mobile phase additives. For
this study, the LOQ for R-954 and BQ123 had to be as low as possible and MS settings were especially optimized
for these two analytes. Therefore, multiple product ions were selected for SRM analysis and the Q1 resolution was
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
maintained at 0.7 in order to keep the high absolute intensity at the LOQ concentration level. Heating of the
ESI-probe successfully helped to evaporate the solvent without degradation of the compounds and improved the signal of
the peptides at the applied flow rate of 700 µL/min. BQ123 generated the single charged ion as the most abundant
peak, [M+H]
1+at m/z = 611.3 in full scan MS (fig. 1A, left). The
13C peaks (insert) with a 1 unit of difference on the
m/z scale showed that the charge state of the peptide ion that produce this peak cluster was 1. Following the selection
of the precursor, the collision energy was optimized and the full product ion scan was recorded (fig.1A, right). The
fragment ions obtained with maximum intensity were at m/z = 169.2 and m/z= 197.2. The ion product mass
spectrum of the protonated molecule showed fragments arising from a five cyclic ring-opened peptide. A stepwise
fragmentation of BQ123 by loss of aspartic acid (D), triphtophane (W) and leucine (L), produced the asylum
H-Proline (P)-Valine (V)
+ion, m/z = 197.1. The spectrum was dominated by the m/z = 169.2 ion formed from the loss
of the elements of CO of the fragment m/z = 197.1.
R-954, with monoisotopic mass 1193.6 u, gave the double-charged ion [M+2H]
2+as the most abundant peak at m/z =
597.9, figure 1B-left. The insert shows the typical isotopic pattern for small peptides and their double-charged ions
with a distance between the main peak and the isotope of 0.5 amu. Biological molecules as R-954 of mass 1000 units
and larger show a considerable contribution of the
13C isotope [34], due the naturally occurring
12C/
13C ratio (98.89%
12C, 1.11%
13C). Fragment ions with maximum intensity obtained after collision energy optimization were at m/z =
170.0, and m/z= 134.3. Both fragments are immonium ions derived from the detachment, breakdown and the loss of
CO of the individual amino acids naphthyl-Alanine (naphthyl-A) and methyl-phenylalanine (methyl-F), respectively.
Due to the good charge stabilizing properties of the conjugated pi-electron systems, the most stable products ions are
those coming from the two amino acids with aromatic rings, giving rise to intense peaks [35]. This was confirmed by
measuring the accurate mass of the fragments using the Q-Exactive Orbitrap mass spectrometer and calculating the
mass deviations reporting the single mass measurement error [36] of 2.1 ppm for m/z= 170.1 and 2.0 ppm for m/z=
134.3.
Figure 2. Full scan MS (left) and ion product (right) mass spectra of the protonated ion under positive mode for
BQ123 [M+H]
+(A) and the doubly charged R-954 [M+2H]
2+(B). Inserts showing the isotope distribution of
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
BQ123 and R-954 taking into account the naturally occurring
12C/
13C ratio, (monoisotopic mass 610.3 u , C
31
H
42N
6O
7and 1193.6 u, C
61H
87N
13O
12, respectively). MS settings were as listed in table 1.
Liquid chromatography
Obtaining a fast analysis with good chromatographic performance for a mixture of greatly different peptides, each
showing distinct characteristics and in presence of certain endogenous compounds from the matrix, was quite
challenging. The BQ123 peptide showed high retention and poor peak shape using a C18 analytical column, even at
higher percentages of organic solvent in either the initial mobile phase or the injection solvent. The hybrid
polymer-silica C8 column allowed for a good retention of the four compounds in comparison to the other stationary phases
used. The length of the chromatographic run was extended to five minutes, which is sufficient for the elution of the
most retained sample components including a 1-min re-equilibration of the column at the initial chromatographic
conditions. One minute of aqueous phase was run before starting the gradient, to eliminate interferences of
compounds with greater polarity eluting at the same time as the peptides of interest. Moreover, the compounds were
eluted in a mobile phase containing a large amount (80%) of organic solvent, thus ensuring a high ionization
efficiency. The LC–HESI-MS/MS analysis showed suitable reproducibility of the chromatographic separation, and
no shortening or variability of the retention times was observed even after injection of large sample sequences, up to
80–100 injections. In the optimized analytical conditions the mean retention time of BQ123 and R-954 were of 2.27
± 0.03 min and 1.95 ± 0.02, respectively (%R.S.D. = 0.6% and 0.9, n = 20).
Estimation of peptide adsorption to solid surfaces and lipophilicity constants
One of the pitfalls encountered during the early development of this methodology was the immediate adsorption of
the two target peptides to the materials commonly used in the laboratory manipulations. Numerous factors could
impact the compound observed concentrations, accuracy, and precision of a bioanalytical method, including, but not
limited to, matrix effects and losses by adsorption. The first considered factor was the nonspecific binding of the
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
analytes to the wall of the peptide containers. To overcome this aspect and avoid severe underestimation of the
compounds’ concentrations in the assays, an adsorption study containing different anti-adsorptive agents was carried
out. The applied approach was a simple quantitative procedure previously used and described above [37, 38].
Several types of anti-adsorption agents, such as BSA[39, 40], acetonitrile[37], tween 20[41], tween 80[42] and
DMSO [43] among others, have been used for blocking adsorption in HSF and urine assays. In the present study,
DMSO, acetone, and citric acid were tested as anti-adsorptive agents. As observed in figure 3 (6
thpair of bars),
DMSO was identified as the best anti-adsorptive as samples pre-treated with DMSO helped to avoid losses,
producing a relative recovery percentage of 96.4% for BQ123 and 102.0% for R-954.
Figure 3, displays ten pairs of bars showing the relative loss by adsorption in four untreated samples, the test with
four antiadsorption agents and two treated samples with DMSO. Each pair of bar correspond to one experiment in
one matrix showing the response of a first bar without transfer and a second one after seven transfer steps in
polypropylene tubes. All the test were performed at a low level concentration of 0.05 µmol/L, except for the second
pair of bars, it was spiked to 1 µmol/L named high concentration. The first four pairs of bars, correspond to HSF and
plasma prepared without the antiadsorptive agent DMSO. The next four, correspond to the tests with the four
antiadsorptive agents. The last two pairs of bars, are the results when the HSF and PBS pH 6.5 were treated with
DMSO.
The relative loss by adsorption (-35%) was pronounced in untreated HSF samples at low concentration (first pair of
bars), due to the limited binding capacity of the wetted solid surface area, , as previously reported [37] and also,
probably due to HSF high salts levels and relatively low protein concentration. HSF originates from plasma, with
small molecules and smaller proteins freely diffusing from the plasma into the HSF. Although larger molecular
weight proteins are excluded, total protein concentration is somewhat lower in plasma [44]. Another reason is the
high ionic strength of HSF (0.16M)[45]; the high concentration of electrolytes in HSF may produce a “salting out”
effect, driving the studied compounds further to the hydrophobic interface. This is confirmed by the results obtained
in the PBS pH 6.5 test (4
thpair of bars), where the concentration of salts is high and there is no protein. Adsorption
effects were not observed in plasma (3
rdpair of bars), which may be due to the strong interaction protein –peptides,
such as hydrogen bonding and hydrophobic interactions [46].
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
Figure 3. Relative recovery of BQ123 (above) and R-954 (below) without transfer step and after seven transfer steps
in polypropylene tubes. Both peptides concentration 0.05 µmol L
-1, * high concentration 1 µmol L
-1. After every 45
min storage at 10°C, 200 µL were transferred from one reaction vial, the first and seventh transfer were measured by
LC-MS/MS. Initial concentration of DMSO: 1.7 % (v/v) in the treated samples.
To elucidate the propensity to partition between the polypropylene surface of the transfer vials and the aqueous
matrices, HSF or phosphate buffers, the distribution coefficient (D) was experimentally assessed. D is the
appropriate descriptor of lipophilicity of ionisable compounds that may exist in different forms (ionized or not) at
any given pH, such as R-954 and BQ123, determining their propensity to adsorb to hydrophobic surfaces such as
plastic [47].
Distribution was compared to percentage of loss. As showed in figure 4, a good correlation was observed between
the percentage loss and log D for both peptides at pH 1.2, 4.5 and 6.8. In our hands a log D of 1.6, at its neutral
form was experimentally obtained, showing very good agreement with Paladino et al[48], that have reported a log P
= 1.55 for BQ123.
Figure 4. Percentage loss and log D as a function of pH for BQ123 and R-954 in buffer phosphates pH=1.2, 4.5 and
6.8, not treated with DMSO. Each point represents the mean of triplicates. Full lines and dashed lines are the
corresponding results of BQ123 and R-954, respectively.
Matrix effects
As depicted in Fig. 5, the presence of a matrix effect was assessed by comparing the slopes of both neat solution and
HSF calibration curves (both treated with DMSO 1.7% v/v) by means of a Student’s t-test. By comparing the
obtained t values, 4.033 for BQ123 and 1.341 for R-954 with their tabulated t (2.003), the study showed that R-954
was not subjected to matrix effect, while BQ123 was susceptible to signal enhancement. To reduce matrix effects,
internal calibration and matrix-matched calibration curves using HSF standards were used for the rest of the
validation study.
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
Figure 5: Comparison of HSF (dotted line) and neat solution (continuous line) calibration curves of BQ123 and
R-954 obtained during the matrix effect assessment.
Method validation
Method validation results are summarized in Table 2
.
Linearity in neat solution and HSF was suitable for BQ123
with determination coefficients R
2≥0.99 and acceptable for R-954 (R
2>0.98). The method intra-day variation for
R-954 was lower than 18%, while inter-day precision was between 5-13 %. Extraction recovery achieved for both
peptides was suitable (>95%). The LOD ranged from 0.008 to 0.031 µmol L
-1, while the LOQ ranged from 0.028 to
0.102 µmol L
-1. During preliminary experiments, external calibration without IS correction was used, and a higher
degree of variability was observed (data not shown). Therefore, it was decided instead to consider a further IS
correction to reduce variability to acceptable values (<15%).
Table 2. Method validation parameters for the analysis of BQ123 and R-954 in HSF and neat solution.
1
Results obtained using a calibration curve with HSF standards and following all the process.
2Results obtained using a calibration curve in neat solution treated with 1.7% of DMSO (%v/v).
aEquation and coefficient of determination for the calibration curve using SRM#1
b
Accuracy of samples (n=3) spiked for a final concentration of 0.1 µmol L
-1for hsf and mobile phase the same day.
cAccuracy of samples (n=3) spiked for a final concentration of 0.8 µmol L
-1the same day.
d
Relative standard deviation of analytes in hsf and neat solution (n=3)3 spiked for a final concentration of 0.1 µmol
L
-1the same day.
e
Relative standard deviation of analytes in hsf and neat solution (n=3) spiked for a final concentration of 0.1 µmol
L
-1on 3 different days (N=12).
f
Sample spiked for a final concentration of 0.8 µmol L
-1(n=3).
g