Article pubs.acs.org/ac
Sequential Protein and Peptide Immunoaffinity Capture for Mass Spectrometry-Based Quantification of Total Human β‑Nerve Growth Factor Hendrik Neubert,*,† David Muirhead,‡ Musarat Kabir,‡ Chris Grace,‡,∥ Adriaan Cleton,‡,⊥ and Rosalin Arends§ †
Pfizer Inc., 1 Burtt Road, Andover, Massachusetts, 01810, United States Pfizer Ltd., Ramsgate Road, Sandwich, Kent, CT13 9NJ, United Kingdom § Pfizer Inc., 445 Eastern Point Road, Groton, Connecticut 06340, United States ‡
S Supporting Information *
ABSTRACT: Nerve growth factor (NGF) is a neurotrophin that is implicated in the modulation of pain perception. Tanezumab, a humanized monoclonal antibody (mAb) specific for NGF, is highly potent in sequestering NGF and has demonstrated efficacy for treatment of chronic pain in clinical trials. We describe a novel, sensitive immunoaffinity liquid chromatography−tandem mass spectrometry (LC−MS/MS) assay for quantitative determination of human serum NGF levels at baseline and after tanezumab treatment. The assay combines magnetic bead-based NGF immunoaffinity enrichment using a non-neutralizing polyclonal antibody followed by digestion and quantitation of a NGF-derived tryptic peptide via high-flow peptide immunoaffinity enrichment and nanoflow LC−MS/MS. Following validation, the assay was employed to measure total NGF concentrations in samples from clinical studies. The assay had a 95% purity) was the tryptic NGFderived peptide IDTACVCVLSR conjugated to keyhole limpet hemocyanin (KLH) via an additional N-terminal cysteine (sequence CGGIDTACVCVLSR; GG added as spacer; C is carbamidomethylated cysteine). Typically, approximately 1 mg of purified antibody was used for building an immunoaffinity column. Assay Procedure. Bead-Based IA Enrichment of NGF Using Liquid-Handling Robotics. Calibrants were prepared by diluting the recombinant human NGF in 5% bovine serum albumin (BSA)/phosphate-buffered saline (PBS). An aliquot of 600 μL of serum (or 60 μL of serum diluted 1 in 10 in PBS for study samples containing NGF levels above 450 pg/mL), calibrants, or quality control (QC) sample was added to an Eppendorf LoBind deep well block (96 well; 1 mL; Eppendorf North America; Hauppauge, NY), followed by adding 20 μL of biotinylated NGF antibody (37.5 μg/mL) and 100 μL of 3-[(3cholamidopropyl)-dimethylammonio]-1-propane sulfonate (CHAPS) buffer (0.1 M ammonium formate, 0.5 M NaCl, 0.1% CHAPS pH 7.5). The plate was sealed and shaken at ∼4 °C overnight. Sample and magnetic bead-processing was performed on a liquid-handling robot (Hamilton Microlab Star and Starlet; Bonaduz, Switzerland), similar to previously described procedures.14,20,21 Bead separation was achieved using a Magnetight HT96 stand (Novagen; Madison, WI). As a general description of the robotic workflow, the magnetic 1720
dx.doi.org/10.1021/ac303031q | Anal. Chem. 2013, 85, 1719−1726
Analytical Chemistry
Article
standard added). Standard curves were obtained by plotting total quantity of detected NGF-derived peptide (fmol) against NGF concentration calculating a weighted (1/y2) multiparameter curve fit in Labstats (Excel add-in). The quality of the curve fit was evaluated by back-calculating NGF concentrations of calibrants. NGF concentrations reported throughout are derived from back-calculation using the calibration curves. Assay Validation. The following parameters were assessed during assay validation: Calibration with standards at 0, 7, 14, 28, 56, 112, 225, and 450 pg/mL NGF in 5% BSA/PBS (n = 2); imprecision/inaccuracy testing on 4 different days at four levels and with six repeats using validation sample (VS) Low1 (partially NGF immuno-depleted), VS Low2 (human serum pool, no spike), VS Medium (human serum pool + 75 pg/mL spike), and VS High (human serum pool + 350 pg/mL spike); the effect of mAb on recovery using VS Low1 and VS High (n = 6) additionally spiked with 0, 100, and 1000 ng/mL of tanezumab; selectivity using unspiked (neat) serum and serum spiked with 30 pg/mL NGF from six individual healthy volunteers (n = 2); freeze−thaw stability with four freeze−thaw cycles for VS Low1 and VS High at −70 °C (n = 6); stability at RT and 4 °C using VS Low1 and VS High (n = 6) for 0, 3, and 6 h at RT or 4 °C prior to analysis; long-term stability at −20 and −70 °C using VS Low1 (n = 6) stored for 3 months. Validation study acceptance criteria and additional validation sample preparation requirements can be found in the Supporting Information. Clinical Sample Analysis. This study was conducted in compliance with the Declaration of Helsinki and all International Conference on Harmonization Good Clinical Practice Guidelines. Patients provided written informed consent prior to initiation of protocol-specified procedures. In this study, serum samples were collected 1 h prior to dosing with tanezumab (baseline NGF, day 0), at 1, 4, and 12 h post dosing and on days 4, 7, 14, 28, 56, and 91. Typically, human sera were analyzed as single samples. Depending on whether NGF levels fell within the calibration range, samples were either analyzed undiluted or using a 1:10 dilution in PBS. In addition, seven NGF calibrants were analyzed in duplicates from 7.03 to 450 pg/mL in 5% BSA/PBS. A duplicate zero standard without NGF spike was also run. Furthermore, duplicate QC samples at the endogenous NGF level as well as 75 pg/mL and 350 pg/mL NGF spikes on top of endogenous NGF were analyzed in each batch. QC acceptance criteria were at an error of no more than ±25% from four of six QCs with at least one acceptable QC at each concentration. Dilution QC samples containing a 2-ng/mL NGF spike in human serum diluted 10-fold in 5% BSA/PBS were freshly prepared and analyzed in quadruplicates. Dilution QC acceptance criteria were at an error of no more than ±25% from three of four QCs.
Figure 1. Flow diagram of the immunoaffinity LC−MS/MS total NGF assay illustrating these main components: (i) antibody based enrichment of NGF from patient serum using streptavidin-coated magnetic beads; elution of NGF and (ii) addition of a SIL peptide; (iii) reduction, alkylation, and trypsin digestion of NGF to generate target and SIL peptides; and (iv) quantitation by high-flow peptide immunoaffinity linked online to nanoflow LC−MS/MS. LC−MS/MS, liquid chromatography tandem mass spectrometry; NGF, nerve growth factor; SIL, stable isotope-labeled.
■
RESULTS AND DISCUSSION NGF Assay Development. We developed a sensitive IALC−MS/MS assay for measuring total β-NGF (Figure 1). The assay was fully automated using 96 well plates and liquidhandling robotics. The general workflow of the assay can be divided into four main steps: (i) IA enrichment of NGF from serum using a noncompeting, biotinylated polyclonal reagent antibody which can bind to NGF in the presence of tanezumab and streptavidin-coated magnetic beads, followed by elution of NGF from the magnetic beads. This NGF IA enrichment
enhances the accuracy for absolute quantification and allows use of intact NGF protein as the calibration standard. (ii) Addition of synthetic stable-isotope-labeled peptide (NGF and SIL peptide sequence are shown in the Supporting Information, Figure S-2). (iii) Reduction, alkylation, and digestion of eluted NGF with trypsin to generate IDTACVCVLSR and its SIL analogue IDTACVC(V)(L)SR (V* [13C5; 15N] and L* [13C6; 1721
dx.doi.org/10.1021/ac303031q | Anal. Chem. 2013, 85, 1719−1726
Analytical Chemistry
Article
Figure 2. Schematic configuration of the online peptide immunoaffinity LC system used in the NGF assay. LC, liquid chromatography; NGF, nerve growth factor.
Figure 3. (a) Extracted ion chromatogram of IDTACVCVLSR from a 7.03-pg/mL NGF calibration standard and the corresponding SIL internal standard peptide. (b) Sample assay validation calibration curve ranging from 7 to 450 pg/mL NGF. NGF, nerve growth factor; SIL, stable isotopelabeled. 15
Briefly, the chromatography system incorporates three separation stages, the first using a column packed with rabbit polyclonal antibody against the NGF-derived peptide. Target peptide and its stable isotope-labeled analogue contained in the tryptic digest are enriched on the antipeptide antibody column, washed, and then eluted. Eluted peptides are then trapped on a reverse phase cartridge before being forward-flushed to a short reverse phase column into the nanospray source of the mass spectrometer (Figure 2). The sequential IA capture at the protein and at the peptide level ultimately resulted in a highly enriched and purified sample containing the peptide of interest for analysis via the mass spectrometer. Instrumental sensitivity was maximized by using low-resolution settings on both quadrupoles (Q1 and Q3) to enhance ion transmission while ensuring that, in this case, noise did not increase concomitantly and that no other SRM signals interfered with the measurement. Validation. The assay was validated, including assessment of inaccuracy and imprecision, selectivity, effect of tanezumab, freeze−thaw cycles, and stability as detailed in the Experimental Section. The calibration range is 7.03−450 pg/mL. An extracted ion chromatogram for a 7.03-pg/mL NGF calibrant and the corresponding SIL peptide is shown in Figure 3a. A
N]; total mass shift 13 Da). (iv) Quantitation of the NGFderived peptide using high-flow, high capacity peptide immunoaffinity enrichment linked online to nanoflow LC− MS/MS. The NGF capture antibody was used in large molar excess compared to the detected NGF concentration ranges and reagents. In addition, incubation conditions were selected such that NGF recovery was maximized. For confirmation, a second IA enrichment step performed on a sample that already underwent a first NGF IA enrichment typically yielded less than 5% of the original NGF signal indicating that adequate recovery for accurate quantification was achieved by using a single NGF IA enrichment step. The NGF derived near C-terminal sequence IDTACVCVLSR was selected as the target peptide for this assay because it is proteotypic, highly responsive in MS, and suitably antigenic to allow for generation of an antipeptide antibody. Furthermore, this sequence is conserved in NGF across species (i.e., human, monkey, dog, rat) such that the IALC−MS/MS configuration can also be employed for preclinical assays of NGF assisting with translational pharmacology. Characteristics of the online IA-LC−MS/MS assay configuration such as capture efficiency, scalable sensitivity, throughput, and robustness have been described previously.18 1722
dx.doi.org/10.1021/ac303031q | Anal. Chem. 2013, 85, 1719−1726
Analytical Chemistry
Article
Table 1. Summary Statistics: Intra- and Interassay Imprecision and Inaccuracy of the IA-LC−MS/MS Total NGF Assaya human serum
VS Low1
VS Low2
VS Medium
VS High
NGF target concentration (pg/mL)
15.5
26.5
101.5
376.5
batch N intrabatch mean (pg/mL) CV (%) inaccuracy (%) interbatch mean (pg/mL) CV (%) inaccuracy (%)
1 6
2 6
3 6
4 6
1 6
2 6
3 6
4 6
1 6
2 6
3 6
4 6
1 6
2 6
3 6
4 6
14.3 16.0 −7.2
13.6 12.7 −11.8
13.9 12.6 −9.8
15.7 9.7 1.5
26.5 18.5 0.1
23.1 8.2 −12.6
25.5 12.2 −3.5
24.3 12.9 −8.3
105 10.8 4.0
106 13.8 5.0
114 6.1 12.6
115 4.5 12.9
402 13.9 6.7
372 14.4 −1.4
387 12.1 2.8
410 13.3 9.0
14.4 13.3 −6.8
24.9 13.9 −6.1
110.2 9.6 8.6
392.6 13.1 4.3
a CV, coefficient of variation; IA, immunoaffinity; NGF, nerve growth factor; LC−MS/MS, liquid chromatography tandem mass spectrometry; VS, validation sample. NGF levels in the VS Low1 and VS Low2 samples were predetermined using this IA-LC−MS/MS assay prior to assay validation as 15.5 and 26.5 pg/mL, respectively, and defined as nominal values.
typical calibration curve obtained during assay validation is shown in Figure 3b. Intra- and interbatch imprecision and inaccuracy of validation samples are presented in Table 1. Assessment of intrabatch imprecision on four testing occasions using six replicates at each validation sample concentration produced a coefficient of variation (CV) ranging from 4.5% to 18.5%. Intrabatch inaccuracies ranged from 0.1% to 12.9%. Interbatch imprecision was 9.6% to 13.9% CV and interbatch inaccuracy ranged from 4.3% to 8.6% at all concentrations. Inaccuracy and imprecision in human plasma was also tested in a single batch using similar criteria as for human serum (Supporting Information). Imprecision was found to be ≤15.9% and inaccuracy ≤−13.8%. Other validation results can be found in the Supporting Information, including effect of tanezumab, selectivity, freeze/thaw, and benchtop and 3-month stability at approximately −20 and −70 °C. Basal NGF Levels in Healthy Volunteers and Preclinical Species. The amino acid sequence of full length NGF is well conserved between human and various monkey species (100% homology), whereas the sequence similarity between rat and human NGF sequence is approximately 86%. The polyclonal capture antibody raised against human NGF was confirmed to cross-react with NGF from various preclinical species including monkey, dog, and rat. Furthermore, the target tryptic peptide sequence IDTACVCVLSR, validated for detection of human NGF, is identical in all of the above species such that the IA-LC−MS/MS configuration could also be employed to measure basal NGF concentration not only in sera from healthy volunteers but also from cynomolgus monkey, dog, and rat. Mean (±SD) NGF concentrations were found to be 23.1 ± 9.7, 44.6 ± 8.2, 45.0 ± 17.3, and 30.5 ± 8.2 pg/mL, respectively (Table 2). Implementation in a Clinical Trial of Tanezumab. The assay was used to measure total NGF concentrations as part of a randomized, double-blind, dose-escalation, placebo-controlled clinical trial investigating efficacy of a single intravenous (IV) dose of tanezumab in Japanese patients with OA of the knee.22 For illustration of typical total NGF profiles measured in tanezumab treated patients, data are provided for the following cohorts: cohort A, tanezumab 10 μg/kg (n = 6) or placebo (n = 2); cohort B, tanezumab 25 μg/kg (n = 6) or placebo (n = 2); and cohort C, tanezumab 50 μg/kg (n = 6) or placebo (n = 2). Serum NGF samples were collected on days 1, 4/5, 8, 15, 29, 57, and 92.
Table 2. Basal NGF Serum Concentrations Observed in Healthy Volunteers and Various Preclinical Species species
human
Cynomolgus monkey
dog (beagle)
rat
mean NGF (pg/mL) SD % CV n min max
23.1 9.7 42.1 70 7.3 75.9
44.6 8.2 18.4 20 28.3 60.8
45.0 17.3 38.5 8 8.9 66.3
30.5 8.2 26.7 32 16.7 53.3
Interbatch assay inaccuracy, expressed as relative errors (% RE) of the estimated QC concentrations, ranged from −19.2% to −1.5% for endogenous (predetermined at 26.5 pg/mL), medium (endogenous +75 pg/mL NGF), and high (endogenous +350 pg/mL NGF) QC samples. Assay imprecision, expressed as % CV, was
