A generic sample preparation approach for LC–MS/MS bio- analysis of therapeutic monoclonal antibodies in serum applied to Infliximab

In order to quantify proteins in complex biological matrices such as serum or plasma, LC-MS/MS is finding its way to the bioanalytical community [1-6] as a complementary method to ligand binding assays (LBA). Advantages of LC–MS over LBA are that no highly selective biological reagents have to be developed, which can take up to six months to generate and possibly less complications due to immunogenicity [4,7-10]. In contrast to LBAs, the conventional standard for therapeutic protein bioanalysis, which measures the unbound intact 3D domain of the therapeutic protein binding to the assay’s capture reagent, LC–MS measures a unique part of the primary structure of the protein. We developed a generic workflow consisting of immunoprecipitation, using protein A capture followed by LC–MS/MS quantification of a unique signature peptide after tryptic digestion of the target therapeutic monoclonal antibody (mAb) in biological matrices [11]. To optimize the generic approach the “seven critical factors” approach was followed [5]. Essential to the generic methodology are magnetic beads, functionalized with protein A, a 42 kDa surface protein originally found in the cell wall of the bacterium Staphylococcus aureus which binds immunoglobulins. Protein A can be suitable to purify classes, subclasses and fragments of immunoglobulins as well as immunoglobulin complexes [12] when the immunoglobulin of interest has a high affinity for protein A as compared to other affinity capture materials. As a result of the ability of Protein A to capture imJOURNAL OF APPLIED BIOANALYSIS, Jan. 2015, p. 26-34. http://dx.doi.org/10.17145/jab.15.005 Vol. 1, No. 1


Introduction
In order to quantify proteins in complex biological matrices such as serum or plasma, LC-MS/MS is finding its way to the bioanalytical community [1][2][3][4][5][6] as a complementary method to ligand binding assays (LBA). Advantages of LC-MS over LBA are that no highly selective biological reagents have to be developed, which can take up to six months to generate and possibly less complications due to immunogenicity [4,[7][8][9][10]. In contrast to LBAs, the conventional standard for therapeutic protein bioanalysis, which measures the unbound intact 3D do-main of the therapeutic protein binding to the assay's capture reagent, LC-MS measures a unique part of the primary structure of the protein. We developed a generic workflow consisting of immunoprecipitation, using protein A capture followed by LC-MS/MS quantification of a unique signature peptide after tryptic digestion of the target therapeutic monoclonal antibody (mAb) in biological matrices [11]. To optimize the generic approach the "seven critical factors" approach was followed [5]. Essential to the generic methodology are magnetic beads, functionalized with protein A, a 42 kDa surface protein originally found in the cell wall of the bacterium Staphylococcus aureus which binds immunoglobulins. Protein A can be suitable to purify classes, subclasses and fragments of immunoglobulins as well as immunoglobulin complexes [12] when the immunoglobulin of interest has a high affinity for protein A as compared to other affinity capture materials. As a result of the ability of Protein A to capture im-munoglobulins it is suitable to concentrate therapeutic monoclonal antibodies which are usually immunoglobulin based. The digestion is combined with disulfide bond reduction and alkylation. To demonstrate the feasibility of this approach, Infliximab (trade name Remicade), a chimeric monoclonal antibody against tumor necrosis factor alpha (TNF-α) used to treat autoimmune diseases, was used [13].

Chemicals and reagents
Infliximab was obtained from UMC Utrecht as 10 mg/ ml solutions. The concentration was verified using Remicade obtained from a local pharmacy. TPCK-trypsin was from Thermo Scientific (Breda, the Netherlands). An 18-mer synthetic signature peptide corresponding to a tryptic fragment (DILLTQSPAILSVSPGER) of the variable part of the light chain of the therapeutic antibody was synthesized by Sigma-Aldrich (Zwijndrecht, the Netherlands). Its stable isotope labelled analogue (DILLTQSPAILSVSPGER[ 13 C 6 15 N4]), applied as internal standard, was also synthesized by Sigma-Aldrich. Rat serum was obtained from Harlan (Horst, the Netherlands). PureProteome protein A magnetic beads were obtained from Millipore (Amsterdam, the Netherlands). The other chemical reagents and solvents used were obtained from Merck (Amsterdam, the Netherlands; acetic acid, hydrochloric acid, sodium chloride), Biosolve (Valkenswaard, the Netherlands; acetonitrile, formic acid, methanol), Fluka (Zwijndrecht, the Netherlands; ammonium bicarbonate, calcium chloride and sodium hydroxide), Sigma-Aldrich (dithiothreitol, glycine, iodoacetamide, monosodium phosphate, trifluoroacetic acid) and Calbiochem (Amsterdam, the Netherlands; tris(hydroxymethyl)aminomethane).

Preparation of calibration and QC samples
Calibration standards were prepared by serial dilution of Infliximab to 0.05, 0.10, 0.20, 0.50, 1.0, 2.5, 10 and 25 μg/ml using rat serum from a 10 mg/ml stock solution. Quality control samples (QCs) were used to evaluate the accuracy and precision during the qualification and were prepared in the same way as calibration standards at 0.10, 0.25, 2.0, and 10 μg/ml. The stable isotope labelled analogue DILLTQ-SPAILSVSPGER [ 13 C 6 15 N 4 ] used as internal standard (IS) was added to calibration standard and QCs during sample preparation, after elution from the beads. All samples were submitted to the immunocapture and digestion steps prior to LC-MS/MS analysis.

Immunocapture and Digestion
For each sample 50 µl protein A magnetic beads was used. After removing the storage buffer and washing, 25 µl sample volume was loaded (60 min) on the beads in 200 µl PBS (phosphate buffered saline). A magnetic stand (Millipore) was used to separate the beads from the solution. After washing, proteins were eluted from the beads using 50 µl 100 mM glycine, pH 2.5. After removal of the beads, the pH was raised with TRIS (pH 8) and internal peptide solution was added (24 ng/ml end concentration). After reduction with dithiothreitol (50 mM end concentration) and alkylation with iodoacetamide (85 mM end concentration), trypsin (20 µg) was added in ammonium bicarbonate and the samples were incubated for 2.5 hours at 37°C. Finally, acetonitrile, formic acid and trifluoroacetic acid were added at final concentrations of respectively 10%, 1% and 0.5% to give an end volume of 205 µl. Samples were transferred to an HPLC vial and injected (10 μl) into the chromatographic system.

Optimization of trypsin digestion
Digestion was carefully optimized since many factors, such as pH, ionic strength, temperature, and incubation time, can influence the efficiency of proteolysis.
We chose TCPK-trypsin, which exhibits limited self-digestion and retains its activity over a long period. The optimal protein digestion was obtained at pH~8 in 100 mM ammonium bicarbonate buffer at a temperature of 37°C for 2.5 hours using 20 μg of trypsin. Protein denaturation was obtained by heating to 95°C, reduction was performed by addition of 50 mM dithiothreitol and incubation for 20 minutes at 56°C and alkylation was optimized by addition of 85 mM iodoacetamide and incubation for 30 minutes at ambient temperature in the dark.
LC-MS quantitation of signature peptide UPLC Chromatography was performed on an Acquity UPLC (Waters) using an Acquity HSS T3, 50 x 2.1 mm, 1.8 µm column (Waters). The column was eluted at a flow rate of 0.8 ml/min using as mobile phase A MilliQ water and as mobile phase B acetonitrile, both containing 0.1% (v/v) formic acid. The analytical gradient profile was as follows: The column temperature was 50°C, and the injection volume was 10 μl.

Quantification of Infliximab by the LC-MS/MS method in spiked rat serum
Calibration curves were established using linear regression and 1/x 2 weighting, plotting the Infliximab concentration in the calibration standard samples (from 0.05 to 25 μg/ml) vs. the peak area ratio of the signature peptide from Infliximab and the signature peptide IS. Accuracy and precision of the method were evaluated during one day by analyzing the four QC samples (0.1, 0.25, 2.0, and 10 μg/ml) in five replicates. Assay accuracy (ratio of the measured concentration to the nominal concentration multiplied by 100) and precision (percent relative standard deviation, %RSD) were calculated at each concentration. The combined recovery of the immunocapture and digestion step at each concentration was estimated from the signal of the signature peptide obtained after Infliximab extraction from rat serum followed by peptide release after tryptic digestion multiplied by molecular mass ratio of the signature peptide (multiplied by 2) and Infliximab and comparing this with the signal obtained for the added signature peptide IS.

Selection of peptides for quantification
The strategy for quantification involved the selection of tryptic peptides from the variable regions of both the light and heavy chains. The corresponding sequences were submitted to similarity searches (Protein BLAST) [14] to ensure detection specificity. Table 1 lists all the potential peptides after in-silico digestion and BLAST searches vs. the species Rattus Norvegicus.
The following peptide selection criteria were applied: -Length between 6 and 20 amino acids -No adjacent tryptic cleavage sites -No methionine or cysteine in the peptide sequence -No asparagine followed by glycine or serine -Peptide contains part of variable domain There are several other relevant peptide selection criteria, which were not found to be relevant for Infliximab and were thus not applied. Table 2 lists the results of the database search. The first practical step of the development was to identify specific peptides with the most intense MS signal, among those produced by digestion. These different peptides were monitored in an exploratory LC-MS run of an Infliximab digest using the Q-Exactive in the full scan mode. Peptide DILLTQSPAILSVSPGER exhibited a signal approximately three times higher than peptides ASQFVGSSIHWYQQR and TVAAPSVFIFPPSDE-QLK. Finally, peptide DILLTQSPAILSVSPGER was selected for further experiments based on sensitivity and chromatography (e.g. peak shape), see Figure 2. A stable isotope labeled analogue of DILLTQSPAILSVSPGER was synthesized to serve as internal standard.

Immunocapture of Infliximab
We selected magnetic beads coated with protein A to pull down Infliximab via its human Fc part. The approximate absolute recovery observed after all the purification, re- Reason for not selecting a peptide (not exhaustive) or assignment as candidate. Parts of the variable regions are in red, parts of the CDR (complementarity determining regions) are in bold and underlined. Sequence obtained from Application of a mass-spectrometric method for the N-glycosylation analysis of immunoglobulin G molecules, doctoral thesis by DI Johannes Stadlmann [15]. X = unknown amino acid.  Where R is the recovery in %, C prot the concentration of protein in serum in ng/ml, MW pep the average molecular weight of the signature peptide in Da, MW prot the average molecular weight of the protein in Da, D prot the protein dilution factor or the ratio between sample end volume and loading volume, C IS the concentration of the added IS solution in ng/ml, D IS the IS dilution factor or the ratio between end volume and added volume, S anal the sig-nal obtained for the analyte signature peptide and SIS the signal obtained for IS peptide.The results shown in this paper indicate that even with a relatively low absolute recovery, accurate results (with good relative recovery) can be obtained when quantifying proteins using calibration samples which have been prepared with the analyte protein in the matrix of interest and subjecting these to the same conditions as the study samples and QCs. A way to further optimize the repeatability of the method would be to load the protein A beads with stable isotope labeled protein internal standard instead of fortifying with peptide internal standard after elution of the proteins from the beads [16,17]. The parameter which requires further optimization in the presented method is the elution step from the protein A beads. When the protein A beads were subjected to another round of tryptic digestion a significant residual signal was observed for DILLTQ- 1945.020 Yes For GLEWVAEIR a hit was obtained for rat which also exhibited an tryptic cleavage site N-terminally to the hit. SPAILSVSPGER (results not shown). In our experience, the elution step from the protein A beads was also critical in the method development for other monoclonal antibodies (results not shown). It is recommended to optimize these conditions instead of using the standard elution conditions described in protein A protocols.

Optimization of LC-MS/MS method for DILLTQ-SPAILSVSPGER.
The most optimal transitions for DILLTQSPAILSVSP-GER were selected on the basis of signal, background noise and chromatographic separation characteristics in the presence of matrix background. It should be noted that for DILLTQSPAILSVSPGER the most prominent charge state ([M+2H] 2+ or [M+3H] 3+ ) could change during method optimization. Especially, when a sensitive method is required, it is important to follow different precursor ion charge states during method optimization, also for other signature peptides. The results obtained with the calibration samples (prepared once, analyzed twice) are shown in Table 3 and the results of the QC samples are shown in Table 4. The obtained calibration curve is presented in Figure 3. Six blank serum samples of individual rats were analyzed, which exhibited a very low background signal in the SRM transition of triply protonated DILLTQ-SPAILSVSPGER at QC LLOQ level. In Figures 4-6 typical chromatograms are shown for blank rat serum, QC LLOQ and QC high samples, respectively. The obtained RSD for the QC low level was higher than 20%, although the RSD was lower than 20% (12%) when the results were calculated with the preset qualifier ion for DILLTQSPAILSVSPGER. Using the preset qualifier ion, however, the RSD for the      LLOQ level was higher than 25% (28%). The presented method would therefore require further optimization prior to full validation according FDA or EMA guidelines.

Conclusion
This work presents an LC-MS/MS method for the quantification of the therapeutic monoclonal antibody Infliximab in rat serum. The approach is based on immunoprecipitation, tryptic digestion of Infliximab, and LC-MS/MS quantification of the signature peptide DILLTQSPAILSVSPGER and its stable isotopically labeled analogue IS. With state of the art UPLC and triple quadrupole instruments, a lower limit of quantification was of 100 ng/ml in rat serum was readily achievable, as illustrated in Figure 5. Since the peptide DILLTQ-SPAILSVSPGER originates from the variable region, it can also be utilized as a signature peptide for the quantification of Infliximab in higher order species, including Homo sapiens.