The therapeutic use of proteins, especially monoclonal antibodies (mAbs), has increased remarkably over the past decade. At the end of 2017, 57 mAbs and 11 biosimilars in clinical use based on information from the US Food and Drug Administration (FDA) or European Medicines Agency (EMA) [1]. The rapid growth of therapeutic proteins in drug development brings about significantly increased demand for the quantitative bioanalysis of protein therapeutics. There are also increasing needs in monitoring endogenous proteins, e.g., protein biomarkers, the target proteins of drugs, enzymes, and transporters. Traditionally, ligand binding assays or immunoassays are commonly used for the quantitative analysis of proteins in biological matrices. Ligand binding assays are very sensitive, often able to analyze proteins as low as pg/ml level. However, ligand binding assays require the development of suitable capture and detection reagents, which takes time and resources, and may not be affordable in drug discovery and early development. In addition, ligand binding assays are less specific and may be affected by the presence of anti-drug antibodies in the samples [2]. Mass spectrometry (MS)-based assays have unique advantages: e.g., high specificity, wide dynamic range, fast method development, and ability to quantify multiple proteins simultaneously. As a result, MS-based assays have been gaining increasing attention and interest for the quantitative bioanalysis of proteins in recent years [3, 4]. Publication guidelines have been devised for targeted mass spectrometry measurements of peptides and proteins [5]. Table 1 summarizes the advantages and limitations of MS-based assays and ligand binding assays for the quantitation of proteins.

There are two major MS-based strategies, the surrogate peptide approach and the intact protein approach (see Figure 1 for the scheme) for the quantitative bioanalysis of protein. The surrogate peptide approach (Figure 1A) is an indirect approach, in which the target protein is digested to smaller peptides first and then analyzing the generated peptides as surrogate analytes of the target protein. The intact protein approach (Figure 1B) is a ‘true’ measurement of the target protein, directly analyzes the intact protein by mass spectrometry. In this review, important considerations in choosing an appropriate strategy for protein bioanalysis and in developing and optimizing the LC-MS protein bioanalytical assays, including the selection of internal standards, selection of surrogate peptide, protein digestion, sample preparation, and MS detection techniques, will be discussed in detail.

Figure 1. Scheme of the surrogate peptide approach (A) and the intact protein approach (B) for the bioanalysis of proteins.

The surrogate peptide approach was first introduced by Barr et al. [6] in 1996, and was improved by Barnidge, Gerber, and Kuhn et al. [7-10]. It has become the most commonly used strategy for MS-based protein quantitation due to its high sensitivity and specificity, and the wide availability of triple quadrupole mass spectrometers. In this approach, the target protein is digested to smaller peptides first. One (or more) unique peptide(s) with good selectivity and sensitivity are identified from the digests, and used as the surrogate analyte of the protein. The surrogate peptide of the protein is then analyzed by LC-MS. Figure 2 shows the flowchart for developing and optimizing an LC-MS protein quantitation assay using the surrogate peptide approach.

Figure 2. Flowchart of the surrogate peptide approach for the LC–MS bioanalysis of proteins.

The surrogate peptide approach relies on the analysis of the surrogate peptide for the quantitation of the whole target protein. Therefore, the selection of appropriate surrogate peptide is critical to ensure the sensitivity, specificity and robustness of the assay. Usually, ‘signature’ peptides, which are peptides unique for the specific target protein, are chosen as the surrogate peptides. Selection of the appropriate surrogate peptides should [11] : (1) avoid peptides prone to chemical modification (e.g., peptides containing methionine, cysteine or tryptophan); (2) avoid arginine-arginine and lysine-lysine in the peptide sequence to minimize inconsistent tryptic digestion; and (3) select peptides with appropriate length (~ 8-20 amino acids): being too short may cause the lack of selectivity, and too long may affect the sensitivity and significantly increase the difficulty and cost of synthesizing the stable-isotope-labeled peptide (as an internal standard). A typical procedure is to perform an in silico digestion of the target protein to generate a list of potential surrogate peptides. These peptides are then searched against all existing proteins in the biological matrices using online databases (e.g., Standard Protein BLAST, http://www.ncbi.nlm.nih.gov/BLAST/Blast.cgi?PAGE=Proteins) to confirm that the signature peptides only exist in the target protein. The sensitivity, specificity and chromatographic behavior of these signature peptides are then evaluated using actual digested protein samples in biological matrices, and the best one will be chosen as the surrogate peptide for the target protein.

Any assay developed using a signature surrogate peptide can only be applied to one specific target protein, and a new assay needs to be developed for each new protein. Recently, Furlong et al. [12, 13] proposed a universal surrogate peptide strategy to enable the quantitative analysis of different mAbs using a single assay. In this approach, they used peptide VVSVLTVLHQDWLNGK, a peptide contained in all human immunoglobulin G (IgG) 1 and 4, as the surrogate peptide. An LC-MS assay based upon this surrogate peptide is able to quantify various human IgG1 and IgG4 based protein therapeutics in animal species. This strategy can greatly reduce the efforts and needs to develop individual assays for new drug candidates in support of preclinical animal studies.

The stable-isotope-labeled (SIL) analog of the target protein is the preferred internal standard (IS) for LC-MS bioanalysis of proteins [14, 15], since a SIL-protein IS can compensate for any variations in all steps including pre-digestion, digestion, post-digestion sample preparation, and LC-MS analysis. However, a SIL-protein IS may be difficult to obtain, especially during drug discovery or early development. Alternatively, SIL-peptide IS, the SIL surrogate peptide of the target protein, has been commonly used for protein bioanalysis [16-20]. SIL-peptide IS can well compensate for the variations in post-digestion sample preparation and LC–MS analysis, but not for pre-digestion and digestion sample preparation (e.g., immunocapture enrichment of target protein). Therefore, consistent pre-digestion sample preparation and digestion are critical to ensure the accuracy and precision of the assay when using SIL-peptide IS. Another type of SIL-peptide IS is the cleavable SIL-peptide (or flanking SIL-peptide), which contains a few extra amino acids beyond the cleavage sites at both the N- and C-terminus of the SIL-surrogate peptide [21-23]. Cleavable SIL-peptide IS can be digested along with the target protein to release the SIL-surrogate peptide, and thus it can normalize some of the variations in the digestion process. However, using cleavable SIL-peptide IS may not be applicable to all proteins and may result in considerable bias [24]. In direct comparisons between cleavable SIL-peptide IS and SIL-peptide IS, using the cleavable SIL-peptide IS did not show significant improvement [25, 26]. Structural analogous proteins or peptides were also used as internal standards [11, 27-29]. However, their correction for the LC-MS bioanalysis of proteins is limited and may result in significantly reduced assay performance [26]. The internal standards can be concatenated together, called Quantification Concatamers (QconCATs), to quantify several proteins or protein isoforms [30], although the design of flanking regions for each peptide standard should be considered carefully [31].

Proteins can be digested using an enzyme (e.g., trypsin, Lys-C, Glu-C, Asp-N [32] ) or chemically [33]. Trypsin digestion is the most commonly used approach because of its high efficiency, good specificity and low cost. Efficient and reproducible digestion is critical to ensure the accuracy, precision and reproducibility of the assay [34], especially when no SIL-protein IS is available. Digestion conditions (incubation time, temperature, trypsin to protein ratio, etc.) need to be optimized to ensure sufficient and consistent digestion. Conventional protein digestion applies a sequential denaturation, reduction, and alkylation pretreatment before the actual enzymatic digestion. This pretreatment can improve the digestion efficiency and completeness by unfolding the proteins and increasing the accessibility of the digestion enzyme. However, pretreatment is time-consuming and uses reagents (urea, guanidine, dithiothreitol, and iodoacetamide, etc.) undesirable for later digestion or LC-MS analysis. Various approaches have been utilized to accelerate the tryptic digestion process, such as elevated digestion temperature, the addition of organic solvent, and microwave-assisted digestion [33]. Recently, Ouyang et al. [35] developed the ‘pellet digestion’ methodology for simple and fast digestion of large proteins. This approach eliminated the time-consuming pretreatment and had been successfully applied to the LC-MS/MS bioanalysis of large proteins in animal pharmacokinetic studies [35, 36]. Yuan et al. [37] evaluated the effectiveness of pellet digestion for the digestion of a large protein with multiple disulfide bonds. They found that pellet digestion provided much better digestion efficiency compared to direct digestion without any pretreatment. Pellet digestion also provided better or similar digestion efficiency compared to the traditional digestion-with-pretreatment method for different regions of the test protein, including the hard-to-digest regions under direct digestion. The main purpose of the denaturation, reduction, and alkylation pretreatment is to obtain a more complete sequence coverage of the protein, which is critical in qualitative work. However, for quantitative bioanalysis, only one surrogate peptide is required. In most cases, pellet digestion or other no-pretreatment digestion methods are good enough to generate the satisfactory surrogate peptide and achieve consistent digestion. Tandem mass tags (TMT), for example, TMT-11plex [38] or TMT10plex [39, 40] reagents from Thermo Scientific, can also label the peptides and yield global proteomic profiles.

Figure 3. Suggested decision tree for selecting the strategy (surrogate peptide approach or intact protein approach) for the LC-MS quantitative bioanalysis of proteins.

Another major strategy is through the direct mass spectrometric detection of the intact protein. Unlike the surrogate peptide approach, in which only a small piece of the target protein is detected, this approach is a ‘true’ measurement of the whole protein. The mass spectrometric detection of the intact protein can be either achieved by selected reaction monitoring (SRM) using a triple quadrupole mass spectrometer (e.g., the bioanalysis of rK5, a small protein with molecular weight of about 10k Dalton [41] ), also called Multiple reaction monitoring (MRM), or by high-resolution accurate mass spectrometry (HRMS), as demonstrated by the analysis of lysozyme [42] or snake venoms [43] using Orbitrap mass spectrometers. Compared to SRM, HRMS could provide a much better selectivity resulting from its high resolution detection. In addition, some proteins may not have a major production for the SRM detection to obtain satisfactory assay sensitivity and selectivity. The intact protein approach using HRMS also has the potential advantage to simultaneously quantify proteins and their derivatives after biotransformation and post-translational modifications [42], all of which may contain the same surrogate peptide and may not be differentiated using the surrogate peptide approach. One example is the analysis of different isoforms of glycosylated proteins (e.g., apolipoprotein C3) by HRMS detection of the intact proteins [44]. Currently, this approach is limited to the analysis of small proteins (molecular weight usually less than 20,000 Da), and is impractical for larger proteins (e.g., mAbs), due to the difficulties in sample preparation and LC-MS analysis [41]. The continuous improvement in mass spectrometry, chromatographic separation and sample preparation techniques will enable the wider application of this approach for bioanalysis of larger proteins in the future. For an example, Choi SB integrated reversed-phase fractionation into capillary electrophoresis nanoelectrospray ionization HRMS to detect proteins from single neuron [45]. Figure 3 shows the suggested decision tree for selecting the appropriate strategy (surrogate peptide approach or intact protein approach) for the LC-MS quantitative bioanalysis of proteins.

Improving sensitivity is one major challenge for the wider application of MS-based assays for protein bioanalysis. The existence of highly abundant endogenous proteins in biological matrices may cause high background noise and severe ion suppression during MS detection, which will significantly reduce the sensitivity of the assay. A sample preparation method that can remove the abundant background proteins/peptides and selectively extract the target protein or surrogate peptide is critical to improve the assay sensitivity and selectivity. A variety of sample preparation strategies have been evaluated and applied for the purification of the target protein (before digestion) or the surrogate peptide (after digestion), and will be discussed here.

Most of the proteins in serum or plasma samples can be precipitated with water-miscible organic solvents (e.g., aceto¬nitrile and methanol), and collected by centrifugation. Protein precipi¬tation can be used for the extraction of organic soluble proteins (e.g., PEGylated proteins [11] ), since they are retained in the supernatant and can be separated from other precipitated proteins. Large proteins, such as mAbs, will co-precipitate with endogenous proteins in serum/plasma, making the supernatant inappropriate for the analysis. Alternatively, the precipitated protein pellet can be used for the analysis of the large target proteins, as in pellet digestion [35, 37]. One advantage of this approach is that protein precipitation can provide some degree of sample cleanup by removing soluble proteins, salts, and a significant portion of phospholipids [35, 37]. Recently, Liu et al. [46] developed an acid-assisted protein precipitation method that can efficiently remove albumin, the most abundant endogenous protein in serum/plasma samples, while retaining the target proteins, thereby obtaining cleaner samples and achieving improved sensitivity.

Solid-phase extraction (SPE) has been applied for the extraction of peptides from digested samples [18, 27]. The use of strong-cation exchange (SCX) SPE as an orthogonal sample preparation technique to reversed-phase (RP) chromatographic separation was demonstrated to efficiently remove background peptides, and therefore improve the assay sensitivity for the analysis of a test mAb in serum [18]. 2D SPE (RP/SCX SPE) can further improve the purification of peptides from tryptic digests of plasma samples: RP SPE removes salts and highly hydrophobic components, and SCX SPE eliminates background peptides with significantly different basicity [27]. For small proteins such as rK5 [41] and lysozyme [42], SPE was also demonstrated to be an effective approach to extract the intact target protein from plasma samples.

Derivatization of peptides is a widely used technique in proteomics for qualitative work and relative protein quantification. Recently, Yuan et al. proposed a novel selective peptide derivatization (SPD) strategy to improve assay sensitivity for the LC-MS/MS quantitative bioanalysis of proteins [47]. The SPD strategy works by selectively derivatizing the surrogate peptide of the target protein while not derivatizing background peptides. Selective derivatization can increase physicochemical property differences between derivatized target peptides and underivatized peptides, therefore improving their separation during extraction and chromatography, and enhancing the assay sensitivity. By applying malondialdehyde (MDA) to selectively derivatize the arginine residue in the surrogate peptide SLIY and SPE cleanup, they achieved a 5-fold increase in sensitivity for the analysis of a test mAb in monkey serum. SPD can provide an alternative sample preparation approach for developing sensitive LC-MS/MS assays for proteins, especially when a suitable immunocapture antibody is not available.

Immunoaffinity capture of target proteins or peptides, coupled with LC-MS detection, is the current method of choice to achieve the most sensitive LC-MS assay, often reaching a lower limit of quantitation (LLOQ) at low ng/ml level. Using anti-idiotypic antibodies or the receptor/ligands of the target proteins to capture the target protein is a highly specific immunocapture approach. This approach provides highly efficient purification and enrichment for the target protein, and therefore significantly increases the LC-MS assay sensitivity. For example, Dubois et al. [28] used soluble epidermal growth factor receptor, the pharmacological target of Erbitux (a therapeutic mAb), to specifically purify Erbitux from serum samples and obtained an LLOQ similar to that of ELISA methods at 20 ng/ml. Highly specific immunocapture enrichment can also be achieved by using an anti-peptide antibody to capture the surrogate peptide in digested samples, as in the stable-isotope standards with capture by anti-peptide antibodies (SISCAPA) technique [16]. This approach can provide >100 fold enrichment of target peptides [16] and increase sensitivity by 100–1000 fold [19]. The samples are digested before immunocapture enrichment, which minimizes the potential interference from the anti-drug antibody (ADA). In addition, multiple anti-peptide antibodies targeting different peptides can be used to achieve multiplexed enrichment of peptides and the simultaneous analysis of multiple proteins [19, 20]. To achieve even better sensitivity, sequential immunocapture of target protein and its surrogate peptide can be applied [22, 48]. This approach generated highly purified and enriched samples, which significantly improved the sensitivity (e.g., low pg/ml level) for the analysis of proteins in serum and tissues [48].

The aforementioned immunocapture approaches require specifically developed antibodies as the capture reagents, making the developed immunocapture method only applicable to the specific target protein. In addition, the capture reagents may not be readily available, especially in drug discovery and early drug development. One more generic strategy is to use capture reagents that capture a common sequence of amino acids or region of protein in different proteins for sample purification. For example, protein A or protein G can bind to the Fc region of immunoglobulin G (IgG) with high affinity, and therefore, can be used for the immunocapture purification of antibody drugs containing the IgG Fc region [21, 49]. Similarly, an anti-human IgG Fc antibody can specifically bind to the Fc region of human IgGs, whereas not bind to IgGs from other species. As a result, various human mAbs can be extracted using this generic immunocapture method; especially from animal matrices. Li et al. [29] developed a generic method using anti-human IgG Fc antibody to purify eight different mAbs from animal plasma or serum samples. They were also able to simultaneously extract and analyze four different mAbs in rat plasma using this approach [50]. Bogdan et al. [51] developed a generic immunocapture procedure for the quantification of human mAbs in mouse tissue samples tissues using a commercially available anti-human IgG Fc antibody. The generic approach can also be applicable to the extraction of PEGylated proteins/peptides. Xu et al. [52] utilized an anti-PEG antibody, which can specifically capture the PEG portion of the PEGylated protein, for the immunocapture enrichment of a PEGylated therapeutic peptide.

SRM of the surrogate peptide(s) using a triple quadrupole mass spectrometer is currently the most commonly used technique for the quantitation of proteins, since SRM, in combination with LC separation, can offer both good specificity and high sensitivity. In optimizing the SRM of peptides, one key is that unlike small molecule analytes, peptides usually do not form a predominant singly charged ion during ionization, but form ions at multiple charge states. The fragmentation pattern of the peptide ions under different charge states also may vary significantly [53]. Therefore, to improve the assay sensitivity and selectivity, SRM transitions of peptide precursor ions at various charge states need to be compared, preferably in the presence of biological matrices to mitigate potential matrix effects on the ionization. Another detection method, called parallel reaction monitoring (PRM), measures all transitions for a peptide with a high resolution mass spectrometer.

HRMS has drawn growing interest as an alternative detection technique to SRM for the quantitation of proteins in recent years [42, 54-57]. The sensitivity gap between HRMS and SRM is decreasing with the continuous advancement of HRMS instrumentation. HRMS has been shown to provide better assay sensitivity than SRM (triple quadrupole instruments) in some cases, e.g., for peptides with poor fragmentation [55]. For example, Ciccimaro et al. [58] developed survivor-selected ion monitoring (survivor-SIM) approach to selectively filter background ions and reduce chemical noise. They achieved significantly improved sensitivity and specificity compared to SRM for the detection of disulfide-rich cyclic peptides extracted from plasma. In addition to its superior specificity and good sensitivity, HRMS can provide qualitative information in the same analysis (e.g., the biotransformation products of the target protein [42] ). HRMS could also be used in the intact protein approach for protein quantitation, while the application of SRM is often limited by its relatively low selectivity. Orbitrap [42, 59] and time-of-flight mass spectrometers [55-57] are the two most commonly used HRMS instruments for protein bioanalysis. Table 2 summarizes the features of SRM and HRMS for the bioanalysis of proteins.

Non-targeted and targeted quantification are two major HRMS analysis approaches. The non-targeted approach generically acquires all ions over a mass range, requiring no (or minimum) MS optimization for individual analytes. This approach can provide qualitative information in addition to the quantitative one, but may not provide the best sensitivity [56]. The targeted quantification approach acquires productions of specific precursor ions. This approach will miss the qualitative information, but can significantly improve the sensitivity of HRMS, and may even obtain similar or better sensitivity compared to SRM [55-57].

MS-based assays have rapidly gained momentum and more acceptances for protein quantitation in recent years due to unique advantages. LC-MS has been more extensively applied for the quantitation of various proteins, including protein therapeutics (monoclonal antibodies, PEGylated proteins, etc.) and protein biomarkers. Improving sensitivity is still one major challenge for the application of MS-based assays for protein bioanalysis. Various sample preparation and LC-MS technologies are evaluated and developed to improve the sensitivity and selectivity of MS-based assays, therefore, enable their wider application. The combination of immuno-capture sample preparation with LC-MS is one promising direction, since it combines the advantages from both LBA and LC-MS assays, and can provide both significantly improved sensitivity and high specificity. The availability of data from ProteomeXchange Consortium can also help the progress of mass spectrometry application of biological analysis [60].

The authors thank Drs. Mark E. Arnold and Qin C. Ji for their helpful review of the manuscript.