1 Introduction

Mass spectrometry (MS) has already become an indispensable tool in quantitative proteomics and is beginning to enjoy popularity in other areas where reliable protein quantitation is critical, such as pharmacokinetic studies of protein drugs [1, 2]. To achieve accurate quantitation, most of the MS-based methods apply differential stable isotope labeling to introduce the internal standard with a specific isotope label. Although the isotope label can be introduced with relative ease in prokaryotic systems metabolically by using isotope-enriched (e.g., N15 or C13) cell culture medium, production of labeled proteins in eukaryotic expression systems is significantly more costly and labor-intensive [3, 4]. Alternative methods of placing isotopic label on internal standards include production of surrogate standards [5], or chemical derivatization of specific amino acid residues using isotope labeling reagents (such as ICAT [6], iTRAQ [7]). Enzyme-catalyzed O18 labeling [8] (as well as its modifications that use acid catalysis [9–11]) is another elegant way to introduce isotope labeling. All those approaches have gained wide acceptance in the field of quantitative proteomics, and also provide powerful tools for targeted protein quantitation, especially in the pharmacokinetic studies of protein drugs. Currently, the most commonly applied method for targeted protein quantitation is known as absolute quantification of proteins (AQUA) [12], in which synthesized stable isotope-labeled peptides are used as internal standards. However, an intrinsic disadvantage of using synthesized peptide standards for protein quantitation is the combination of the standards and endogenous peptides at the very end of the process, which inevitably introduces more sources of quantitation error [13]. Although the use of isotope-labeled protein standards [14] can eliminate this problem, its application is limited due to high costs and difficulties of producing eukaryotic protein standards.

We propose a new O18 labeling strategy where the labeling is carried out at the intact protein level by using O18-labeled iodoacetic acid (IAA) to cap the cysteine residues. Since the labeling is done at the intact protein level, the sample and the internal standard can be combined at the very beginning of the process, which eliminates possible discriminatory effects and quantitation errors caused by differential processing of the protein within the sample and the internal standard prior to MS analysis. Feasibility of the new method is demonstrated using as a model an 80 kDa protein non-glycosylated human serum transferrin (hTf), which is a part of a number of biopharmaceutical products that are currently under development [15].

2 Experimental

Non-glycosylated human serum transferrin (hTf) was provided by Professor Anne B. Mason (University of Vermont College of Medicine, Burlington, VT, USA); H2O18 (97 % purity), iodoacetic acid (IAA), and proteomic-grade trypsin were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO, USA). All other chemicals and solvents used in this work were of analytical grade or higher.

The O18-labeled IAA solution was prepared by dissolving 80.0 mg of IAA in 495 μL of O18 enriched water and incubating in the presence of 1 % trifluoroacetic acid at 50 °C for 1 day. Progress of O18 incorporation into IAA was monitored by an API-2000 triple quadrupole mass spectrometer (AB SCIEX, Toronto, Canada) in negative ion mode. Following pH adjustment to near-neutral level, the O18-labeled IAA solution was sealed to prevent the oxidation from air and covered by aluminum foil to avoid light, and kept at −20 °C for future use.

The O18 labeling of hTf was introduced by alkylation of free cysteines. Briefly, 20 μg of hTf was prepared in 200 μL of 100 μM ammonium bicarbonate (pH 8.0), and then reduced with 10 mM of DTT at 50 °C for 45 min. The alkylation of cysteine residues was followed by addition of O18-labeled IAA stock solution to a concentration of 30 mM, and incubation at 50 °C for 45 min in the dark. The digestion was initiated by addition of trypsin at a 20:1 substrate to enzyme ratio and followed by incubation at 37 °C for overnight.

To investigate the back exchange of O18-labeled peptides under typical RP-HPLC conditions and near neutral pH, a 20 μL aliquot of the tryptic digests was diluted into 100 μL of H2O16 with 0.1 % formic acid or 100 μL of 100 mM ammonium bicarbonate, respectively, and sampled hourly or daily for LC/MS/MS analysis.

To use this strategy for protein quantitation, two batches of hTf were alkylated by O18-labeled and unlabeled IAA, respectively. Right after the alkylation, two samples were mixed in ratios of 1:50, 1:20, 1:10, 1:5, 1:2, 1:1, 2:1, 5:1, 10:1, 20:1, 50:1 and digested by trypsin using the procedure outlined above.

The digested sample was analyzed with LC/MS/MS using an LC Packings Ultimate (Dionex/Thermo Fisher Scientific, Sunnyvale, CA, USA) nano-HPLC system coupled with a QStar-XL (AB SCIEX, Toronto, Canada) hybrid quadrupole/TOF MS. A previously reported setup and method were used for the nano-LC/MS/MS analysis [9].

3 Results and Discussion

The O18-labeled IAA acid was prepared by exchanging oxygen atoms in its carboxylic group with H2O18 at low pH. The labeling results are shown in Figure 1, where the isotopic distribution of IAA after 1 day’s incubation at 50 °C, in the presence of 1 % TFA, shows >85 % exchange of both carboxylic oxygen atoms and >98 % exchange of at least one oxygen atom. These numbers yield the overall extent of O18 incorporation being 93 %, whereas the theoretical value is calculated to be 93.9 % after taking into account the presence of residual O16 in the reaction mixture. The consistent results suggested that the O18 exchange between IAA and water had already reached the equilibrium. The presence of iodide ions were also observed in the mass spectra, which was quite common even for freshly prepared IAA solution.

Figure 1
figure 1

ESI MS O18-labeled IAA in the negative ion mode showing the isotopic distribution of IAA anion

Full size image

The alkylation of cysteine residues with the O18-labeled IAA was further investigated using hTf as the model protein, which has 38 cysteine residues. Figure 2 presents a nano-LC/MS/MS analysis of the mixture of tryptic digests from two batches of hTf capped by unlabeled and O18-labeled IAA, respectively. The mass spectrum (Figure 2b) of a tryptic peptide KPVEEYANC*HLAR indicates a maximum mass shift of 4 Da between the unlabeled (black trace) and the labeled (grey trace) peptides. Extracted ion chromatograms plotted for their most abundant isotopic peaks, I0 and I4 (Figure 2a), clearly show identical elution time for the O18-labeled peptide and its unlabeled counterpart, which is critically important for reliable quantitation. The tandem MS data (Figure 2c and d) suggest that the presence of IAA moiety on a peptide ion does not interfere with its identification using CID (contrary to ICAT labeling [16]). Furthermore, the isotopic distributions of y +4 and y +5 ions provide clear indication as to the localization of O18 atoms within the peptide (cysteine residue).

Figure 2
figure 2

Nano-LC/MS/MS analysis of tryptic peptides derived from a mixture of hTf capped with unlabeled and O18-labeled IAA. (a) Total ion chromatogram of the digest and extracted ion chromatograms of peak I0 and I4 representing unlabeled (black trace) and O18-labeled (gray trace) tryptic fragment KPVEEYANC*HLAR. (b) Isotopic distribution of a triply charged peptide ion KPVEEYANC*HLAR. (c) MS/MS (CID) of peptide ion KPVEEYANC*HLAR shown in (b). (d) A zoom view of the MS/MS spectrum (c) showing y +4 and y +5 fragment ions

Full size image

The O18 labeling of other cysteine-containing tryptic peptides is summarized in Table 1. For peptides containing one cysteine, the full labeling products (with both two O18 atoms incorporated) made up to 87 %–90 %, consistent with the labeling results of IAA (vide supra). Unlike enzyme-catalyzed O18 labeling which has peptide-specific O18 incorporation rate [17], the use of O18-labeled IAA generates almost the same labeling efficiency for all the cysteine-containing peptides. This is particularly helpful in quantitative measurements when the mass shift is not sufficient to completely separate the isotopic clusters of the labeled and unlabeled peptides [18]. The ratio between unlabeled and labeled peptides can be easily estimated by the following equation:

Table 1 Efficiency of O18 incorporation into cysteine-containing tryptic fragments of human serum transferrin (hTf)
Full size table
$$ {\text{peptide ratio}} = {\text{I1I5}} \cdot {1} + \alpha $$

where α (α = 0.14 in this work) is a fraction of partially labeled peptides (only one O18 atom incorporated); and I 1 and I 5 represent the observed relative intensities of the second isotopic peaks for the unlabeled peptides and fully O18-labeled peptides, respectively (see Figure 2b).

Since the O18 labeling is introduced at the protein level, any consecutive steps, including purification, digestion, and LC/MS analysis will become sources of back exchange by exposing the O18 labeled carboxylic acid group in O16 water. Thus, the stability of the O18 labeling was investigated at the RP-HPLC conditions (0.1 % FA) as well as at pH 8 for tryptic digestion (see Supplementary Material). The relative abundance of the isotopic peak I 2 (as percentage of the total ionic signal) was calculated and monitored over time. No noticeable back exchange was observed in RP-HPLC conditions for at least 8 h, or under near neutral pH for at least 6 d. The good stability of O18 labeling under near neutral pH allows overnight tryptic digestion as well as any desired purification process as long as it does not require a long period of time in extreme pH conditions. Furthermore, the good stability of O18 labeling in RP-HPLC conditions allows the feasibility of applying long LC gradient separation for complex samples.

To demonstrate the feasibility of using O18-labeled IAA for protein quantitation, we generated hTf capped by unlabeled and O18-labeled IAA at various ratios covering three orders of magnitude, and performed nano-LC/MS analysis. The quantitation was achieved using two cysteine-containing tryptic peptides from hTf respectively (see Supplementary Material). High correlation of the theoretical and experimentally observed protein ratios was achieved (R2 > 0.99). Two reference peptides both showed good linearity in up to three orders of magnitude. Furthermore, the quantitation results generated by these two peptides were highly consistent, highlighting the reliability of the quantitation.

In contrast to enzyme-catalyzed O18 labeling, which universally incorporates O18 into carboxylic terminus of each peptide, the O18-labeled IAA only targets the cysteine-containing tryptic peptides and, therefore, cannot be applied to quantitate proteins containing no cysteine residues. Nevertheless, even the most conservative estimates of the fraction of completely cysteine-free proteins in the proteomes of higher organisms put this number as low as 8 % [19], while the SWISS-Prot database puts the number of human proteins that have at least one cysteine residue in the sequence as high as 97 %. Obviously, the frequency of occurrence of cysteine residues in prokaryotic organisms is significantly lower, but quantitation of prokaryotic proteins can be easily achieved by generating isotopically labeled internal standards at the protein expression step [4].

4 Conclusion

O18-labeled IAA was prepared in O18-enriched water at low pH, and then used for MS-based protein quantitation. This approach offers several advantages over enzyme catalyzed O18 labeling. First, full labeling can be easily achieved at the equilibrium of the oxygen exchange reaction. Furthermore, the O18 incorporation efficiency for peptides is consistent rather than peptide-specific, which makes the ‘unlabeled/labeled’ calculation much easier. Finally, the combination of unlabeled and labeled samples at protein level excludes all sources of quantitation error introduced afterwards. The feasibility of this method was demonstrated by using hTf as the model protein; linear quantitation with dynamic range spanning three orders of magnitude was achieved. Due to the thiol-specific feature, its utility is limited to cysteine-containing proteins. However, since only a very small percentage of eukaryotic proteins contain no cysteine residues, this approach can still be generally applied to most of the protein quantitation works. Furthermore, virtually all of the existing protein therapeutics contain cysteine residues, making this new technique particularly suitable for tasks ranging from pharmacokinetic studies to quality control of biopharmaceutical products, a vast emerging field in protein quantitation [1].