O labeling is a simple technique for tagging peptides in the presence of H218
O. It typically relies on class-2 proteases (e.g. trypsin) to catalyze the exchange of two 16
atoms for two 18
atoms at the C-terminal carboxyl group of proteolytic peptides, resulting in a mass shift of 4 Da between singly charged, differentially labeled peptide ions observed in MS1
mode (). The first study describing an enzyme-catalyzed oxygen exchange in the presence of H218
O was reported in 1951 by Sprinson and Rittenberg [6
], while MS spectra obtained by Antonov et al.
using electron-beam MS explicitly showed a mass shift resulting from enzyme-catalyzed 18
O incorporation at the carboxylic group of proteolytic peptides [7
]. Desiderio and Kai employed enzyme-catalyzed 18
O exchange for the preparation of internal standards for MS-based quantitation of peptides in biological extracts [8
]. Mirgorodskaya et al.
and Stewart et al.
] proposed the use of 16
O labeling for MS-based quantitation of proteins; the application of this technique as an effective quantitative solution-based, shotgun proteomic tool was first reported by Yao et al.
]. Coupling the SDS–PAGE-based quantitative approach with post-digestion 18
O exchange for differential proteomics of protein complexes was first proposed by Bantscheff et al.
MALDI-MS depicting natural isotopic pattern of selected pair of differentially 16O/18O-labeled peptides, exhibiting complete incorporation of both 18O atoms.
O labeling has also been used for non-quantitative proteomic investigations. Shevchenko et al.
] described a method for de novo
peptide sequencing that employs protein tryptic digestion in the presence of equal ratios of 16
O water for derivatization of tryptic peptides; this method greatly facilitates de novo
sequencing due to simplicity of MS/MS spectra interpretation assisted by the presence of long Y ion series showing characteristic 16
O ratio throughout the spectrum. Kosaka et al.
] employed tryptic digestion in the presence of 50% H218
O for C-terminal characterization of proteins resolved by 2D-PAGE, while Park et al.
] applied this approach to characterize plasma gelsolin as a substrate for matrix metalloproteinase and its potential role in the context of severe trauma. Back et al.
] proposed the use of 18
O labeling for detecting cross-linked peptides within protein complexes. El-Shafey et al.
] further developed this technique and applied it to protein–protein interaction analysis and characterization of the 3D structure of freeze-dried protein complexes. Mirgorodskaya et al.
] proposed an interesting approach for analysis of protein–protein interactions, which employs differential 16
O labeling to distinguish between endogenous protein-complex components and those that were non-specifically co-purified. These non-quantitative studies depict the variety of applications of trypsin-catalyzed 18
O tagging for functional profiling of peptides/proteins mixtures.
With the advent of this technique, it instantly became evident that the enzyme-catalyzed 18O exchange is not always homogeneous (complete) and results in a mixture of peptides having one [16O118O1] or both [18O2] oxygen atoms exchanged at their C-termini. The variable 18O incorporation alters the natural isotopic distribution and forms a complex isotope pattern, depicted in , complicating the calculation of the 18O/16O ratios. Many factors are responsible for the variable degree of 18O incorporation, including variable enzyme substrate specificity, oxygen back-exchange, pH dependency and peptide physical–chemical properties.
Figure 2: MALDI-MS depicting altered isotopic pattern of selected pair of differentially 16O/18O-labeled peptides, indicting the presence of peptides with single 18O atom incorporation [16O118O1] characteristic for variable oxygen incorporation (marked by asterisk). (more ...)
Diverse upstream labeling approaches were developed to optimize oxygen exchange and achieve homogenous (complete oxygen) incorporation. Significant advancement was reported by Yao et al.
], who proposed decoupling of 18
O tagging from the digestion step. This modification allowed targeted optimization of conditions for incorporating 18
O and minimized H218
O consumption. This study also confirmed that trypsin-facilitated 18
O exchange of both C-terminal 16
O atoms is a two-step catalytic reaction; the first hydrolytic reaction, RC16
ONHR’ + H218
O → RC16
NR′, is followed by the second hydrolytic reaction, RC16
O → RC18
O. Both trypsin-catalyzed oxygen exchanges were confirmed to be strictly substrate (Lys and Arg)-specific. This investigation showed weaker substrate binding for Lys-ending peptides than for Arg-ending ones. Subsequently, Hajkova et al.
] showed that the incorporation of the second 18
O atom can be substantially accelerated if the post-digestion 18
O labeling is carried out at a pH in the range of 5–6, depending on the enzyme used in this step. Storms et al.
] observed that prohibition of 18
O back-exchange can be efficiently accomplished by heating differentially labeled samples at 80°C for 10 min before combining them for subsequent MS analysis. Sevinsky et al.
] proposed the use of immobilized trypsin for both the proteolysis and the labeling step to provide protection for the isotopic tags throughout the IPG–IEF process and prevent the 18
O back-exchange. A significant increase of the 18
O labeling rate was reported by Mirza et al.
], describing accelerated oxygen-exchange if the trypsin was immobilized in the micro-spin column. Wang et al.
] proposed inverse 18
O labeling for improved peptide/protein quantitation accuracy, particularly for peptides/proteins exhibiting extreme abundance changes.
For the past several years, our laboratory has been investigating the utility of 16
O for proteomic profiling of a complex membrane protein mixture that relies on buffered methanol to facilitate solubilization and proteolysis of membrane proteins. We have shown, using an α-N-benzoyl-l
-arginine ethyl ester (BAEE) assay, that trypsin exhibits higher activity in 20% MeOH than in pure aqueous buffer, resulting in improved labeling efficiency when used for post-digestion labeling of membrane proteins [24
]. The workflow depicting this modification is shown in .
Figure 3: A workflow depicting differential 16O/18O labeling of membrane proteins. Isolated membrane samples (control and modified one) are first solubilized and digested in the buffer containing 60% MeOH/H216O. After lyophilization, compared sample is digested (more ...)
In addition to efforts focused on optimizing the labeling conditions, several advanced computational tools were developed with the aim of accounting for the variable oxygen incorporation. Halligan et al.
] developed an algorithm that employs a calculation method previously described by Yao et al.
]. The algorithm relies on differences between experimentally obtained isotope abundances and those obtained theoretically, while the method developed by Johnson and Muddiman [26
] relies on average-based calculations to account for variable oxygen incorporation. Eckel-Passow et al.
] described a method for estimating the 18
O incorporation directly, relying on a multivariable regression model in the context of post-digestion 18
O exchange. Ramos-Fernandez et al.
] describe a kinetic exchange model that is incorporated within the quantification algorithm and is able to eliminate artifacts caused by variable oxygen incorporation; this model is readily amenable to quantitative profiling of complex protein mixtures. The algorithm developed by Mason et al.
] utilizes a linear regression model to automatically interpret the spectra of 18
O-labeled isotope clusters, correcting for artifacts caused by variable 18
O incorporation. This approach uses centroid peak data obtained by MS with high-resolution power. We are in the process of testing software developed in-house that accounts for variable 18
O incorporation. The assumption that the integrated area of each peak within the isotopic manifold represents overlapping Poisson distributions is used as a basis for accurate 18
O peptide ratio calculation.