Regression analysis for comparing protein samples with 16O/18O stable-isotope labeled mass spectrometry

MOTIVATION: Using stable isotopes in global proteome scans, labeled molecules from one sample are pooled with unlabeled molecules from another sample and subsequently subjected to mass-spectral analysis. Stable-isotope methodologies make use of the fact that identical molecules of different stable-isotope compositions are differentiated in a mass spectrometer and are represented in a mass spectrum as distinct isotopic clusters with a known mass shift. We describe two multivariable linear regression models for (16)O/(18)O stable-isotope labeled data that jointly model pairs of resolved isotopic clusters from the same peptide and quantify the abundance present in each of the two biological samples while concurrently accounting for peptide-specific incorporation rates of the heavy isotope. The abundance measure for each peptide from the two biological samples is then used in down-stream statistical analyses, e.g. differential expression analysis. Because the multivariable regression models are able to correct for the abundance of the labeled peptide that appear as an unlabeled peptide due to the inability to exchange the natural C-terminal oxygen for the heavy isotope, they are particularly advantageous for a two-step digestion/labeling procedure. We discuss how estimates from the regression model are used to quantify the variability of the estimated abundance measures for the paired samples. Although discussed in the context of (16)O/(18)O stable-isotope labeled data, the multivariable regression models are generalizable to other stable-isotope labeled technologies.     

An isotope coding strategy for proteomics involving both amine and carboxyl group labeling

This paper describes a heavy isotope coding strategy for the analysis of all types of tryptic peptides, including those that are N-terminally blocked and from the C-terminus of proteins. The method exploits differential derivatization of amine and carboxyl groups generated during proteolysis as a means of coding. Carboxyl groups produced during proteolysis incorporate 18O from H218O. Peptides from the C-terminus of proteins were not labeled with 18O unless they contained a basic C-terminal amino acid. Primary amines from control and experimental samples were differentially acylated after proteolysis with either 1H3- or 2H3-N-acetoxysuccinamide. When these two types of labeling were combined, unique coding patterns were achieved for peptides arising from the C-termini and blocked N-termini of proteins. This method was used to (1) distinguish C-terminal peptides in model proteins, (2) recognize N-terminal peptides from proteins in which the amino terminus is acylated, and (3) identify primary structure variations between proteins from different sources.     

Properties of 13C-substituted arginine in stable isotope labeling by amino acids in cell culture (SILAC)

We have recently described a method, stable isotope labeling by amino acids in cell culture (SILAC) for the accurate quantitation of relative protein abundances. Cells were metabolically labeled with deuterated leucine, leading to complete incorporation within about five cell doublings. Here, we investigate fully substituted 13C-labeled arginine in the SILAC method. After tryptic digestion, there is a single label at the C-terminal position in half of the peptides. Labeled and unlabeled peptides coelute in liquid chromatography-mass spectrometric analysis, eliminating quantitation error due to unequal sampling of ion profiles. Tandem mass spectrum interpretation and database identification are aided by the predictable shift of the y-ions in the labeled form. The quantitation of mixtures of total cell lysates in known ratios resolved on a one-dimensional SDS-PAGE gel produced consistent and reproducible results with relative standard deviations better than five percent under optimal conditions.     

Trypsin is the primary mechanism by which the (18)O isotopic label is lost in quantitative proteomic studies

Labeling with (18)O is currently one of the most commonly used methods for incorporating a stable isotopic label into samples for comparative proteomic studies. In this approach, isotopic labeling involves the enzymatic digestion, typically performed with trypsin, of a protein population in (18)O-water, which incorporates the stable isotope into the C termini of the newly formed peptides. Although trypsin is often used to facilitate isotopic incorporation after digestion, it is typically overlooked that this same mechanism can lead to isotopic loss even under conditions such as low pH where it is assumed that trypsin is inactive. To examine the role that trypsin plays in isotopic loss, several experiments were performed on the rate of delabeling under conditions relevant to multidimensional proteomic experiments. Results from these studies demonstrate that enzyme-facilitated exchange of (18)O in the peptide with (16)O in the aqueous solvent was the major process by which the label is removed from the peptides, even under conditions of low pH and temperature where trypsin is thought to be inactive. This study brings the rapid, tryptic-facilitated exchange to the attention of laboratories using this scheme to prevent inaccuracies in quantitative labeling due to loss of the isotopic label.     

Optimization and quality assessment of the post-digestion O labeling based on urea for protein denaturation by HPLC/ESI-TOF mass spectrometry

The post-digestion ¹⁸O labeling method decouples protein digestion and peptide labeling. This method allows labeling conditions to be optimized separately and increases labeling efficiency. A common method for protein denaturation in proteomics is the use of urea. Though some previous studies have used urea-based protein denaturation before post-digestion ¹⁸O labeling, the optimal ¹⁸O labeling conditions in this case have not been yet reported. Present study investigated the effects of urea concentration and pH on the labeling efficiency and obtained an optimized protocol. It was demonstrated that urea inhibited ¹⁸O incorporation depending on concentration. However, a urea concentration between 1 and 2 M had minimal effects on labeling. It was also demonstrated that the use of FA to quench the digestion reaction severely affected the labeling efficiency. This study revealed the reason why previous studies gave different optimal pH for labeling. They neglect the effects of different digestion conditions on the labeling conditions. Excellent labeling quality was obtained at the optimized conditions using urea 1–2 M and pH 4.5, 98.4 ± 1.9% for a standard protein mixture and 97.2 ± 6.2% for a complex biological sample. For a 1:1 mixture analysis of the ¹⁶O- and ¹⁸O-labeled peptides from the same protein sample, the average abundance ratios reached 1.05 ± 0.31, demonstrating a good quantitation quality at the optimized conditions. This work will benefit other researchers who pair urea-based protein denaturation with a post-digestion ¹⁸O labeling method.     

Differential dimethyl labeling of N-termini of peptides after guanidination for proteome analysis

We describe an enabling technique for proteome analysis based on isotope-differential dimethyl labeling of N-termini of tryptic peptides followed by microbore liquid chromatography (LC) matrix-assisted laser desorption and ionization (MALDI) mass spectrometry (MS). In this method, lysine side chains are blocked by guanidination to prevent the incorporation of multiple labels, followed by N-terminal labeling via reductive amination using d(0),(12)C-formaldehyde or d(2),(13)C-formaldehyde. Relative quantification of peptide mixtures is achieved by examining the MALDI mass spectra of the peptide pairs labeled with different isotope tags. A nominal mass difference of 6 Da between the peptide pair allows negligible interference between the two isotopic clusters for quantification of peptides of up to 3000 Da. Since only the N-termini of tryptic peptides are differentially labeled and the a(1) ions are also enhanced in the MALDI MS/MS spectra, interpretation of the fragment ion spectra to obtain sequence information is greatly simplified. It is demonstrated that this technique of N-terminal dimethylation (2ME) after lysine guanidination (GA) or 2MEGA offers several desirable features, including simple experimental procedure, stable products, using inexpensive and commercially available reagents, and negligible isotope effect on reversed-phase separation. LC-MALDI MS combined with this 2MEGA labeling technique was successfully used to identify proteins that included polymorphic variants and low abundance proteins in bovine milk. In addition, by analyzing a mixture of two equal amounts of milk whey fraction as a control, it is shown that the measured average ratio for 56 peptide pairs from 14 different proteins is 1.02, which is very close to the theoretical ratio of 1.00. The calculated percentage error is 2.0% and relative standard deviation is 4.6%.     

Proteolytic 18O labeling by peptidyl-Lys metalloendopeptidase for comparative proteomics

The potential capabilities of a new proteolytic 18O labeling method employing peptidyl-Lys metalloendopeptidase (Lys-N) have been demonstrated for use in comparative proteomics. Conditions (pH>or=9.5) have been found such that Lys-N incorporates only a single 18O atom into the carboxyl terminus of each proteolytically generated peptide. This 18O labeling method has a major advantage over current protelytic 18O labeling methods that generate a mixture of isotopic isoforms resulting from the incorporation of one or two 18O atoms into each peptide species by the proteases (trypsin, Lys-C, or Glu-C) used. We demonstrate that the single 18O atom incorporation property of Lys-N overcomes the major problem of the current proteolytic 18O labeling methods and provides accurate quantification results for isotopically labeled peptides.     

Quantitative Proteomics Using Reductive Dimethylation for Stable Isotope Labeling

Stable isotope labeling of peptides by reductive dimethylation (ReDi labeling) is a method to accurately quantify protein expression differences between samples using mass spectrometry. ReDi labeling is performed using either regular (light) or deuterated (heavy) forms of formaldehyde and sodium cyanoborohydride to add two methyl groups to each free amine. Here we demonstrate a robust protocol for ReDi labeling and quantitative comparison of complex protein mixtures. Protein samples for comparison are digested into peptides, labeled to carry either light or heavy methyl tags, mixed, and co-analyzed by LC-MS/MS. Relative protein abundances are quantified by comparing the ion chromatogram peak areas of heavy and light labeled versions of the constituent peptide extracted from the full MS spectra. The method described here includes sample preparation by reversed-phase solid phase extraction, on-column ReDi labeling of peptides, peptide fractionation by basic pH reversed-phase (BPRP) chromatography, and StageTip peptide purification. We discuss advantages and limitations of ReDi labeling with respect to other methods for stable isotope incorporation. We highlight novel applications using ReDi labeling as a fast, inexpensive, and accurate method to compare protein abundances in nearly any type of sample.     

Conformational changes of recombinant monoclonal antibodies by limited proteolytic digestion,stable isotope labeling,and liquid chromatography–mass spectrometry

Limited proteolytic digestion is a method with a long history that has been used to study protein domain structures and conformational changes. A method of combining limited proteolytic digestion, stable isotope labeling, and mass spectrometry was established in the current study to investigate protein conformational changes. Recombinant monoclonal antibodies with or without the conserved oligosaccharides, and with or without oxidation of the conserved methionine residues, were used to test the newly proposed method. All of the samples were digested in ammonium bicarbonate buffer prepared in normal water. The oxidized deglycosylated sample was also digested in ammonium bicarbonate buffer prepared in ¹⁸O-labeled water. The sample from the digestion in ¹⁸O–water was spiked into each sample digested in normal water. Each mixed sample was subsequently analyzed by liquid chromatography–mass spectrometry (LC–MS). The molecular weight differences between the peptides digested in normal water versus ¹⁸O–water were used to differentiate peaks from the samples. The relative peak intensities of peptides with or without the C-terminal incorporation of ¹⁸O atoms were used to determine susceptibility of different samples to trypsin and chymotrypsin. The results demonstrated that the method was capable of detecting local conformational changes of the recombinant monoclonal antibodies caused by deglycosylation and oxidation.     

Method for qualitative comparisons of protein mixtures based on enzyme-catalyzed stable-isotope incorporation

Determining which proteins are unique among one or several protein populations is an often-encountered task in proteomics. To this purpose, we present a new method based on trypsin-catalyzed incorporation of the stabile isotope (18)O in the C-termini of tryptic peptides, followed by LC-MALDI MS analysis. The analytical strategy was designed such that proteins unique to a given population out of several can be assigned in a single experiment by the isotopic signal intensity distributions of their tryptic peptides in the recorded mass spectra. The method is demonstrated for protein-protein interaction analysis, in which the differential isotope labeling was used to distinguish endogenous human brain proteins interacting with a recombinant bait protein from nonbiospecific background binders.     

Quantitative profiling of the detergent-resistant membrane proteome of iota-b toxin induced vero cells

Enzyme-mediated 18O/16O differential labeling of proteome samples often suffers from incomplete exchange of the carboxy-terminus oxygen atoms, resulting in ambiguity in the measurable abundance differences. In this study, an 18O/16O labeling strategy was optimized for and applied to the solution-based comparative analysis of the detergent-resistant membrane proteome (DRMP) of untreated and Iota-b (Ib)-induced Vero cells. Solubilization and tryptic digestion of the DRMP was conducted in a buffer containing 60% methanol. Unfortunately, the activity of trypsin is attenuated at this methanol concentration hampering the ability to obtain complete oxygen atom turnover. Therefore, the incorporation of the 18O atoms was decoupled from the protein digestion step by carrying out the trypsin-mediated heavy atom incorporation in a buffer containing 20% methanol; a concentration at which trypsin activity is enhanced compared to purely aqueous conditions. After isotopic labeling, the samples were combined, fractionated by strong cation exchange and analyzed by microcapillary reversed-phase liquid chromatography coupled on-line with electrospray ionization tandem mass spectrometry. In total, over 1400 unique peptides, corresponding to almost 600 proteins, were identified and quantitated, including all known caveolar and lipid raft marker proteins. The quantitative profiling of Ib-induced DRMP from Vero cells revealed several proteins with altered expression levels suggesting their possible role in Ib binding/uptake.     

In-depth quantitative cardiac proteomics combining electron transfer dissociation and the metalloendopeptidase Lys-N with the SILAC mouse

In quantitative proteomics stable isotope labeling has progressed from cultured cells toward the total incorporation of labeled atoms or amino acids into whole multicellular organisms. For instance, the recently introduced (13)C(6)-lysine labeled SILAC mouse allows accurate comparison of protein expression directly in tissue. In this model, only lysine, but not arginine, residues are isotope labeled, as the latter may cause complications to the quantification by in vivo conversion of arginine to proline. The sole labeling of lysines discourages the use of trypsin, as not all peptides will be quantifiable. Therefore, in the initial work Lys-C was used for digestion. Here, we demonstrate that the lysine-directed protease metalloendopeptidase Lys-N is an excellent alternative. As lysine directed peptides generally yield longer and higher charged peptides, alongside the more traditional collision induced dissociation we also implemented electron transfer dissociation in a quantitative stable isotope labeling with amino acid in cell culture workflow for the first time. The utility of these two complementary approaches is highlighted by investigating the differences in protein expression between the left and right ventricle of a mouse heart. Using Lys-N and electron transfer dissociation yielded coverage to a depth of 3749 proteins, which is similar as earlier investigations into the murine heart proteome. In addition, this strategy yields quantitative information on ～ 2000 proteins with a median coverage of four peptides per protein in a single strong cation exchange-liquid chromatography-MS experiment, revealing that the left and right ventricle proteomes are very similar qualitatively as well as quantitatively.     

18O Labeled Nucleosides. 2. A General Method for the Synthesis of Specifically Labeled Pyrimidine Ribonucleosides1

Abstract

A facile method for the synthesis of highly enriched ¹⁸O labeled pyrimidine ribonucleosides is described using uridine as a model compound. The isotopic label may be selectively incorporated into the base moiety at O² or into the ribose portion of the molecule at the 5′ position. In addition, both positions may be labeled and this is the first report of a method for labeling of both the base and sugar moieties of pyrimidine ribonucleosides. The site and level of isotope incorporation may be determined mass spectrometrically.     

Mechanism of formation of the ester linkage between heme and Glu310 of CYP4B1: 18O protein labeling studies

Cytochrome P450s in the CYP4 family covalently bind their heme prosthetic group to a conserved acidic I-helix residue via an autocatalytic oxidation. This study was designed to evaluate the source of oxygen atoms in the covalent ester link in CYP4B1 enzymes labeled with [18O]glutamate and [18O]aspartate. The fate of the heavy isotope was then traced into wild-type CYP4B1 or the E310D mutant-derived 5-hydroxyhemes. Glutamate-containing tryptic peptides of wild-type CYP4B1 were found labeled to a level of 11-13% 18O. Base hydrolysis of labeled protein released 5-hydroxyheme which contained 12.8 +/- 1.9% 18O. Aspartate-containing peptides of the E310D mutant were labeled with 6.0-6.5% 18O, but as expected, no label was transmitted to recovered 5-hydroxyheme. These data demonstrate that the oxygen atom in 5-hydroxyheme derived from wild-type CYP4B1 originates in Glu310. Stoichiometric incorporation of the heavy isotope from the wild-type enzyme supports a perferryl-initiated carbocation mechanism for covalent heme formation in CYP4B1.     

Improved method for differential expression proteomics using trypsin-catalyzed 18O labeling with a correction for labeling efficiency

Quantitative strategies relying on stable isotope labeling and isotope dilution mass spectrometry have proven to be a very robust alternative to the well established gel-based techniques for the study of the dynamic proteome. Postdigestion 18O labeling is becoming very popular mainly due to the simplicity of the enzyme-catalyzed exchange reaction, the peptide handling and storage procedures, and the flexibility and versatility introduced by decoupling protein digestion from peptide labeling. Despite recent progresses, peptide quantification by postdigestion 18O labeling still involves several computational problems. In this work we analyzed the behavior of large collections of peptides when they were subjected to postdigestion labeling and concluded that this process can be explained by a universal kinetic model. On the basis of this observation, we developed an advanced quantification algorithm for this kind of labeling. Our method fits the entire isotopic envelope to parameters related with the kinetic exchange model, allowing at the same time an accurate calculation of the relative proportion of peptides in the original samples and of the specific labeling efficiency of each one of the peptides. We demonstrated that the new method eliminates artifacts produced by incomplete oxygen exchange in subsets of peptides that have a relatively low labeling efficiency and that may be considered indicative of false protein ratio deviations. Finally using a rigorous statistical analysis based on the calculation of error rates associated with false expression changes, we showed the validity of the method in the practice by detecting significant expression changes, produced by the activation of a model preparation of T cells, with only 5 microg of protein in three proteins among a pool of more than 100. By allowing a full control over potential artifacts, our method may improve automation of the procedures for relative protein quantification using this labeling strategy.     

A novel quantitative proteomics workflow by isobaric terminal labeling

Quantification by series of b, y fragment ion pairs generated from isobaric-labeled peptides in MS2 spectra has recently been considered an accurate strategy in quantitative proteomics. Here we developed a novel MS2 quantification approach named quantitation by isobaric terminal labeling (QITL) by coupling (18)O labeling with dimethylation. Trypsin-digested peptides were labeled with two (16)O or (18)O atoms at their C-termini in H(2)(16)O or H(2)(18)O. After blocking all ε-amino groups of lysines through guanidination, the N-termini of the peptides were accordingly labeled with formaldehyde-d(2) or formaldehyde. These indistinguishable, isobaric-labeled peptides in MS1 spectra produce b, y fragment ion pairs in the whole mass range of MS2 spectra that can be used for quantification. In this study, the feasibility of QITL was first demonstrated using standard proteins. An accurate and reproducible quantification over a wide dynamic range was achieved. Then, complex rat liver samples were used to verify the applicability of QITL for large-scale quantitative analysis. Finally, QITL was applied to profile the quantitative proteome of hepatocellular carcinoma (HCC) and adjacent non-tumor liver tissues. Given its simplicity, low-cost, and accuracy, QITL can be widely applied in biological samples (cell lines, tissues, and body fluids, etc.) for quantitative proteomic research.     

Catalytic and noncatalytic nucleotide binding sites of the Escherichia coli F1 ATPase. Amino acid sequences of beta-subunit tryptic peptides labeled with 2-azido-ATP

Under appropriate conditions tight, noncovalent binding of 2-azido-adenine nucleotides to either catalytic or noncatalytic binding sites on the E. coli F1-ATPase occurs. After removal of unbound ligands, UV-irradiation results primarily in the covalent incorporation of nucleotide moieties into the beta-subunit in both catalytic and noncatalytic site labeling experiments. Minor labeling of the alpha-subunit was also observed. After trypsin digestion and purification of the labeled peptides, microsequencing studies identified two adjacent beta-subunit tryptic peptides labeled by 2-azido-ADP or -ATP. These beta-subunit peptides were labeled on tyrosine-331 (catalytic sites) and tyrosine-354 (noncatalytic sites) in homology with the labeling patterns of the mitochondrial and chloroplast enzymes.     

Identifying and quantifying proteolytic events and the natural N terminome by terminal amine isotopic labeling of substrates

Analysis of the sequence and nature of protein N termini has many applications. Defining the termini of proteins for proteome annotation in the Human Proteome Project is of increasing importance. Terminomics analysis of protease cleavage sites in degradomics for substrate discovery is a key new application. Here we describe the step-by-step procedures for performing terminal amine isotopic labeling of substrates (TAILS), a 2- to 3-d (depending on method of labeling) high-throughput method to identify and distinguish protease-generated neo-N termini from mature protein N termini with all natural modifications with high confidence. TAILS uses negative selection to enrich for all N-terminal peptides and uses primary amine labeling-based quantification as the discriminating factor. Labeling is versatile and suited to many applications, including biochemical and cell culture analyses in vitro; in vivo analyses using tissue samples from animal and human sources can also be readily performed. At the protein level, N-terminal and lysine amines are blocked by dimethylation (formaldehyde/sodium cyanoborohydride) and isotopically labeled by incorporating heavy and light dimethylation reagents or stable isotope labeling with amino acids in cell culture labels. Alternatively, easy multiplex sample analysis can be achieved using amine blocking and labeling with isobaric tags for relative and absolute quantification, also known as iTRAQ. After tryptic digestion, N-terminal peptide separation is achieved using a high-molecular-weight dendritic polyglycerol aldehyde polymer that binds internal tryptic and C-terminal peptides that now have N-terminal alpha amines. The unbound naturally blocked (acetylation, cyclization, methylation and so on) or labeled mature N-terminal and neo-N-terminal peptides are recovered by ultrafiltration and analyzed by tandem mass spectrometry (MS/MS). Hierarchical substrate winnowing discriminates substrates from the background proteolysis products and non-cleaved proteins by peptide isotope quantification and bioinformatics search criteria.     

Characterization of growing microorganisms in soil by stable isotope probing with H218O

A new approach to characterize growing microorganisms in environmental samples based on labeling microbial DNA with H(2)(18)O is described. To test if sufficient amounts of (18)O could be incorporated into DNA to use water as a labeling substrate for stable isotope probing, Escherichia coli DNA was labeled by cultivating bacteria in Luria broth with H(2)(18)O and labeled DNA was separated from [(16)O]DNA on a cesium chloride gradient. Soil samples were incubated with H(2)(18)O for 6, 14, or 21 days, and isopycnic centrifugation of the soil DNA showed the formation of two bands after 6 days and three bands after 14 or 21 days, indicating that (18)O can be used in the stable isotope probing of soil samples. DNA extracted from soil incubated for 21 days with H(2)(18)O was fractionated after isopycnic centrifugation and DNA from 17 subsamples was used in terminal restriction fragment length polymorphism (TRFLP) analysis of bacterial 16S rRNA genes. The TRFLP patterns clustered into three groups that corresponded to the three DNA bands. The fraction of total fluorescence contributed by individual terminal restriction fragments (TRF) to a TRFLP pattern varied across the 17 subsamples so that a TRF was more prominent in only one of the three bands. Labeling soil DNA with H(2)(18)O allows the identification of newly grown cells. In addition, cells that survive but do not divide during an incubation period can also be characterized with this new technique because their DNA remains without the label.     

Lysyl residues of cardiac myosin accessible to labeling with a fluorescent reagent, N-methyl-2-anilino-6-naphthalenesulfonyl chloride

Two mol of N-methyl-2-anilino-6-naphthalenesulfonyl (Mns) groups was preferentially incorporated into pig cardiac myosin in the absence of divalent metal ions, 1 mol rapidly and 1 mol slowly. In the presence of divalent metal ions. 1 mol was rapidly incorporated but subsequent incorporation was strongly suppressed. No substantial effect of incorporation of Mns groups in the presence or absence of divalent metal ions on the Ca2+- and K+-ATPase activities of myosin was found. However, the fluorescence spectra due to attached Mns groups were different in the two cases. Extensive pronase digestion of labeled myosin indicated that the Mns groups were attached predominantly to lysyl residues, regardless of the labeling conditions. Peptide mapping of the labeled myosin digested with subtilisin, pepsin or trypsin uniformly showed the selective incorporation of an Mns group into essentially one species of peptide. However, the peptide labeled in the absence of divalent metal ions was clearly different from that labeled in their presence. The present results confirm that pig cardiac myosin heavy chains contain two distinct lysyl residues, which are both accessible to labeling with Mns groups only when divalent metal ions are absent. The results also suggest that conformational changes occur around these residues when divalent metal ions are added.