Trypsin is the Primary Mechanism by which the ¹⁸O Isotopic Label is Lost in Quantitative Proteomic Studies

Abstract

Labeling with ¹⁸O 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 ¹⁸O 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 trypsin plays in isotopic loss, several experiments were performed on the rate of de-labeling under conditions relevant to multidimensional proteomic experiments. Results from these studies demonstrate that enzyme facilitated exchange of ¹⁸O in the peptide with ¹⁶O 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 in order to prevent inaccuracies in quantitative labeling due to loss of the isotopic label.

Introduction

Quantitative proteomics attempts to provide changes in protein expression between two biologically different states for purposes of disease research or as a complement to genomic research. The field of shotgun proteomics has developed numerous methods for correlating various features of the eluting peptides to the abundance levels of the proteins in the proteome. Label free methods using normalized ion intensities [1], spectral counts [2], mass, scan number and signal intensity [3], or accurate mass plus retention time [4] have all been used with success to link these aspects of eluting peptides to protein expression level for comparative quantitation. However, ion intensities and retention times are variable, limiting the accuracy of these approaches. The alternative strategy for relative quantitation involves the simultaneous analysis of isotopically labeled and unlabeled peptides. This approach has advantages over label free methods in that the heavy and light labeled peptide pairs elute together, allowing a direct comparison of relative abundances for that peptide and that two or more differential populations may be combined for a single analysis, greatly decreasing the time spent analyzing the samples.

Recent years have seen a broad array of strategies for introduction of the stable isotope to a particular population. The isotope may be introduced in vivo by metabolic incorporation of a stable isotopically labeled amino acid in cell culture (SILAC) [5], or by using media enriched for particular isotope(s) such as 13C, 15N or 2H [6,7]. These methods produce consistent incorporation of an isotope into a population. However, metabolic incorporation is not practical for differential comparison of whole animals. Methods to introduce a stable isotope in vitro have thus been developed. For example, isotope-coded affinity tags (ICAT) [8] chemically target specific amino acids, typically cysteine, in the peptide sequence for differential labeling. A potential limitation of strategies targeting specific amino acids, such as ICAT, is that only populations that contain the target residue can be selected for quantification purposes. Other in vitro approaches target functional groups common to all peptides, such as the primary amines of the N-termini and/or the carboxyl groups of the C-termini [9-14]. This allows all observable peptides from each protein to be used for quantification purposes.

¹⁸O labeling utilizing tryptic activity to label the C-termini of peptides with the stable isotope [9] is the most widely used universal labeling procedure for quantitative proteomics and has been applied to a wide variety of analyses, most notably including samples of limited amount [15] and membrane populations [16]. The resulting mass shift from ¹⁸O incorporation does not alter the chromatographic separation or the ionization efficiency of the labeled peptides, thus providing a versatile, nondiscriminatory, global labeling system for relative quantitative proteomics. ¹⁸O labeling for proteomics is different from other methods of isotopic incorporation that target universally present functional groups in that the mechanism is enzyme driven, allowing fast incorporation of the label without jeopardizing biologically derived modifications that may be of interest. The technique is performed by enzymatically digesting the proteins in ¹⁸O enriched water using trypsin or another serine protease, thus producing a +4 Da shift with the incorporation of the two ¹⁸O isotopes. Samples may also be labeled after digestion has occurred, which permits smaller volumes of ¹⁸O water to be used while allowing conditions to be optimized separately for digestion and the labeling of a particular population [17,18]. Enzyme facilitated labeling after digestion is possible due to the cleaved peptide acting as a pseudo substrate to serine proteases (reactions III_F+IV_F/III_R+IV_R in Figure 1) [9,18,19]. It is for this reason that trypsin [9], chymotrypsin, [18], Lys-C [9], and Glu-C [9,20] catalyze the addition of two ¹⁸O atoms into the C terminus of the proteolytic peptides. It should be noted that other classes of proteases can lead to the incorporation of a single ¹⁸O isotope, as has been reported using the metalloendopeptidase Lys-N [10]. However, trypsin is by far the most commonly used enzyme because it consistently produces uniform peptides that are amenable to interpretable fragmentation by tandem mass spectrometry [21].

Figure 1.

A mechanism depicting the process by which peptides are cleaved from the protein and isotopically labeled, where E is the enzyme, Pro is the intact protein, Pep-K/R is the tryptic peptide terminated with either lysine or arginine, and Pep is the other cleaved peptide. It is likely that the optimal conditions for Reaction III where the enzyme is interacting with the cleaved peptide, are different than the conditions for Reaction I and Reaction II where the enzyme is interacting with the intact protein to induce cleavage of the peptide backbone.

Comparative proteomics relies on minimal loss of the isotopic label for accurate relative quantitation. For ¹⁸O labeling, the loss of the isotopic label has been reported via chemical reaction at low pHs [9,22] as well as through the same enzyme facilitated mechanisms that were used to apply the isotopic label [9, 18,19,22], which appears to be the faster process. The use of immobilized trypsin eliminates this process as it allows the trypsin to be removed from the sample after the labeling step. [23] Other approaches have minimized the loss of the isotopic label via the enzymatic route by employing conditions that deactivate trypsin, such as reducing and alkylating after labeling [22], boiling the enzyme at digestive pH followed by lowering the pH [16] or the more common practice of inhibiting the tryptic mechanism by lowered pH of the sample. [9, 15,16, 19] These disabling approaches are based on the prediction that conditions which minimize the proteolytic rate of trypsin (reactions I - IV in Figure 1) will lead to a similar decrease in the rate of isotopic loss from enzymatic back exchange (reactions III_F+IV_F/III_R+IV_R in Figure 1).

Here we present a study demonstrating that trypsin is the main facilitator of loss of the ¹⁸O label even at low pH and temperature, conditions that are often used in a proteomic study. Since ¹⁸O isotopic labeling is currently one of the most utilized isotopic labeling techniques available for a wide range of in vitro samples types, it is our aim to bring the rapid, tryptic facilitated exchange to the attention of laboratories using this scheme in order to prevent inaccuracies in quantitative labeling due to loss of the isotopic label.

Materials and Methods

Materials

α-Casein, ammonium bicarbonate, trifluoroacetic acid, α-cyanohydroxy cinnamic acid (CHCA), L-1-Chloro-3-[4-tosyl-amido]-7-amino-2-heptanone-HCI (TLCK) were purchased from Sigma (St. Louis, MO). Isotopic water (95% ¹⁸O) was supplied by Isotec (Miamisburg, OH). Urea, formic acid and acetonitrile were supplied by J.T. Baker (Phillipsburg NJ). Sequencing grade trypsin was purchased from Promega (Madison, WI). Immobilized trypsin beads were purchased from Pierce (Rockford, IL).

Experimental

A solution of approximately 50 pmole/μL α-casein in 50 mM ammonium bicarbonate was digested overnight at 37°C with trypsin using a 1:50 ratio of protease to substrate (w/w). For all cases, ¹⁸O labeling was performed after this initial digest using the following protocol: A 60 μL portion of the digested protein was dried under vacuum for 30 min at < 37°C. A 60 μL volume of 50 mM ammonium bicarbonate in ¹⁸O enriched water, prepared by adding the appropriate amount of ammonium bicarbonate salt to the isotopically enriched water, was added to the samples, and labeling allowed to proceed overnight. No additional trypsin was needed as previous studies had shown that trypsin remained viable after the drying process had occurred.

Loss of label in the absence of trypsin, molecular weight cutoff filtration

A 400 μL portion of the 50 picomole/μL tryptically digested α-casein peptides was filtered with a molecular weight cutoff filter of 10,000 (Amicron; Beverly, MA) to remove trypsin. After filtration, the ¹⁸O labeled peptides were diluted 1:15 with 5% formic acid to reach a pH of 2. A sample at a pH of 7 was prepared in the same fashion with dilution of the labeled peptides in natural abundance water. Back exchange was measured once every 10 min for the first hour, once an hour for 5 h after that, and then once every 24 h thereafter.

Loss of label in the absence of trypsin, immobilized beads

Exactly 20 μL of immobilized trypsin beads was added to a 50 μL solution of 50 picomole/μL solution of α-casein protein in natural abundance water. The sample was dried by vacuum centrifugation and reconstituted in 40 μL of 50 mM ammonium bicarbonate prepared in 95% H₂¹⁸O with 10 μL acetonitrile and incubated at room temperature overnight with shaking. After incubation, the sample was spun at 10,000 g for 3 min. A portion of the supernatant was removed and diluted 1:10 in either formic acid or water to attain the desired pH of either 2 or 7. Back exchange was tracked by 1:5 dilution in the MALDI matrix at time zero, 6 h after time zero, and every 24 h thereafter.

Loss of label, temperature dependent

For pH studies, 20 μL of the 50 picomole/μL labeled solution was diluted 1:10 with varying concentrations of formic acid to reach a pH of either 2 or 7. The pH of each solution was checked using litmus paper (Colorphast, EMD Chemicals, Gibbstown, NJ). Back exchange was measured at room temperature (approximately 23°C) and at 4°C every 10 min for the first 50 min, every 25 min for the next 75 min, and once 24 h after time zero. At the specified time, a portion of the sample was withdrawn and diluted 1:5 with MALDI matrix, and analyzed by MALDI-MS, as described below. For the experiment performed at 4° C, samples and matrix were kept on ice during the course of the investigation, with overnight storage on ice inside a refrigerator maintained at 2-8 °C.

Loss of label in the presence of high organic and low pH

The effects of lyophilization and reconstitution in strong cation exchange buffer were examined. A fully labeled solution of α-casein sample was diluted 1:1 in natural abundance water. The sample was instantly frozen within seconds after addition of the natural abundance water by suspending the sample in liquid nitrogen for 15 sec. The sample was immediately lyophilized. To prevent melting before sublimination of the sample, the sample was maintained on an alcohol/dry ice bath. Visual inspection confirmed that no melting of the samples occurred before the drying process had begun. After lyophilization, samples were reconstituted in 20% methanol, 0.1% formic acid, imitating a strong cation exchange separation. Back exchange was measured at room temperature (approximately 23°C) every 10 min for the first 50 min, every 25 min for the next 75 min, and once 24 h after time zero. At the specified time, a portion of the sample was withdrawn and diluted 1:5 in MALDI matrix, and analyzed by MALDI-MS, as described below.

Loss of label, lowered pH with heat denaturation

The effects of simultaneous heat denaturation and lowering of pH versus isotopic loss were measured. The fully labeled solution of α-casein was adjusted to a pH of 2 using concentrated formic acid. The small amount of natural abundance water in the formic acid did not impact the study, since the relative change of label was being recorded. The solution was then placed in boiling water for 10 min. After cooling to room temperature, a fraction of this solution was diluted in natural abundance water to a final pH of 7. Back exchange was measured at room temperature (approximately 23°C) every 10 min for the first 50 min, every 25 min for the next 75 min, and once 24 h after time zero. At the specified time, a portion of the sample was withdrawn and diluted 1:5 in MALDI matrix, and analyzed by MALDI-MS, as described below.

Loss of label, lowered pH with inhibitor

L-1-Chloro-3-[4-tosyl-amido]-7-amino-2-heptanone-HCI (TLCK) was used to inhibit tryptic activity. This molecule is not stable at pHs greater that 7.5, therefore a stock solution of 1 mg/mL TLCK was prepared in 5% formic acid/natural abundance water. Exactly 1 μL of the 1 mg/mL TLCK stock solution was diluted 1:50 in 95% ¹⁸O labeled water to a final concentration of 20 μg/mL TLCK. The pH of a labeled sample solution was lowered to below pH 6.5 by addition of 1 μL of 10% formic acid. TLCK was added in a ratio of 1:30 enzyme: inhibitor to a 50 μL portion of labeled α-casein sample. The labeled sample with TLCK was incubated overnight at room temperature. A 10 μL aliquot of the sample was diluted to 100 μL in 0.1% formic acid/natural abundance water to a measured pH of 2. Back exchange was measured at time zero, every 10 min for the first 50 min, followed by time points at 75 min and 100 min. A final measurement was taken at 24 h from time zero.

MALDI-MS

All samples were analyzed by MALDI-ToF on an ABI 4700 Proteomics Analyzer (Applied Biosystems, Applied Biosystems, Foster City, CA) equipped with an Nd:YAG laser operating at 355 nm with a 200 Hz laser rate. All samples were acquired in the positive ion reflector mode with an acquisition mass range from 900 - 2000 m/z and a focus at 1500 m/z. Each spectrum was an accumulation of 1000 shots obtained with a laser setting of 3600, accelerating voltage of 20 kV, source chamber pressure of 6.0 × 10⁻⁸ torr, and a reflector chamber pressure of 2.0 × 10⁻⁸ torr. External calibration was performed using four standards des-arg¹-bradykinin (904.468), angiotensin I (1296.685), Glu¹-fibrinopeptide B (1570.678), and neurotensin (1672.92) ( singly charged monoisotopic species denoted).

Samples for all measurements were mixed with α-cyanohydroxy cinnamic acid (CHCA) to a 1 picomole/μL concentration and spotted onto a MALDI target plate. Mixing with the MALDI matrix and spotting were performed in less than 1 min. Back exchange of a sample dried onto the MALDI target was measured at time zero and after 24 h and found to be negligible. For all experiments, where possible, the same stock solution of labeled digest was used to minimize variances due to enzyme, peptide, and ¹⁸O water concentration. Peptide mass fingerprinting was performed using Protein prospector (http://prospector.ucsf.edu) against the Swissprot databank with the settings enzyme- trypsin, species bos taurus, one missed cleavage, peptide tolerance 200 ppm, instrument setting MALDI-TOF. Calculation of isotopic contributions was performed using MS-Isotope (http://prospector.ucsf.edu), using the peptide sequence function for known peptides, and the averagine function to calculate isotopic contribution for unknown peptides.

Results

Proteolysis with trypsin begins with association of the polypeptide chain with the enzyme by formation of a hydrogen bond between the peptide backbone and the enzyme [24] and subsequent locking of the charged residue onto the catalytic serine residue [25] (Fig. 1, reaction I). This step is followed by cleavage of the amide bond after arginine (Arg) and lysine (Lys) residues (Fig. 1, reaction II). During this process, the newly created peptide terminating with Lys/Arg becomes covalently attached to the enzyme. Nucleophilic attack by water on the peptide-enzyme complex liberates the peptide from the enzyme and results in the incorporation of a single oxygen atom from the solvent into the C-terminus of the newly formed peptide (Fig. 1, reactions III_F and IV_F). The Lys/Arg terminated peptide may again bind to the active site of the enzyme releasing one of the two equivalent C-terminal oxygen atoms (Fig. 1, reactions III_R and IV_R). Once again this peptide-enzyme complex dissociates after hydrolysis, incorporating another oxygen atom from the solvent into the C-terminus. Since Lys/Arg terminated peptides are in constant equilibrium with the peptide-enzyme complex, these two reactions (Fig. 1, reactions III_F + IV_F and III_R + IV_R) occur repeatedly and explain why two ¹⁸O labels are incorporated into the C-terminus of trypsin generated peptides when this process is performed in water highly enriched with the isotope. This equilibrium implies that the extent of isotopic incorporation on the peptides will proceed towards that of the solvent in which they are dissolved. In other words, ¹⁸O containing peptides will lose their isotopic labels, i.e., be delabeled, when the isotopic content of the solvent is decreased. A similarly problematic, although rarely discussed, consequence of this equilibrium is that non-labeled peptides will incorporate ¹⁸O (i.e., become labeled) when they are introduced to an isotopically enriched solvent. This situation can occur when a ¹⁶O labeled sample dissolved in ¹⁶O water is combined with an ¹⁸O labeled sample in ¹⁸O water. These concerns prompted us to evaluate the extent of isotopic scrambling resulting from tryptic activity under conditions applicable to a proteomic scheme.

Initial studies on the loss of isotopic label were performed with α-casein that was post digestively labeled and thus contained the maximum amount of double isotopic incorporation [16-18]. The extent of ¹⁸O delabeling was determined by MALDI-TOF analysis. MALDI was selected for these studies because it allows a shorter time period from sample preparation to analysis than either ESI-MS or LC-MS, and thus offered the smallest possible window for loss of the isotope and a more accurate representation of the isotopic abundances of the peptides in solution. Prior to these studies, the extent of back exchange in the MALDI matrix was experimentally determined to be below the detection limits (data not shown), which is in agreement with a previous report [19].

Early studies identified that the primary process limiting isotopic incorporation was the initial proteolytic cleavage (reactions I-IV_f) [18]. The use of a post-digestive labeling strategy circumvents this issue. [17,18] In this latter approach, it is expected that the rate at which ¹⁸O is incorporated will be similar to the rate of enzyme facilitated back exchange since both of these processes proceed via reactions III_f+IV_f/III_r+IV_r, implying that peptides which are slow to label will also experience slow back exchange. To account for variability introduced by differing rates of back exchange due to peptide length or composition [18,19], and thus establish a general trend, five peptides with mass to charge ratios (m/z) of 971, 1134, 1271, 1384, and 1759 were followed under the various conditions described in subsequent paragraphs. An average of these five peptides was used to calculate the relative percent double label remaining at the selected time point. All isotopic incorporation percentages were calculated for the double labeled peptides via dividing the peak area from the double labeled ion by the sum of peak areas from the unlabeled, single labeled and double labeled ions after adjusting these ion areas for the naturally occurring isotopes of the peptides analyzed. Differences in labeling efficiencies between experiments were accounted for by calculating the loss of the double labeled ion relative to the value obtained at time zero for each set of conditions. This allowed for direct comparisons of the extent of isotopic loss from each set of experiments. Loss of the double ¹⁸O label was followed as the resulting mass shift (+4 Da) minimizes isotopic overlap and thus produces more reliable quantitation.

The conditions for this investigation were selected to imitate those that would be typically encountered in a proteomic study. In particular, a pH of 2 was intended to mimic a sample being re-suspended in an acidic solution prior to injection onto a reversed phase column, while a pH of 7 mimics combining samples immediately after digesting/labeling without adjustment of pH. Temperature was also taken into account, with the de-labeling process being investigated at both room temperature and at 4°C to account for cases where the sample is placed in an auto sampler with a cooling system. While these conditions do not represent every conceivable possibility, they do cover a wide spectrum of those typically used in proteomic studies and demonstrate general trends for trypsin catalyzed isotopic loss.

To set a benchmark for chemical exchange versus enzymatic exchange, the loss of the double ¹⁸O label was first observed at acidic and neutral pH in the absence of trypsin while at room temperature. Two different samples were prepared for this study. One of these samples consisted of α-casein digested/labeled in ¹⁸O water, followed by removal of the trypsin with a 10,000 molecular weight cutoff filter. Once trypsin was removed, samples were diluted 1:10 with natural abundance water or a formic acid solution to reach a pH of 7 or 2. In this sample, the relative percent loss of double label after 9 d storage at room temperature was 3.6 ± 6.3 at pH 7, and 16.2 ± 3.1% at pH 2 (Table 1 and Fig. 2). A second sample was prepared by digesting/labeling α-casein in ¹⁸O water using immobilized trypsin. This solution was centrifuged and the supernatant removed for examination of back exchange, as previously reported [23]. The loss of label from the sample at pH 7 was 4.2 ± 3.3% after 9 d (Table 1 and Fig. 2), while that of the same sample stored at pH 2 was 11.2 ± 6.4% for this time period (Table 1 and Fig. 2). The extent of back exchange for these two samples are not experimentally distinguishable, and agree well with previous reports on chemical methods for incorporating/exchanging oxygen atoms into the carboxylic groups of peptides under acidic conditions [19,26]. These values are also in agreement with a study that utilized acid catalyzed labeling of the peptide [27].

Table 1.

Summary of the results for the relative percent loss of double ¹⁸O label. Percent double label at each time point equals (¹⁸O2 peak/[¹⁶O peak + ¹⁸O2 peak]) - isotopic contributions × 100. The percent double label for each peptide was calculated and an average over all peptides was taken for each time point. Relative percent change is calculated relative to the initial labeling value.

Condition	Final Timepoint	Relative % Change
Trypsin, pH 7, RT*	2 hr	95.2 ± 6.3
Trypsin, pH 2, RT*	24 hrs	14.9 ± 7.9
Trypsin, pH 7, 4°C*	2 hr	22.1 ± 14.3
Trypsin, pH 2, 4°C*	4 Days	17.9 ± 6.7
Beads, pH 7*	9 Days	4.2 ± 3.3
Beads pH 2*	9 Days	11.2 ± 6.4
Trypsin, Filtered pH 7ŧ	9 Days	3.6 ± 6.3
Trypsin, Filtered pH 2ŧ	9 Days	16.2 ± 3.1
Trypsin, TLCKŧ	3 Days	22.1 ± 14.1
Trypsin, heat pH 2ŧ	3 Days	16.7 ± 12.2
Trypsin, 20% methanol, pH 2	9 Days	6.8 ± 12.3

Figure 2.

Relative percent loss of the double ¹⁸O label from the fully labeled α casein digest as a function of time stored in solutions free of trypsin at room temperature after being mixed in a 1:10 ratio with a solution of H₂¹⁶O at either a pH of 2 or 7.

Investigation of the delabeling process in the presence of trypsin at a neutral pH was performed. At room temperature, loss of the ¹⁸O isotope labels was found to be very rapid, with a 78.0 ± 11.6 % loss of the double isotopic label in 10 min, when a fully labeled α-casein digest was mixed in a 1:10 ratio with a solution of H₂¹⁶O (Table 1, Fig. 3). Although de-labeling at this pH can be expected, the rapidity of loss of the isotopic label is surprising. In particular, the time for de-labeling is much shorter than the timeframe normally used for tryptic digestion in solutions with a pH near the optimum for this protease. An additional experiment with the same conditions but at a temperature of 4°C showed a 22.1 ±14.3% loss of double label after 2 h and a 74.8 ± 19.4% loss of label after 24 h (Table 1). The decrease in temperature at neutral pH thus slows down the tryptically facilitated de-labeling process, but significant isotopic loss still occurs. The rate of isotopic loss in both of these experiments is much greater than that observed in the solutions which did not contain trypsin, leading us to conclude that the enzyme is catalyzing this loss. These observations show that significant isotopic losses occur quickly after adding an ¹⁸O labeled sample to a solution of H₂¹⁶O, as would happen when an ¹⁸O labeled sample is mixed with a ¹⁶O labeled sample in H₂¹⁶O at pH 7 in the presence of trypsin. Furthermore, if this solution is allowed to stand at a neutral pH, the ¹⁶O/¹⁸O ratio in the tryptic peptides will quickly approach that of the bulk solution. The rapidity of isotopic exchange observed under these conditions can clearly lead to erroneous results and is therefore unsatisfactory for quantitative proteomic studies.

Figure 3.

(A) Region of the MALDI mass spectrum showing the peptide with a molecular mass of 1759 (HQGLPQEVLNENLLR) from the analysis of α casein digested in 95% H₂¹⁸O. (B) The same region of the MALDI mass spectrum obtained after the fully labeled α casein digest was mixed in a 1:10 ratio with a solution of H₂¹⁶O at a pH of 7 for (B) 10 minutes and (C) 2 hours.

Loss of the isotopic label at a pH of 2 in the presence of trypsin was also investigated. Acidic conditions are often used to limit tryptic activity [9,15,16,19], yet past work [28] has reported that trypsin retains activity even at low pH. The tryptically facilitated delabeling process at acidic pH was explored by diluting the fully labeled α-casein digest 1:10 in a 5% formic acid solution. At room temperature, a 14.9 ± 7.9% loss of label was observed after 24 h (Table 1 and Fig. 4). A second sample cooled to 4°C, acidified to pH 2, and stored at this temperature showed the loss of double label to be 6.1 ± 6.4% after 24 h and 17.2 ± 6.7 % after 4 d. Obviously the isotopic loss is more rapid than can be accounted for by acidic back exchange as shown in the previous experiment where loss of label was examined in the absence of trypsin. The overall sum of these results leads to the conclusion that trypsin is the main facilitator of the delabeling process, even at low pH and temperature. These results imply that some loss of ¹⁸O to the solvent can occur at pH 2, and that this loss can become significant when the samples are stored for longer than several hours even at a lowered temperature. This situation can arise when samples are deposited into a queue on an auto injector or subjected to further treatment in H₂¹⁶O based solvents.

Figure 4.

Relative percent loss of the double ¹⁸O label from the fully labeled α casein digest under conditions as a function of time stored at room temperature after being mixed in a 1:10 ratio with a solution of H₂¹⁶O at a pH of 2. In this figure, conditions were used to inactivate trypsin by lowering the pH, heat denaturation, and the addition of a trypsin inhibitor (TLCK).

Since strong cation exchange and reverse phase are currently the two primary forms of separation for proteomics, another experiment examined the delabeling process using conditions characteristic of a peptide separation by strong cation exchange chromatography. In this case, a fully labeled α-casein digest containing trypsin was combined with an equal amount of natural abundance water, imitating a 1:1 combination of two populations. The sample was immediately snap frozen for 15 sec in liquid nitrogen and lyophilized for 8 h. The sample was then reconstituted in 20% methanol, 0.1% formic acid. After 9 d at room temperature, a 6.8 ± 12.3 relative percent loss of label was calculated (Table 1). The extent of back exchange in this case is comparable to the investigation where trypsin was removed from solution and maintained at a neutral pH, indicating the presence of 20% methanol at low pH minimizes the tryptically catalyzed nucleophilic attack of water on the C-terminus of the cleaved peptide (Fig. 1, Reactions III and IV). Based on these observations, lowering the pH and adding methanol appears to be a good method to minimize isotopic loss. However, these conditions preclude subsequent procedures, such as enzymatic de-glycosylation, from being performed after isotopic labeling.

A series of experiments were performed to examine other methods for minimizing tryptic activity, thus reducing isotopic loss. The first of these evaluated was heat denaturation, as this was the simplest conceivable procedure to inactivate the enzyme. In these experiments, the fully labeled α-casein digest in H₂¹⁸O containing the trypsin used to digest/label the peptides was boiled for 10 min. Because heating was performed in 95% H₂¹⁸O, exchange of oxygen atoms on the C-terminus with water via a chemical process would not lead to loss of the isotopic label. This solution was allowed to cool, then diluted 1:10 with natural abundance water. Experiments performed at a pH near the enzymes optimal activity (pH ~7.8) showed a rapid loss of isotope (data not shown for brevity). This rate matched that seen previously for samples that were not boiled, suggesting that trypsin quickly re-natures after being boiled. Thus, simply boiling the solution is not a satisfactory method to minimize loss of the isotopic label. Heat denaturing under low pH was also examined. This was accomplished by acidifying the solution of labeled peptides in H₂¹⁸O with formic acid prior to boiling for 10 min. This solution was allowed to cool, then diluted 1:10 with natural abundance water. Once again, heating was performed in 95% H₂¹⁸O water to negate any chemical exchange which may result from the elevated temperature and reduced pH. Results from this investigation revealed that this procedure slows the rate of isotopic exchange with the solvent compared to simply lowering the pH (Table 1 and Fig. 4). The isotopic loss observed under these conditions could indicate that trypsin renatures after heating, albeit, more slowly than at a pH of 7. The similarity in isotopic loss with and without boiling suggests that this is not an effective method to reduce isotopic loss. In addition, boiling peptides under acidic conditions can lead to numerous unwanted reactions, such as the hydrolysis of aspartic acid-proline linkages, the loss of sialic acid, and deamidation of aspartic acid residues. For these reasons heat denaturation was deemed unacceptable.

Chemical inhibition of trypsin with TLCK was also examined as a means of slowing/stopping the enzymatic facilitation of isotopic exchange with the solvent. In these experiments, TLCK was added (30:1 inhibitor to enzyme ratio) to a solution of fully labeled α-casein digest in H₂¹⁸O containing the trypsin used to digest/label the peptides. This solution was allowed to stand at room temperature for 24 h, then diluted 1:10 with natural abundance water. TLCK is not stable at pHs over 7.5, therefore, the pH of this solution was lowered to a pH of 2 with concentrated formic acid prior to addition of the TLCK solution. Under these conditions, it is expected that trypsin would be significantly inhibited due to the high ratio of inhibitor: enzyme. MALDI-MS revealed that after 3 d, the relative loss of double label was 22.1 ± 14.1% (Table 1 and Fig. 4). This level of isotope loss is experimentally indistinguishable from that observed after heat denaturation at this pH (discussed above), although it is several orders of magnitude slower than that observed without trypsin denaturation. The need to maintain a low pH to reduce isotopic loss, both with TLCK and heat denaturation, prohibits additional steps, such as enzymatic de-glycosylation, from being performed after isotopic labeling. This limitation, combined with the substantial isotopic loss, lead us to conclude that TLCK treatment to reduce isotopic exchange is not a satisfactory method.

Discussion

The use of enzyme facilitated ¹⁸O labeling as a method of stable isotope incorporation for quantitative proteomics requires control of isotopic loss for accurate quantitation. The continued interaction of the cleaved peptide with trypsin, even under conditions previously thought to severely limit proteolytic activity, adds further evidence [22] that the mechanism for proteolysis (reactions I-IV in Fig. 1) and delabeling/labeling (reactions III_F+ IV_F/III_R + IV_R in Fig. 1) can occur under different conditions. In particular, the labeling/delabeling reactions appear to readily occur at a lower pH than proteolysis. Using a serine protease such as trypsin to digest and label peptides for quantitative proteomics is useful in that it can de-couple the digestion step from the labeling procedure. Unfortunately, this continued interaction with the cleaved peptide allows enzyme facilitated exchange of the isotope on the peptides with the solvent anytime the sample and enzyme are present in the same sample. Therefore, the experimental design for a quantitative proteomic scheme utilizing a ¹⁶O/¹⁸O enzyme facilitated labeling scheme to differentiate populations must take into account the capabilities for continued interaction between the cleaved peptide and the serine protease. As demonstrated here, loss of the label becomes negligible after removal of the trypsin from solution and maintaining the solution at a neutral pH prior to analysis. This study serves to alert investigators to the fact that the enzymatic mechanism of incorporating the ¹⁸O stable isotope also serves as the main process by which the stable isotope tag is removed. In conclusion, trypsin is the primary mechanism by which the ¹⁸O isotopic label is lost in quantitative proteomic studies, even under conditions of low pH and low temperature.