Identification of the Sites of Phosphorylation in Proteins Using High Performance Liquid Chromatography and Mass Spectrometry

Abstract

After incubation of cells with ³²P-labeled inorganic phosphate, it is possible to identify in vivo radiolabeled phosphoproteins. Generally, after the cells are lysed the phosphoprotein can be separated by sodium dodecyl sulfate (SDS) gel electrophoresis and a rough estimate of the size of the phosphoprotein can be gainnted. In order to determine the phosphorylated residues in the protein, the radiolabeled band can be transferred to a membrane, hydrolyzed with trypsin (or another suitable enzyme), and the two-dimensional (2D) map establishnted. The phosphopeptides observed on the 2D map can be tentatively correlated with expected tryptic fragments, based on their hydrophobicity and charge. A number of protocols have been developed to refine the correlation of the expected fragments to phosphopeptides present on the 2D map. For example, phospho-amino acid analysis of individual species present on the 2D map can be used to identify the type of phosphorylated residues present. In addition, manual Edman degradation can be performed on phosphopeptides after they are removed from the 2D map in order to identify the position of the ³²P-containing residue. This information can be used to help determine the position of the phosphorylated residue when more than one such residue is present in the tentatively assigned fragment sequence. Armed with this information, mutational analysis can confirm the site of phosphorylation, again using 2D map analysis. Although the above protocols have been successfully utilized, difficulties can arise. For example, the 2D map may not have sufficient resolution to separate two different phosphopeptides. As a result, the manual Edman analysis may not take into account the heterogeneous nature of the sample extractnted. Alternatively, ambiguous results can be obtained from manual Edmanwhen more than five cycles are required to discriminate between two possible “tentatively assigned” fragments. Partial oxidation of cysteine, methionine and tryptophan residue-containing peptides may also complicate the interpretation of the 2D map. Finally, the correlation between the pI and the hydrophobicity of the fragments may not be sufficient information to direct the mutational analysis. Many of these problems can be resolved by high-sensitivity mass spectrometry (MS) analysis. Analysis of peptide mixtures with MS will generally clarify whether a sample is heterogeneous and delineate modifications such as phosphorylation and oxidation.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

After incubation of cells with 32P-labeled inorganic phosphate, it is possible to identify in vivo radiolabeled phosphoproteins. Generally, after the cells are lysed the phosphoprotein can be separated by sodium dodecyl sulfate (SDS) gel electrophoresis and a rough estimate of the size of the phosphoprotein can be gained. In order to determine the phosphorylated residues in the protein, the radiolabeled band can be transferred to a membrane, hydrolyzed with trypsin (or another suitable enzyme), and the two-dimensional (2D) map established. The phosphopeptides observed on the 2D map can be tentatively correlated with expected tryptic fragments, based on their hydrophobicity and charge. A number of protocols have been developed to refine the correlation of the expected fragments to phosphopeptides present on the 2D map. For example, phospho-amino acid analysis of individual species present on the 2D map can be used to identify the type of phosphorylated residues present. In addition, manual Edman degradation can be performed on phosphopeptides after they are removed from the 2D map in order to identify the position of the 32P-containing residue. This information can be used to help determine the position of the phosphorylated residue when more than one such residue is present in the tentatively assigned fragment sequence. Armed with this information, mutational analysis can confirm the site of phosphorylation, again using 2D map analysis. Although the above protocols have been successfully utilized, difficulties can arise. For example, the 2D map may not have sufficient resolution to separate two different phosphopeptides. As a result, the manual Edman analysis may not take into account the heterogeneous nature of the sample extracted. Alternatively, ambiguous results can be obtained from manual Edman when more than five cycles are required to discriminate between two possible “tentatively assigned” fragments. Partial oxidation of cysteine, methionine and tryptophan residue-containing peptides may also complicate the interpretation of the 2D map. Finally, the correlation between the pI and the hydrophobicity of the fragments may not be sufficient information to direct the mutational analysis. Many of these problems can be resolved by high-sensitivity mass spectrometry (MS) analysis. Analysis of peptide mixtures with MS will generally clarify whether a sample is heterogeneous and delineate modifications such as phosphorylation and oxidation.

Several approaches for mapping phosphopeptides with MS have been proposed based on ionization techniques available in the 1980s (1,2). The introduction of matrix-assisted laser desorption (3,4) and electrospray (5,6) ionization techniques a decade later has led to substantial reductions in the amount of sample needed for MS analysis. However, the sensitivity with MS remains significantly lower than that available with radioactive labeling. As a result, it is generally not possible to use MS to analyze directly the minute amounts of phosphopeptides that can be visualized by autoradiography. MS mapping strategies for identifying the phosphopeptides generated by proteolysis of phosphoproteins generally make use of reverse phase high-performance liquid chromatography (RP-HPLC) to remove buffer salts, because their presence inhibits MS ion formation. RP-HPLC also separates the peptides based on hydrophobicity, thereby reducing the bias in the MS ionization process that may lead to inhibition of a peptide that is not readily ionized. Peptides are eluted from RP-HPLC columns using an aqueous mobile phase containing an ion pairing agent such as trifluoroacetic acid (buffer A) and an organic modifier, typically acetonitrile, containing the same ion pairing agent (buffer B). In conventional RP-HPLC, the peptides are detected by their ultraviolet (UV) absorption. Recently, Verma et al. used an alternative means of detecting phosphopeptides to identify the site of phosphorylation in S-phase Cdk inhibitor (Sic1p) (7). In this method, developed by Carr and coworkers (8–10) an electrospray triple quadrupole instrument operated in the negative ionization mode is used to analyze tryptic hydrolysis fragments, separated by RP-HPLC (11,12). MS identification of the phosphopeptide-containing fractions uses 10% of the eluent from the RP-HPLC (first pass), whereas the remainder of the stream is split for collection to be analyzed subsequently. In the first pass, fragmentation of all species exiting the RP-HPLC is induced and the instrument is set to monitor solely for the presence of a single ion, the m/z 79 fragment ion. Because this fragment ion is a specific marker for phosphopeptides, it unequivocally identifies the phosphopeptide-containing fraction. The high sensitivity available with this technique is due, in large part, to using this nonUV method of phosphopeptide detection. Because all peptides absorb in the UV, only the phosphopeptides produce this intense fragment ion. This advantage is important, as the remaining material is required for second and third dimension passes to fully characterize the phosphopeptide (7). Another advantage of Carr’s protocol is that it obviates the need for ³²P labeling. The electrospray triple quadrupole instrument applied in the above method has also been used by other investigators identifying the sites of phosphorylation in phosphoproteins (e.g., Mann and coworkers characterized IKK−2) (NF-KB inhibitory subunit kinase−2) (13). Because these accomplishments are impressive, the instruments used are expensive, and not always readily available, compared to other MS instruments.

The method described herein utilizes more generally available MS instrumentation, RP-HPLC and the classical ³²P label to identify the phosphopeptide-containing fraction(s). Typically, biological observations are made on experiments that utilize ³²P. As such, the ³²P radioactive trace offers an excellent marker to substitute for monitoring the m/z 79 fragment ion. In addition, successive steps of enzymatic hydrolysis are incorporated as a means of localizing the site of phosphorylation. In order to carry out additional stages of enzymatic hydrolysis, without requiring elaborate scale-up procedures, use is made of the significant improvements in sensitivity that have recently been made with reduced diameter columns for RP-HPLC (14,15). The products of enzymatic hydrolysis are analyzed by MALD and/or nanospray MS (16,17). In Matrix Assisted Laser Description (MALD), a UV laser is used to irradiate the sample dispersed in a matrix. The matrix is typically a small organic molecule which absorbs at the wavelength of the laser so that rapid volatilization occurs. The sample molecules trapped within the matrix are ionized and mass analyzed based on the time it takes to travel a known distance (time-of-flight). Alternatively, using nanospray ionization, a solution flows (0.02 μL/min−0.2 mL/min) through a fine capillary needle which sprays clusters made up of sample molecule ions and solute molecules. After the solute molecules are stripped away, the ions are transported into a mass spectrometer and collected in an ion trap where different mass to charge ratio (m/z) ions are ejected from the trap and detected. MALD time-of-flight and nanospray ion trap instruments are techniques which are inexpensive to commercialize and therefore have the potential to become more widespread in biochemistry /biology laboratories (18).

In the strategy presented, the labeled phosphoprotein is immunoprecipitated and separated by SDS gel electrophoresis prior to being transferred onto a membrane where enzymatic hydrolysis is performed. The primary isolation step is accomplished using RP-HPLC with a millibore diameter column. Based on the presence of the radioactive label, phosphopeptide-containing fractions are analyzed with MS to identify the general site of phosphorylation. In order to localize the site of phosphorylation further, an additional enzymatic hydrolysis step using a different specificity enzyme is often required. The fragments generated from the secondary digest are again purified with RP-HPLC, utilizing a smaller microbore diameter column, and analyzed with MS. By using MS to determine the intact mass of phosphopeptides present in the map, it is possible to obtain accurate masses that will augment the identification of phosphopeptide fragments present in the 2D map. In the strategy outlined below, this information is used in combination with phospho-amino acid analysis, manual Edman protocols and the peptide pI to interpret 2D maps.

2 Materials

2.1 Extraction and Millibor e RP-HPLC Equipment (see Note 1).

• 1. HP1090 (Hewlett Packard, Palo Alto, CA) 200 μL injection loop, 200 μL/min flow rate, 2.1 × 150 mm C₁₈ Vydac (5 μ 300 Å particle size) column.

• 2. Ultrasonic bath, FS3 (Fischer Scientific).

• 3. Filter pipet tips (Molecular Bio-Products, San Diego, CA).

• 4. Microcentrifuge tubes (1.5 and 0.5 mL) (Sarstedt, Germany).

• 4. Trifluroacetic acid (TFA) (redistilled in house).

• 5. A-buffer : 0.055% TFA in DI H2O (Millipore).

• 6. B-buffer: 0.05% TFA in acetonitrile:DI H₂O (9:1).

• 7. Extraction buffer: formic acid, acetonitrile, isopropanol, and DI H₂O mixture (1:1:1:1).

2.2 Phosphoprotein (Primary) Proteolysis

• 1. Membrane pretreatment buffer: 0.5% PVP360 (Sigma Chemical Co., St Louis, MO) in 100 mM acetic acid.

• 2. Trypsin stock: 25 μg sequencing grade trypsin (Boehringer Mannheim, Germany, cat. no. 1418475) in 50 μL 0.055% TFA.

• 3. Lys-C stock: 5 μg Asp-N (Boehringer Mannheim, Germany, cat. no. 84212321) in 50 μL DI H2O.

• 4. Trypsin and Lys-C hydrolysis buffer: pH 8.5, 50 mM NH₄HCO₃.

• 5. 10%aqTFA.

• 6. 100 mM sodium acetate.

• 7. 100 mM TCEP (Diagnostic Chemicals Ltd., Oxford, CT) in 50 mM sodium acetate.

2.3 Mass Spectrometry Equipment (see Note 2)

• 1. Reflex MALD time-of-flight instrument (Bruker Daltonics, Billerica, MA) operated at +31 kV accelerating voltage and +30 kV reflector voltage, 100 MHz digitizer, mass accuracy better than 500 ppm.

• 2. Esquire nanospray quadrupole ion trap instrument (Bruker Daltonics, Billerica, MA) operated at 800 V, mass accuracy better than 200 ppm.

• 3. Nanospray capillaries (Bruker Daltonics, Billerica, MA).

• 4. Electrospray tuning/calibration mix (G2421, Hewlett Packard, Palo Alto, CA).

• 5. Adjustable volume (0.5–10 μL) pipet (Eppendorf varipette 4710, Germany) and pipet tips (Eppendorf geloader 0030 001.222, Germany).

• 6. UV absorbing matrix: Nitrocellulose (1 mg /mL) (Schleicher and Schuell, Germany) is added to a saturated solution of α-cyano−4-hydroxy cinnamic acid (Aldrich Chemical Co., Milwaukee, WI) in isopropanol and acetone (1:1).

• 7. 1% formic acid in methanol:DI H₂O (1:1).

2.4 Phosphopeptide (Secondary) Proteolysis and Microbore RP-HPLC

• 1. Michrom Bioresources UMB100 (Michrom Bioresources, Auburn, CA) 100 μL injection loop, 30 μL/min flow rate, 0.5 × 150 mm C₁₈ Vydac (5 μ 300 Å particle size) column.

• 2. Vacuum centrifuge (Savant Speed-Vac, Farmingdale, NY).

• 3. Asp-N stock: 2 μg Asp-N (Boehringer Mannheim, Germany, cat. no. 1054589) in 50 μL DIH₂O.

• 4. Chymotrypsin stock solution: 25 μg chymotrypsin (Boehringer Mannheim, Germany, cat. no. 1418467) in 50 μL DI H₂O.

• 5. Asp-N hydrolysis buffer: 100 mM Na₂HPO₄, pH 7.5.

• 6. Chymotrypsin hydrolysis buffer: 50 mM Et₃NH₄HCO₃ (Aldrich Chemical Co., Milwaukee Wisconsin), pH 8.2.

• 7. 10%aqTFA.

3 Methods

3.1 Determining an Appropriate HPLC Gradient and Identifying the Retention Time of the Phosphopeptide of Interest

In this protocol we rely on detection of the radioactively labeled phosphopeptide eluting from RP-HPLC. The first step, while the background counts on the HPLC are low, is to determine an appropriate gradient for analysis of the phosphopeptide of interest. This can be accomplished by carrying out 2D map analysis of a tryptic (or other compatible) enzyme digest (see Note 3 ) as described in detail in this volume, and elsewhere (19). From the 2D map between 500 and 1000 cpm of the cellulose powder containing a phosphopeptide is collected, the peptide is extracted from the cellulose and the sample applied to RP-HPLC. The retention time of the phosphopeptide eluting fraction is thereby determined.

• 1. Phosphopeptides eluted on a 2D map are localized by autoradiography and removed from the plate and collected (the cellulose is loosened from the plate by scraping with a clean spatula and collected in a filter-containing pipet tip attached to a vacuum system). The pipet tip (including filter and cellulose) is placed in a microcentrifuge tube and centrifuged to transfer the cellulose powder to the tube.

• 2. The extraction buffer (30–40μL) is added to the cellulose powder containing the phosphopeptide and the microfuge tube is floated in the sonication bath and sonicated for (30–60 min).

• 3. The decant is removed and counted. Between 500 and 1000 counts are sufficient to carry out steps 4 and 5. The decant is diluted (1:5) with A-buffer prior to step 4.

• 4. The sample is injected onto RP-HPLC with a standard gradient (isochratic at 100% A for 5 min followed by a 40-min gradient from 0% B to 80% B). The eluent from the column, while under isochratic conditions, is collected in one 1.5-mL microfuge tube; followed by gradient fractions collected at 30-s intervals in 0.5-mL microfuge tubes (for the first 20–25 min of the gradient) and then at 5-min intervals in 1.5 mL microfuge tubes (over the remainder of the gradient and the column wash) (see Note 4 ).

• 5. The retention time of the phosphopeptide is determined by counting the fractions (see Note 5 ).

• 6. The dilution of the extracted peptide (step 3), or the collection times, may need to be adjusted and the procedure repeated with a phosphopeptide from a fresh 2D map.

3.1.1 Isolation and Hydrolysis of Phosphoprotein SDS gel electrophoresis (see Note 6) is used to isolate the ³²P labeled phosphoprotein of interest (a sample containing approx 0.2–2.0 μg of a 50 kDa phosphoprotein and 100,000 cpm ³²P labeled protein is required, see Note 7 ), the phosphoprotein is transferred to a PVDF membrane as described elsewhere (20) and hydrolyzed with trypsin or another suitable enzyme.

• 1. The PVDF membrane is exposed to X-ray film and the region containing the labeled protein excised using a clean disposable razor blade. Other regions are excised (see Note 8 ) as control samples. While keeping the membrane pieces moist with DI H₂O, use the razor blade to cut the excised membrane into 1 mm² or smaller squares. Using clean pinceps place these membrane pieces into a 1.5mL microfuge tube.

• 2. Add 1 mL of the pretreatment buffer to the membrane pieces, shake the microfuge tube for 15 min at 32°C, remove the buffer, and rinse three times with 1 mL DI H₂O.

• 3. After removing the final DI H₂O rinse, add the enzyme buffer (100 μL) and measure the total cpm in the membrane pieces. An aliquot of the trypsin stock solution (3 μL equivalent to 1.5 μg) (or Lys-C, 0.5 μg) is then added every 8 h to the microfuge tube incubated at 37°C (see Note 9 ). The same amount of buffer and enzyme are added to the control samples at appropriate intervals.

• 4. The release of phosphopeptides is checked by removing the membrane pieces and counting the hydrolysis buffer after 16 and 24 h incubation (see Note 10 ).

• 5. When the digestion is complete (see Note 11 ), the membrane pieces are removed.

• 6. A small aliquot of the protein hydrolysate (corresponding to approx 100–500 cpm) is removed and a 2D map analyzed to verify that the map is comparable with that obtained previously.

• 7. The remainder of the hydrolysis solution is acidified by adding sufficient volume (typically 10–20 μL) of 10% TFA to adjust the pH of the hydrolysis buffer solution to approx pH 4 and then sufficient (typically 50–80 μL) 100 mMNaOAc to adjust to pH 5.0 (see Note 12 ).

• 8. Reduction of the protein is carried out by adding 10 μL of 100 mM TCEP in 50 mM NaOAc and incubating at 25 °C for 60 min.

3.2 Millibore RP-HPLC Isolation of Phosphopeptide(s)

The resulting hydrolysis mixture containing the phosphopeptide(s) of interest are separated by RP-HPLC using the predetermined conditions. Fractions collected are prepared for MS and 2D map analysis.

• 1. The hydrolyzed phosphoprotein sample is analyzed with RP-HPLC using gradient conditions and manually collecting fractions with the same time intervals as determined in Subheading 3.1 . (see Note 13 and Fig. 1 ). The fractions collected are stored directly on ice.

• 2. Based on the RP-HPLC retention time of the phosphopeptide (as determined in Subheading 3.1. ), the fraction in which the phosphopeptide of interest elutes is identified (see Fig. 2 ).

• 3. Using a hand held counter it is quickly verified that this fraction contains the phosphate label (see Note 14 ).

• 4. After the phosphopeptides of interest have eluted (but not necessarily before the gradient and collection are finished) a 1−μL aliquot of the samples of interest (representing 1% of the total sample) is transferred onto a preprepared MALD target (see Notes 15 and 16).

• 5. In order to confirm that the phosphopeptide isolated corresponds with the species identified in the 2D map, the RP-HPLC fraction from the primary digest and the phosphopeptide mixture are analyzed alone on separate 2D maps, and mixed together on a third 2D map. This mixture represents a coelution experiment and is used to verify that the intensity of the spot of interest is increased and that no new spots appear (see Note 17 ).

Fig. 1.

The UV trace observed from milli-bore RP-HPLC using the standard gradient (see 2.1) of (A) enzyme blank and (B) approximately 0.5 μg of a phosphoprotein. Some UV absorption is observed (Figure 2 shows the ³²P label observed for these fractions).

Fig. 2.

Comparison of the ³²P label observed from (?) a phosphopeptide removed from a 2D tryptic map containing 1000 counts (cpm × 10) and (?) and a phosphoprotein hydrolyzed with trypsin. The retention time of the extracted phosphopeptide (11.0 minutes) was used to direct the investigation to the equivalent retention time fraction from the phosphoprotein digest. ST is the empty sample tube and the reading before that is the injector rinse.

3.3 MS Analysis and Interpretation

The phosphopeptide(s) of interest, and both earlier and later eluting fractions, are analyzed with MALD and subsequently with nanospray MS.

• 1. All samples prepared in step 6 are analyzed with MALD (see Note 18 and Fig. to identify other closely eluting peptide fragments (see Note 19 ).

• 2. Tentative assignments are then made between species observed in the MALD mass spectra and expected hydrolysis fragments of the protein generated by a suitable computer program (see Note 20 ). Allow a suitably wide tolerance range of masses (approximately ^+- twice the expected accuracy of mass measurement) to ensure that all possible candidates are considered.

• 3. It is important to determine whether there are any other possible ambiguous assignments conflicting with the tentative assignments made in step 2. The observed species should be compared with singly phosphorylated and singly oxidized fragments (this is easily done by subtracting either 80 or 16 Da from the observed species and checking against the computer generated nonphosphorylated fragments). Multiply phosphorylated or oxidized fragments (subtracting multiples of either 80 or 16 Da from the observed masses) must also be considered. Searches should also be carried out to identify fragments corresponding with combinations of these modifications. In addition, for tryptic digests, compare the expected trypsin autolysis fragments with the species observed in the MALD mass spectra (see Note 21 ). Based on these calculations some of the tentative assignments made in step 2 may be considered ambiguous (e.g., an observed species may be consistent with either a singly phosphorylated fragment and a trypsin autolysis fragment).

• 4. Based on the MALD results, the expected mass and approximate charge based on the tentative assignments (charge = 1 for free amino terminus + 1 for every arginine or lysine residue in the proposed sequence) of the fragments can be calculated and the electrospray MS instrument calibrated (using electrospray tuning/calibration mix) over an appropriate mass range. Next, the instrument focusing can be optimized for detection of a control peptide having similar mass and charge compared to that of the proposed species.

• 5. A 1−μL aliquot of the sample (representing 1% of the total sample) is transferred to a nanospray capillary, using an adjustable volume pipet with geloader tips, for Nanospray MS analysis (see Note 22 ).

• 6. Nanospray MS analysis of fractions (see Note 23 ), particularly after tentative assignments have been made, allow the presence of a particular mass and charge species to be verified/identified (see Fig. 4 ).

• 7. The information gained in steps 1–6 and Subheading 3.3. , step 5 is used to supplement the predicted positions of phosphopeptides on the 2D map. For example, based on size, charge, pI and phospho-amino acid analysis certain fragments may be predicted in regions of the 2D map. The successful MS analysis of one or more of these phosphopeptides will help allow reinterpretation of the expected positions of other fragments in the map. Even the analysis of nonphosphorylated peptides, in phosphopeptide containing fractions, gives additional information. The observed species may be the nonphosphorylated form of the phosphopeptide or it may have a very similar retention time to the phosphopeptide. If the latter is the case, then a computer search of expected phosphoprotein hydrolysis fragments with similar retention time to the observed species will include the phosphopeptide and this information will help direct an assignment. All the information gained can be used in an iterative fashion to help refine other assignments in different regions of the 2D map.

• 8. Resolution of remaining ambiguities may require further information. Approximately 2% of the phosphopeptide sample is consumed for MS analysis (in steps 1 and 6) whereas a further 5–10% of the sample is reserved for 2D map analysis (see Subheading 3.5 ). Rather than attempting to preconcentrate the remaining phosphopeptide solution (reducing the volume typically results in large sample losses) the remaining sample is used to perform a secondary digest (see Note 24 ).

Fig. 3.

The MALD mass spectrum of the ³²P containing fraction with retention time of 19 min shown in Fig. 2. In the RP-HPLC fractions recovered from the primary digest it is not unusual to observe closely eluting nonphosphorylated peptides as indicated.

Fig. 4.

Nanospray mass spectrum of a ³²P containing fraction recovered from a secondary digest (approximately 1.0mg of phosphoprotein was hydrolyzed with trypsin (primary proteolysis), purified with milli-bore RP-HPLC and then hydrolyzed with Asp-N (secondary proteolysis) and purified with milli bore RP-HPLC), in which M, Mp and M′ species were clearly observed. It is unusual to observe a peptide (indicated as M′) which is not the non-phosphorylated form of a phosphopeptide closely eluting with the ³²P label in the purification after the second proteolysis, but it does occur! M and M′ were generated from the same precursor phosphopeptide isolated after the primary proteolysis step but not separated on micro bore RPHPLC because both M and M′ have the same theoretical HPLC retention time. (e.g., an observed species may be consistent with either a singly phosphorylated fragment and a trypsin autolysis fragment).

3.4 Microbore RP-HPLC Isolation of Secondary Enzyme Hydrolysis Phosphopeptide(s)

For reasons identified in Subheading 3.4 ., a secondary enzyme hydrolysis (with differing enzyme specificity compared to the first hydrolysis) is performed to conclusively identify the phosphopeptide. It is also advantageous to carry out a secondary enzyme hydrolysis when an intact nonphosphorylated fragment is observed from the ³²P containing fraction, but not the corresponding phosphopeptide fragment (see Note 25 ). By reducing the size of the phosphopeptide, while retaining the approximate concentration of the phosphopeptide in the solution, the intact phosphopeptide ion is often observed. In order to keep the phosphopeptide concentration roughly comparable, the secondary digest is purified with microbore RP-HPLC at a lower flow rate while retaining the same time interval for collection. The choice of enzyme depends on the alternative assignments that need to be distinguished. For example, if one or more assignments involve sequences containing aspartic acid or glutamic acid residues (see Note 26 ), then an Asp-N digest may be an appropriate means of confirming an assignment. Other factors to consider are the size of the fragments that would be produced (see Note 27 ) and whether an alternative enzyme would better localize the site of phosphorylation (see Note 28 ). Chymotrypsin is one such alternative enzyme which has a complementary, if somewhat broader, specificity compared to either trypsin or Asp-N. The resulting hydrolysis of the phosphopeptide(s) of interest, after separation by microbore RP-HPLC, is prepared for MS analysis.

• 1. The HPLC purified fraction(s) isolated in Subheading 3.3 . is reduced in volume to 5–10 μL by drying in a vacuum centrifuge. An earlier eluting fraction with low counts is also reduced in volume for use as an enzyme blank, and an appropriate control peptide is prepared in a separate microfuge tube as a positive control.

• 2. An aliquot (100 μL) of the enzyme buffer solution is added to the fraction of interest and the control samples.

• 3. Asp-N (0.2 μg) or alternatively, chymotrypsin (2.0 μg) are added and the solutions incubated at 37°C.

• 4. The enzyme hydrolysis of the positive control peptide is stopped after 16 h by addition of 10 μL of 10% TFA. RP-HPLC analysis and MALD MS analysis are used to verify enzyme hydrolysis of the control peptide.

• 5. Provided satisfactory hydrolysis is observed in step 4, the secondary enzyme hydrolysis of the phosphopeptide and the enzyme blank are stopped by addition of 10 μL of 10% TFA.

• 6. The hydrolyzed phosphoprotein sample is analyzed with microbore RP-HPLC using a standard gradient (isochratic at 100% A for 5 min followed by a 20-min gradient from 0% B to 60% B; see Note 29 ). The eluent from the column while under isochratic conditions is collected in one 0.5-mL microfuge tube followed by separate gradient fractions collected at 30-s intervals. The phosphopeptidecontaining fractions are identified as in Subheading 3.2 . A 1 μL aliquot of this sample (representing 6% of the total sample) is immediately transferred onto a preprepared MALD target(s) prepared as in Subheading 3.3 .

• 7. Repeat steps in Subheading 3.4 ., as applicable.

3.5 Further Interpretation

Usually, the second enzyme hydrolysis will have resolved ambiguities and a single site of phosphorylation is present. For example, a ³²P-containing fraction may again be isolated in which the MS species observed is assigned as a phosphorylated peptide fragment. This fragment is consistent with hydrolysis of a fragment observed after the first hydrolysis, and the specificity of the enzyme used in the second hydrolysis step. It is worthwhile to compare the results obtained with other information, such as the fragment pI, the phosphoamino acid analysis or the manual Edman analysis. Provided these are all consistent, the site of phosphorylation can be considered identified.

However, a second hydrolysis step may not resolve all ambiguities. Consider that a nonphosphorylated peptide fragment that meets t he above criteria (i.e., that it is consistent with further hydrolysis and the specificity of the second enzyme) is observed from the ³²P-containing fraction. The method presented is designed, in part, to address this situation. If the level of incorporation of phosphate is sufficiently low (see Note 30 ), in conjunction with the ionization bias (see Note 25 ), then we may isolate and observe a nonphosphorylated peptide after both the first and second hydrolysis steps ( Subheadings 3.3 . and 3.5.). Because the millibore RP-HPLC step ( Subheadings 3.3 ) is localizing the ³²P label, we deduce that the nonphosphorylated species observed is closely eluting with the phosphopeptide (which is not observed). After the second proteolysis step ( Subheadings 3.5 ), if the fragment observed from the ³²P-containing fraction corresponds with proteolysis of this first hydrolysis fragment, then we have a critical piece of information. Because the second proteolysis step is again localizing the 32P label, and the second proteolysis fragment is generated from the first hydrolysis fragment, it is proposed that the observed fragments are the nonphosphorylated forms of the phosphopeptide fragments. The site of phosphorylation is thereby localized to the sequence of the second hydrolysis fragment.

• 1. The mass observed from analysis of the ³²P label-containing second hydrolysis fraction should be compared with expected hydrolysis fragments from the species observed in the primary hydrolysis fraction (see Note 20 ). The question asked is “does the observed second hydrolysis species correspond ONLY to an observed and expected fragment formed from a primary hydrolysis fragment ?”

• 2. If the answer is “yes,” then go to step 5.

• 3. If the answer is no, then all the possibilities cannot be detailed herein, and even if they were, the interpretation may still not be clear cut. Therefore, check for ambiguities, i.e., compare the observed species with the intact phosphoprotein sequence and trypsin autolysis fragments (as in Subheading 3.4. , step 3) since it is possible that coeluting impurities were present in the first hydrolysis fraction. The question asked is ”is it plausible that the observed species is a hydrolysis fragment of a species coeluting with the phosphopeptide in the millibore RP-HPLC, which coincidentally coelutes with the phosphopeptide fragment on the microbore RP HPLC ?“

• 4. 2D map analysis of the phosphopeptide fractions obtained from RP-HPLC purification after the first and second hydrolysis reactions (each alone and mixed together) can be used to confirm that the second proteolysis reaction cleaved the phosphopeptide and help answer the question posed in step 3.

• 5. Next, ask if the secondary hydrolysis fragment contains an amino acid consistent with the phosphoamino acid analysis and whether the results are consistent with the manual Edman analysis (carried out on either the primary or secondary hydrolysis fragment).

• 6. Test your assignment wherever possible. For example, by changing the primary enzyme from trypsin to Lys-C you should change the 2D map position (and RP-HPLC retention time) of a fragment which incorporates an arginine residue at either the N- or C- terminal cleavage site. If the fragment you observe is the nonphosphorylated form of the phosphopeptide then the change of primary enzyme should be reflected in the mass observed. Alternatively, if a tentative assignment contains a residue which may be oxidized, then confirming the presence of an oxidized residue in the isolated phosphoprotein (by analyzing the 2D map of the fraction with and without performic acid treatment) may allow this assignment to be confirmed.

4 Notes

• 1. It is necessary to have access to a millibore HPLC, such as that described in Subheading 2. , for Subheadings 3.1. and 3.2 and 3.2. (to obtain reproducible retention times for comparisons). In order to carry out a secondary enzyme digest and localize the site of phosphorylation (see Subheading 3.3 ), without requiring scale up of the starting material, it is necessary to have access to a microbore HPLC such as that described in the Methods section.

• 2. It is not necessary to have access to both MS instruments described in Subheading 3. . to perform phosphopeptide mapping successfully. In Subheading 3.4. , omitting the nanospray MS analysis (steps 5 and 6) reduces the chances of identifying the phosphopeptide and thereby increases the reliance on the deductive arguments discussed in Subheading 3.6.

• 3. Lys-C can be a useful alternative to trypsin when trying to distinguish between different possible assignments (see Subheading 3.6. ,step 6). The buffer concentration and volatility are important constraints in this protocol.

• 4. Always condition the column and allow it to equilibrate. Avoid elution conditions where a peptide from one run may be eluted in the following run (e.g., if a higher percentage of organic modifier is used). HPLC initialization protocol: an Abuffer injection is measured first to verify that the chromatogram is blank, followed by an enzyme blank, another Abuffer blank and the phosphoprotein/phosphopeptide hydrolysate. Record and mark the UV trace to indicate fraction number, when the fraction cut started (even though there is no UV absorbance) and time after injection start at suitable intervals.

• 5. If most of the radioactivity eluted in the column crash, then the peptide is hydrophilic. It can typically be dried down to 0 μL and taken up in 100% Abuffer and reinjected onto the column with minimal losses (check this by counting the empty tube afterwards). If it again crashes off the column, then a different ion pairing agent (e.g., hexaflurobutyric acid) can be used, or an alternative column designed for hydrophilic peptides (e.g., Alltech Lichrosorb Select B 5U) may be used. The collection time can be adjusted but it should remain as narrow (30 s) as possible between 0–40% B or 0–50% B before it is widened out (above 40% B or as determined in Subheading 3.1. ) to collect the column wash.

• 6. It may be advantageous to determine the minimal SDS gel electrophoresis time that will separate the phosphoprotein of interest from other immunoprecipitated proteins. Longer duration SDS gel electrophoresis will increase the area of the phosphoprotein in the gel, and therefore, the area of the phosphoprotein on the membrane. The physical size of the membrane determines the volume of the enzymatic digestion buffer that should be kept 100 μL.

• 7. The success in measuring the intact phosphopeptide is also dependent on the degree of phosphorylation. However, with this method it may be possible to identify the site of phosphorylation even when the phosphopeptide is not directly detected (as explained in Note 28 and Subheading 3.6. ).

• 8. Based on autoradiogram exposure of the membrane, excise the phosphoproteincontaining portion of the membrane. From the same membrane, excise a similar size area of membrane either above or below phosphoprotein being careful to avoid the phosphoprotein region (and the antibody light or heavy chain regions). A second control region can be excised from the membrane from an area which is not in contact with the gel.

• 9. To minimize the contribution of enzyme autolysis fragments in the HPLC chromatograms and in the MS, the enzyme stock is kept at 4°C for a maximum of 12 h (or a fresh stock is used). In addition, enzyme to substrate conditions that maximize hydrolysis of the phosphoprotein (as indicated by elution of ³²P from the membrane), and minimize interference of the enzyme in the subsequent chromatographic isolation and MS analysis, are employed.

• 10. Intermittent shaking and gentle centrifugation of the membrane pieces is carried out to ensure that the membrane pieces are dispersed and completely immersed in the enzyme buffer solution.

• 11. If significant counts remain in the membrane (>50% of the total number of counts prior to hydrolysis) return the membranes to the buffer and add the third aliquot of enzyme.

• 12. The actual volumes required to adjust the pH should be determined first using the enzyme blank. The same volumes are then employed for the phosphoprotein.

• 13. Ideally, the volume to be injected should be less than that which can be loaded in a single injection (and the same as that used in Subheading 3.1. . Multiple sample injections may increase the loss of hydrophilic peptides. Fractions are collected while marking the UV trace to indicate fraction number, when the fraction cut started (irrespective of the UV absorbance) and time after injection start at suitable intervals. These records will help when comparing these results to those obtained in Subheading 3.1.

• 14. In addition to rapid scanning with a handheld counter, using a scintillation counter accurately determines the number of cpm in all fractions and relevant material. The total cpm for the eluent of the HPLC column (i.e., including the column crash and the collected fractions) plus the total cpm lost (i.e., counts left in the sample microfuge tube, syringe, and injector wash) should approximate to the cpm of the hydrolyzed phosphoprotein sample. Counting all of these microfuge tubes, or the closest equivalent (e.g., rinse the syringe and count the rinse), allows minimization of losses and optimization of the recovery of the phosphopeptide. Less than 10% of the hydrolyzed phosphoprotein counts should be lost. You can also evaluate the benefit of rinsing the membrane vs counts lost (in the column crash) as the total injection volume increases.

• 15. MALD targets are first cleaned by sonicating in 10% HNO₃ for 15 min and then in DI H₂O (15 min). The targets are removed from the H₂O, rinsed with methanol and dried. The target is pretreated with a (0.5 μL) aliquot of the UV absorbing matrix.

• 16. Take care that the matrix is not disturbed (with the pipet tip) when applying the sample aliquot. If the phosphopeptide is sufficiently hydrophobic that the solution contains more than approx 25% acetonitrile, drying of the sample can be facilitated by aiming a stream of nitrogen gas at the solution on the matrix.

• 17. When a number of consecutive HPLC fractions have significant levels of ³²P, then presumably the phosphopeptides have similar retention times under the RP-HPLC conditions employed. It is essential to carry out the 2D map analysis to verify that the fraction of interest contains the appropriate 2D map phosphopeptide and also determine the purity of this fraction. Even when a single or isolated ³²P label fraction is recovered, it is important to confirm the homogeneity of this fraction.

• 18. The MALD target is inserted into the MS instrument vacuum chamber and analyzed, then withdrawn from the instrument and rinsed by adding 10 μL of DI H₂O to the matrix surface and immediately blowing the solution off the matrix surface with a stream of N₂ gas. Extensive (50–200 independent laser shots) analysis of the sample from different positions on the matrix surface are carried out both prior to, and after, rinsing of the sample target.

• 19. By analyzing pre- and posteluting fractions, closely eluting peptides can be distinguished from coeluting peptides. If a peptide is observed in a preeluting fraction, then it is presumed that its presence in the later fraction is because of “tailing” of the species as it elutes off the column. If it is observed in a posteluting fraction with equal or greater intensity, it is presumed that the fraction was cut while the peptide was eluting.

• 20. A number of computer programs can be used to generate the calculated masses of hydrolysis fragments from a known protein, e.g., MacBioSpec (21) (available through PESciex Internet address: http://www2.perkin-elmer.com/sc/index.htm) or MS-Digest (available from UCSF http://prospector.ucsf.edu/htmlucsf/msdigest.htm) or GPMAW (http://130.225.147.138/gpmaw/default.htm).

• 21. Checking whether any species observed could be owing to autolysis fragments can be done with programs listed in Note 20, and the appropriate enzyme sequence. For example, loading the trypsin sequence and setting trypsin as the enzyme will generate trypsin autolysis fragments. In order to check for alternative specificity cleavage sites, it is often preferable to determine whether the observed mass species would correspond to any peptide fragment generated from the precursor and then subsequently determine if any of these fragments make “sense” on a chemical proteolysis basis. This is accomplished with a search based on “mass.”

• 22. Prior to transfer of the sample for analysis, the nanospray capillary is rinsed with isopropanol, and then dried with a stream of N₂ gas. To check that a valuable sample is not loaded into a damaged or contaminated capillary, a 1-μL aliquot of 1% formic acid in methanol is loaded into the capillary using the geloader tips. The sample is inserted into the MS instrument and positioned. Once inserted into the source housing, the capillary is examined with a 25× microscope and, if necessary, the aperture enlarged by carefully touching the capillary against the end plate. The mass spectrum is measured in both positive and negative ionization modes (which will serve as background spectra), the position of the capillary is optimized and the flow rate checked.

• 23. An aliquot (1 μL) of the HPLC purified sample is inserted into the capillary using the geloader tips. The capillary is inserted into the source housing, taking care not to damage or contaminate the capillary. Final repositioning of the capillary may be necessary. Both positive and negative ionization mode spectra are measured.

• 24. No discussion of PSD or MSⁿ capabilities available with certain MALD time-of-flight or nanospray MS instruments is included. The ability to discriminate between more than one possible assignment, or further localize the site of phosphorylation with PSD or MSⁿ techniques has been well documented but is considered “icing on the cake.” The method presented here is applicable for researchers who have access to MALD time-of-flight, but not necessarily to PSD or alternatively, access to a nanospray single quadrupole instrument (i.e., without MSn).

• 25. An ionization bias is often observed in MS where the relative intensity of a mass in the mass spectrum does not reflect the overall relative abundance of this species. This ionization bias is not only observed for different peptides (with differing sequences), but also peptides of the same sequence where one form of the peptide is modified (e.g., peptide and phosphopeptide mixtures). For example, a mixture of a peptide and the corresponding phosphopeptide (9:1) may not result in a detectable signal for the phosphopeptide (22,x23

• 26. Using V8 S. aureas or Asp-N at high enzyme to substrate conditions will hydrolyze peptide bonds at the carboxy or amino-terminal side of glutamic acid residues, respectively.

• 27. Low-mass fragments are often difficult to detect in MALD time-of-flight mass spectra because of the presence of matrix ions.

• 28. Selection of an appropriate enzyme may be made on the basis of the peptide observed in the fraction, irrespective of whether a phosphopeptide is observed. A low level of phosphorylation, or the ionization bias (see Note 25 ), may prohibit observation of the phosphopeptide.

• 29. As illustrated in Subheading 3.1. , for the millibore RP-HPLC system the retention time of the phosphopeptide can be determined on the microbore RP-HPLC prior to injection of the actual sample. Although this strategy works very well with the millibore RP-HPLC, it is less reliable due to the reduced reproducibility of the retention times with microbore systems and is mentioned here only as an option to be considered.

• 30. An underlying axiom when using MS to identify phosphopeptides is that MS requires a high level of incorporation of phosphate at a particular site to identify that particular phosphopeptide. The level of incorporation of phosphate at individual sites is not readily revealed by autoradiography after SDS gel electrophoresis of a phosphoprotein. If the phosphoprotein contains multiple sites of phosphorylation many of the protein molecules may contain phosphate, but an isolated phosphopeptide may be present together with a significant excess of the corresponding nonphosphorylated peptide. For example, a protein which has ten or more sites of phosphorylation and a stoichiometry of incorporation of phosphate at each site of 30% will, on average, generate at least one phosphate (and therefore a label) on every molecule. However, MS analysis of one of the phosphopeptides will be attempting to detect a phosphopeptide in the presence of an excess amount of the corresponding nonphosphorylated peptide.