Straightforward and de Novo Peptide Sequencing by MALDI-MS/MS Using a Lys-N Metalloendopeptidase^,

Abstract

In this work, we explore the potential of the metalloendopeptidase Lys-N for MALDI-MS/MS proteomics applications. Initially we digested a HEK293 cellular lysate with Lys-N and, for comparison, in parallel with the protease Lys-C. The resulting peptides were separated by strong cation exchange to enrich and isolate peptides containing a single N-terminal lysine. MALDI-MS/MS analysis of these peptides yielded CID spectra with clear and often complete sequence ladders of b-ions. To test the applicability for de novo sequencing we next separated an ostrich muscle tissue protein lysate by one-dimensional SDS-PAGE. A protein band at 42 kDa was in-gel digested with Lys-N. Relatively straightforward sequencing resulted in the de novo identification of the two ostrich proteins creatine kinase and actin. We therefore conclude that this method that combines Lys-N, strong cation exchange enrichment, and MALDI-MS/MS analysis provides a valuable alternative proteomics strategy.

In proteomics, peptide sequencing is mainly performed by CID-based tandem mass spectrometry (1,2). Generated peptide fragmentation spectra are matched against in silico-derived spectra from amino acid sequences in proteomics and genomics databases. Trypsin is the most frequently utilized protease as it generates peptides in the preferred mass range for effective fragmentation by CID (3). Trypsin has high cleavage specificity and is stable under a wide variety of conditions generating peptides with a C-terminal arginine or lysine. This C-terminal positioning of the basic residue has consequences for fragment ion formation in CID. According to the “mobile proton” model, dissociation upon excitation is initiated by a proton that weakens an amide bond in the peptide backbone (4–7). The proton affinity/gas phase basicity of the two conjugate fragments will then dictate which fragment will inherit the amide-breaking proton, leading to the formation of b- or y-ions, respectively (8). In MALDI-MS/MS of singly charged tryptic peptides, fragmentation results in complex spectra containing not only b- and y-ions but also some a- and immonium ions, internal fragments and ions resulting from neutral loss of ammonia or water (9,10). Although all these fragment ions are used in typical database search strategies they often complicate and hamper de novo sequencing, e.g. sequencing of peptides from species of which no genome sequence is available (11,12). Therefore, several attempts have been made to simplify MALDI-CID spectra (11, 13–19). For example, the peptide N terminus can be derivatized with sulfonic acid as in chemically assisted fragmentation (CAF)1 (16,17) or with 4-sulfophenyl isothiocyanate (SPITC) (20,21) to establish a fixed negative charge. After this reaction, primarily protonated C-terminal fragment ions are detected, giving rise to y-ion ladder series. The loss of sensitivity caused by adding a negative charge while performing positive ion mode analysis can, although only in part, be compensated for by increasing the basicity of lysine residues (22). Another approach to simplify MALDI spectra is to add a fixed positively charged tag to the peptide N terminus (13) while modifying internal arginine residues (14) or removing the C-terminal lysine or arginine (11). These modified peptides fragment to generate spectra with mainly a- and b-ions. One can also modify the basicity of the peptide to promote formation of a single series of ions. Lysine can be made more basic by guanidation (18) or treatment with 2-methoxy-4,5-dihydro-1H-imidazole (19). In this way, the C-terminal fragment of a tryptic peptide is more likely to be protonated after fragmentation yielding spectra with more intense y-ions. Such techniques, however, require additional sample handling. Moreover chemical derivatization of minute amounts of sample is more difficult and is often hampered by the formation of unwanted side products (23).

Recently we explored a new method for mass spectrometry-based sequencing of peptides using a little explored metalloendopeptidase with Lys-N cleavage specificity (24). We showed that the combination of this protease with ESI-MS using electron transfer-induced dissociation (ETD) for peptide fragmentation produced spectra that were completely dominated by c-type fragment ions, providing simple sequence ladders of the peptides of interest (24). In ETD with supplemental collisional activation (commonly referred to as ETcaD), doubly charged peptide ions generated by ESI are charge-reduced during the electron transfer process, resulting in the remainder of a single free proton (25). As the N terminus of Lys-N peptides accommodates two basic entities, primarily N-terminal fragments are protonated, therefore leading to the detection of mainly c-ions. However, ETD fragmentation requires multiply charged ions (i.e. ESI) and specific instrumentation that is not readily available. Therefore, in this study, we comprehensively explored the use of the Lys-N metalloendopeptidase and CID fragmentation using MALDI-MS/MS. A HEK293 cellular lysate was digested by Lys-N and for a direct comparison also with Lys-C, which produces tryptic-like peptides with the basic lysine at the C terminus. The resulting peptides were separated and enriched for peptides containing a single lysine residue by strong cation exchange (SCX) and analyzed by MALDI-TOF/TOF. The combination of Lys-N and MALDI-MS/MS resulted in spectra with clear and straightforward sequence ladders, consisting of almost exclusively b-ions. We also performed a direct comparison of the following four combinations: (i) Lys-C, SCX enrichment, and MALDI-MS/MS analysis; (ii) Lys-N, SCX enrichment, and MALDI-MS/MS analysis; (iii) Lys-N, SCX enrichment, and ESI-CID MS/MS analysis; and (iv) Lys-N, SCX enrichment, and ESI-ETD MS/MS analysis. The comparison clearly demonstrated that only the combinations of Lys-N-generated peptides with MALDI-CID and ESI-ETD MS/MS provided very clear sequence ladders. Furthermore the potential of this method for facilitating de novo sequencing was illustrated by the successful identification of proteins from an SDS-PAGE band of an ostrich tissue lysate where ostrich represents a species with an unsequenced genome.

EXPERIMENTAL PROCEDURES

Materials—

Protease inhibitor mixture and Lys-C were purchased from Roche Diagnostics. Metalloendopeptidase Lys-N was obtained from Seikagaku Corp. (Tokyo, Japan). Iodoacetamide, TFA, and α-cyano-4-hydroxycinnamic acid were purchased from Sigma-Aldrich. DTT was obtained from Fluka Biochemical (Buchs, Switzerland). HEK293 cells were a gift from the ABC Protein Expression Center (Utrecht University, The Netherlands). Ostrich steak was purchased at the local butcher. Water that was used in these experiments was obtained from a Milli-Q purification system (Millipore, Bedford, MA). All other chemicals were purchased from commercial sources and were of analysis grade.

Sample Preparation—

HEK293 cells were harvested at a density of ∼1.5 × 10⁶ cells/ml and stored at −20 °C. Cells were thawed and resuspended in ice-cold lysis buffer (15 ml of PBS, 150 μl of Tween 20, and protease inhibitor mixture). After Dounce homogenizing on ice, the lysate was stored at 0 °C for 10 min. Subsequently centrifugation at 20,000 × g at 4 °C yielded separation of soluble and insoluble protein fractions. The soluble fraction was collected, and the concentration was determined by a Bradford assay. The lysate was dissolved in 50 mm ammonium bicarbonate to a concentration of 4 mg/ml.

Approximately 200 μg of ostrich muscle tissue was frozen in liquid nitrogen and pulverized with a mortar and pestle after which 8 m urea was added, and the sample was homogenized by microtip sonication. 30 μg of lysate was then separated by 1D SDS-PAGE.

In-solution Digestion—

HEK293 lysate was reduced with 45 mm DTT (50 °C, 15 min) followed by alkylation using 110 mm iodoacetamide (in the dark, room temperature, 15 min). Buffer exchange was performed using 5-kDa spin columns. The resulting solutions were dried in a vacuum centrifuge and resuspended in 50 mm ammonium bicarbonate. One part was digested with Lys-C, and an equal amount was digested with Lys-N. Lys-C was added to the samples at a 1:50 (w/w) ratio, whereas Lys-N was added at a ratio of 1:85 (w/w). Both solutions were incubated overnight at 25 °C.

In-gel Digest—

A gel band at ∼42 kDa was cut out of the gel and washed with water. After shrinking the gel piece with acetonitrile the contents were reduced with 10 mm DTT (60 °C, 1 h) followed by alkylation using 55 mm iodoacetamide (in the dark, room temperature, 30 min). After shrinking the gel pieces with acetonitrile the gel was incubated with Lys-N (10 ng/μl) overnight at 37 °C. Supernatant was transferred to new Eppendorf tubes. Peptides were extracted by adding 50% acetonitrile, 5% formic acid to the gel pieces. The supernatant was added to the previous supernatant.

SCX—

SCX was performed using an Agilent 1100 series LC system with a C₁₈ Opti-Lynx (Optimize Technologies) guard column and Polysulfoethyl A SCX column (PolyLC, Columbia, MD; 200 mm × 2.1-mm inner diameter). Sample was dissolved in 0.05% formic acid and loaded onto the guard column at 100 μl/min and consecutively eluted onto the SCX column with 80% ACN, 0.05% formic acid. SCX buffer A was 5 mm KH₂PO₄, 30% ACN, pH 2.7; SCX buffer B was 350 mm KCl, 5 mm KH₂PO₄, 30% ACN, pH 2.7. Gradient elution was performed as follows: 0–85% B in 45 min, 85–100% B in 6 min, and 100% B for 4 min. A total of 53 1-min fractions was collected, and fractions were dried in a vacuum centrifuge.

Off-line Nano-RP-LC and MALDI Preparation—

Nano-RP-LC separation of SCX fractions 31–33 and of the Lys-N in-gel digest of ostrich muscle tissue was performed on a Famos/Ultimate LC instrument (LC Packings, Naarden, The Netherlands) using a vented column setup (26). The trapping column was Aqua C₁₈ (Phenomenex, Torrance, CA; 0.1 × 20 mm), the analytical column was Aqua C₁₈ (0.075 × 230 mm). All columns were packed in-house. Trapping was performed at 5 μl/min for 10 min, and analytical separation was performed at 0.2 μl/min, passively split from 200 μl/min. Buffer A was 95% H₂O, 5% ACN, 0.05% TFA; buffer B was 5% H₂O, 95% ACN, 0.05% TFA. The gradient was 0–32% B in 35 min, 32–100% B in 2 min, and 100% B for 5 min. 20-s fractions were automatically mixed with 0.5 μl of MALDI matrix (3 mg/ml α-cyano-4-hydroxycinnamic acid, 80% ACN, 0.1% TFA) and spotted onto a MALDI target using a Probot Microfraction collector (LC Packings).

MALDI-TOF/TOF—

MALDI-TOF/TOF analysis was performed with a 4700 Proteomics Analyzer (Applied Biosystems, Darmstadt, Germany). Spectra were acquired in positive and reflectron ion modes in the m/z range 900–4000. Maximally 1500 shots were averaged for each spectrum. Data were acquired at a laser repetition rate of 200 Hz, an acceleration voltage of 20 kV, a grid voltage of 70%, and a digitizer bin size of 0.5 ns. The calibration of the spectra was done using a standard peptide calibration mixture (Applied Biosystems). CID spectra were obtained with a collisional energy of 1 keV and averaging maximally 15,000 shots. Maximally five MS/MS precursors were selected per MS run and were excluded from further selection once sequenced.

Nano-LC-ESI-CID MS/MS and Nano-LC-ESI-ETD MS/MS—

An aliquot of SCX fractions 31–33 was also analyzed by nano-LC-CID/ETD MS/MS. An Agilent 1100 HPLC system was connected to an LTQ XL linear ion trap mass spectrometer with an ETD source at the back (Thermo Fisher Scientific Inc., Waltham, MA). The instrument was equipped with a 20-mm × 100-μm-inner diameter Aqua C₁₈ trap column (Phenomenex) and a 200-mm × 50-μm-inner diameter Reprosil C₁₈ RP analytical column (Dr. Maisch, Ammerbuch-Entringen, Germany). The fractions were separated by using a 95-min 100 nl/min linear gradient from 0 to 60% solvent B (0.1 m acetic acid in 80% acetonitrile (v/v)) in which solvent A was 0.1 m acetic acid. The MS instrument was operated in positive ion mode, and parent ions were isolated for fragmentation by CID or ETD in data-dependent mode. ETD fragmentation was performed with supplemental activation, fluoranthene was used as reagent anion, and ion/ion reaction in the ion trap was allowed to take place for 100 ms.

Peptide Identification—

MALDI data analysis and peak list generation was performed with the Data Explorer™ software version 4.5 (Applied Biosystems). Raw ESI-CID and ESI-ETD MS data were converted to peak lists using Bioworks Browser software, version 3.3.1. For the work on the HEK293 lysate, spectra were searched against the IPI human (v3.37, 69,164 entries searched) using Mascot (version 2.1.0) with Lys-C or Lys-N cleavage specificity allowing one missed cleavage, carbamidomethyl (Cys) as fixed modification, and oxidation (Met) as variable modification. Peptide tolerance was set to 100 ppm for 1+ peptide charge (MALDI) or 0.5 Da for 2+ and 3+ peptide charges (ESI), and MS/MS tolerance was 0.2 (MALDI) or 0.9 Da (ESI). Peptides were identified with a minimum Mascot score of 30, and at these settings the false discovery rate was less than 0.75% as estimated by using the Mascot decoy database function. For further data analysis, Mascot data were imported into Scaffold 1.7.

For the de novo sequencing of ostrich proteins the most abundant peptides in the MALDI-MS spectra were fragmented. Manual annotation of CID spectra was performed using the mass differences between adjacent fragment ions. The obtained sequences were BLAST-searched against a human and chicken UniProt database. Several assigned peptides were identical to peptides from chicken and/or human creatine kinase and actin. Other peptides were very similar but revealed ostrich-specific single or double amino acid differences.

RESULTS

In a typical proteomics experiment, digestion is performed by trypsin generating peptides with a C-terminal arginine or lysine. A mixture of N- and C-terminal fragment ions can be detected after CID of these peptides, but as the arginine and lysine residues are more basic than the α-amino group, generally y-ions are more abundant (10,19). This study evaluated the unique situation that arises when proteolysis is performed by Lys-N as peptides are yielded with an N terminus accommodating both α- and ɛ-amino basic entities. Fragmentation of Lys-N peptides is thus likely skewed toward the production of N-terminal ions (see Fig. 1) as has been hinted at before when looking at a few individual peptides (27,28). To more comprehensively assess the fragmentation behavior of Lys-N-produced peptides, a whole cellular lysate was digested in parallel with Lys-N and Lys-C. Lys-C, like trypsin, generates peptides with a C-terminal basic amino acid residue. However, Lys-C has no specificity for arginine, and so comparing Lys-N with Lys-C is more appropriate than comparing it with trypsin. The digested cellular lysates were first subjected to low pH SCX chromatography to enrich for and isolate peptides with a single basic lysine residue (24,29, 30). A few consecutive fractions of the SCX run will provide a set of peptides with a single C-terminal lysine residue for the Lys-C-digested sample and with a single N-terminal lysine from the Lys-N digest (24, 30–32). Off-line nano-RP-LC separation was performed on these selected SCX fractions. The eluent was subsequently mixed postcolumn with α-cyano-4-hydroxycinnamic acid and automatically fractionated and spotted onto a MALDI target plate, an experimental setup adopted from our previously described off-line zwitterionic hydrophilic interaction liquid chromatography setup (33). The fractionated peptides were then subjected to analysis by MALDI-TOF/TOF-based tandem mass spectrometry. As an illustrative result, fragmentation of the peptide KCQEVISWLDANTLAE, as depicted in Fig. 2a, resulted in a CID spectrum that was typical for Lys-N proteolytic peptides (Fig. 2b) and displayed a complete sequence ladder consisting of b-ions. As shown in Fig. 2b, Lys-N peptide sequences can be easily read off as there are no significant “interfering” ion series present. As indicated in Fig. 2, the b₁-ion is quite abundant, and additionally the b₁-related ions a₁ and a₁ − NH₃ were found to often be the base peak in the spectra. C-terminal fragment ions (y-ions) were detected at a very low frequency and/or intensity.

Fig. 1.

Schematic representation of CID fragmentation of Lys-C- and Lys-N-derived singly charged peptide ions. Lys-C peptides have a basic N and C terminus; therefore, both termini will be protonated, leading to a mixture of b- and y-ions in CID. Lys-N peptides concentrate the basicity at the N terminus, leading to predominantly b-ions in CID. AA, amino acid.

Fig. 2.

Representative MALDI-CID spectra of peptides identified from a Lys-N-digested HEK293 cellular lysate. a, a clean b-series sequence ladder is detected for peptide KC*QEVISWLDANTLAE (m/z 1876.73, 1+; C*, carbamidomethylated cysteine). b, six typical CID spectra of Lys-N peptides annotated by Scaffold dominated by a nearly full series of b-ions: KPGNQNTQVTEAWN (m/z 1586.61, 1+), KGFSEGLWEIENNPTV (m/z 1819.85, 1+), KGQGSVSASVTEGQQNEQ (m/z 1833.79, 1+), KLGGTIDDC*ELVEGLVLTQ (m/z 2059.88, 1+), KC*NEIINWLD (m/z 1304.52, 1+), and KDQIYDIFQ (m/z 1169.54, 1+).

We further examined and compared the effect of the N- or C-terminal position of the lysine on the fragmentation of the peptides in CID. Typical peptide fragmentation spectra are shown in Fig. 3, which incorporates MALDI-CID spectra of the same peptide with a lysine either on the C terminus or the N terminus (respectively generated by Lys-C and Lys-N digestion) as well as spectra of the doubly charged ion of the same Lys-N peptide analyzed by ESI-CID MS/MS and ESI-ETcaD MS/MS. The MALDI tandem mass spectrum obtained for the peptide with a C-terminal lysine contains a mixture of b- and y-ions (and no complete series) and is clearly more complex than the spectrum from the analogous Lys-N peptide, which provides a nearly complete b-ion series. We found that fewer immonium ions were detected, and the number of non-informative background peaks seemed to be lower in the spectra of the Lys-N peptide, possibly related to the reduced number of fragment ion pathways available. The tandem mass spectrum of the doubly charged ion of the Lys-N peptide obtained by ESI-CID also shows a complex spectrum with both b- and y-ions that can be attributed to the availability of two protons for the fragment ions (24). Finally the ETcaD spectrum of the doubly charged peptide ion is comprised of only c-ions in agreement with our recent findings and in appearance similar to the MALDI-CID spectrum (24).

Fig. 3.

Representative mass spectra annotated by Scaffold of the same peptide with lysine on the C (a) or the N terminus (b, c, and d) of the peptide, respectively. In a (m/z 1467.60, 1+) and b (m/z 1467.58, 1+) are shown the MALDI-CID spectra, in c (m/z 734.20, 2+) the ESI-CID spectrum obtained by an ion trap is shown, and in d (m/z 734.26, 2+) the ESI-ETcaD mass spectrum is shown. MALDI-MS/MS of peptides with a C-terminal lysine provides a spectrum with both b- and y-ions, whereas spectra of peptides with an N-terminal lysine are less complex and dominated by b-ions. However, ESI-CID spectra of doubly charged peptides with an N-terminal lysine show a mixture of b- and y-ions due to the double charge, whereas the MALDI-TOF/TOF spectrum of the singly charged peptide results in a full b-ion series. Also the ETD experiments result in mainly c-ion generation (24).

As stated, MALDI-CID of peptides with a single N-terminal lysine yielded mass spectra dominated by b-ions and with only minor y-ions. However, when a proline is present in the sequence this straightforward fragmentation pattern becomes disrupted. Peptide cleavage on the N-terminal side of a proline leads to preferential formation of a y-ion (Fig. 4a). An illustrative example of a CID spectrum of a peptide containing two prolines is shown in Fig. 4b. The two most intense peaks correspond to y-ions with an N-terminal proline residue. Also two intense internal fragments are detected with an N-terminal proline residue. Nevertheless an almost complete and clear b-ion series is still detected.

Fig. 4.

CID fragmentation of proline-containing Lys-N peptides. a, schematic representation of CID fragmentation of Lys-N peptides that do not or do contain proline. Fragmentation of a proline-free peptide results in the detection of mainly b-ions, whereas cleavage N-terminal of a proline residue of a Lys-N peptide results in the detection of dominant y-ions. b, representative MALDI-CID spectrum of a proline-containing Lys-N peptide (KQPAIMPGQSYGLEDGSC*SY, m/z 2187.78, 1+) annotated by Scaffold. The two intense y-ions correspond to ions with an N-terminal proline. Also two intense internal fragments with an N-terminal proline are detected. Note that still a nearly full b-ion series can be detected. AA, amino acid.

MALDI-CID analysis of three SCX fractions, enriched for peptides containing a single, N-terminal lysine, led to the identification of a total of 247 peptides (Mascot score ≥30; false discovery rate, 0.75%) with an N-terminal lysine residue (tandem mass spectra are available in the PRIDE (Proteomics Identifications) database under accession number 3380). Of these, 36 contained an extra basic residue and were initially removed from further analysis. Of the remaining 211 peptides, 119 contained one or more proline residues. To account for the proline effect on ion formation, proline-containing peptides were analyzed separately from proline-free peptides. N-terminal b-ions represented in number 84% of the detected backbone fragment ion types in the 92 spectra of proline-free Lys-N peptides, corresponding to 94% of the total signal intensity of b- and y-ions confirming the dominance of N-terminal fragment ions (see Table I). Furthermore on average, two-thirds of all theoretically obtainable b-ions were detected. A slightly lower percentage (75%) of b-ions was observed for proline-containing peptides. If y-ions with an N-terminal proline were removed, the relative b-ion intensity percentage increased (81%) although not to the level of proline-free peptides. Nevertheless the dominant nature of b-ions in such spectra still allows straightforward interpretation (Fig. 4). The relative intensity of b-ions in fragmentation spectra of Lys-C peptides is in sharp contrast with those of Lys-N peptides. In tandem mass spectra of Lys-C peptides b- and y-type ions are approximately equally abundant (see Table I).

Table I.

Analysis of the frequency and normalized overall intensity of b-ions compared with y-ions in MALDI-CID spectra of Lys-N- and Lys-C-derived peptides

The percent intensity is the intensity of all b-ions divided by the sum of intensity of all b- and y-ions.

b-ion vs. y-ion	Lys-N (211 peptides)		Lys-C (220 peptides)
	No proline	Proline-containing	No proline	Proline-containing
Frequency (%)	84	75	46	48
Intensity (%)	94	71 (81a)	50	43

^aAfter subtracting the intensity of y-ions with an N-terminal proline.

To evaluate the applicability of Lys-N digestion for genuine de novo sequencing, muscle tissue from an ostrich, of which the genome has not (yet) been sequenced, was lysed and separated by 1D SDS-PAGE. It should be noted that in our earlier work no significant difference in yield was found between in-gel digestion with trypsin and that with Lys-N (24) (supplemental Fig. 1 indicates the sequence coverage and peptide signal achieved by LC-MS for an in-gel digestion of equivalent amounts of BSA by Lys-N and trypsin). A band at ∼42 kDa was excised from the ostrich gel, and its content was digested with Lys-N. These peptides were then separated by “off-line” nano-LC with the eluent being mixed postcolumn with α-cyano-4-hydroxycinnamic acid and automatically spotted onto a MALDI target plate. MALDI-TOF/TOF analysis was subsequently performed. CID spectra of ∼20 of the most intense peptides were manually interpreted. Identified peptide sequences were homology-searched against UniProt human and chicken databases using BLAST. Four unique peptides were found to align with creatine kinase (Fig. 5a), whereas five unique peptides were found to align with actin (Fig. 5b). Strikingly one of the actin peptides was found in four different forms. Alongside an unmodified form we observed sequences containing methionine oxidation, methionine oxidation plus tryptophan hydroxylation, and a decomposed carboxymethylated methionine (34). Such modifications would be missed by a database search strategy if one does not, in advance, input these possibilities in the submission criteria. Three peptide sequences of creatine kinase and four sequences of actin were identical to human and chicken protein sequences allowing the identification of the protein. For both proteins we found peptides to have a sequence that differed slightly from the human and chicken sequences. In Fig. 5c the CID spectrum of such a peptide of creatine kinase is depicted. The clear b-ion series facilitated its de novo sequencing. The sequence deviates at the fourth (alanine) residue compared with chicken and human, whereas the 11th (glutamine) residue is similar to chicken but different from human. In Fig. 5d a CID spectrum is depicted of actin peptide KYPIEHAIITNWDDME. This peptide aligns with human and chicken actin except for residue 7 (alanine). Additional annotated spectra can be found in supplemental Fig. 2.

Fig. 5.

Lys-N facilitates de novo sequencing of ostrich creatine kinase and actin. a, multiple sequence alignment of de novo sequenced ostrich peptides with partial creatine kinase sequences of human (P06732, UniProtKB/Swiss-Prot) and chicken (P00565). b, multiple sequence alignment of de novo sequenced ostrich peptides with partial actin sequences of human (P62736) and chicken (P68139). c, MALDI-CID spectrum of creatine kinase peptide KLSAEALNSLEGEF (m/z 1507.71, 1+), which has a Val to Ala mutation of the fourth residue compared with chicken and human creatine kinase. d, MALDI-CID spectrum of actin peptide KYPIEHAIITNWDDME (m/z 1974.84, 1+), which has a Gly to Ala mutation of the seventh residue compared with chicken and human actin. *, identical amino acid; : conserved substitution; ., semi-conserved substitution.

DISCUSSION

In the present study, we show that in MALDI-CID fragmentation of Lys-N peptides the basic N terminus has a strong influence on fragment ion formation and leads to clean b-ion ladder series. These ladders are easily deciphered as the presence of interfering ion series is significantly reduced, or they are altogether missing. We show that this is clearly different from MALDI-CID spectra of peptides with a C-terminal lysine, generated by a Lys-C protease, where a mixture of both b- and y-ion series was detected as expected because both the N terminus and C terminus contain basic entities. Furthermore MALDI provides informative lower m/z ions including the b₁-ion (see Fig. 3). An additional advantage is the good mass accuracy and resolution for these spectra because analyses were performed with a TOF mass analyzer. It should be noted that b₁-ions are typically missing in tandem MS of tryptic peptides (35). The presence of the b₁-ion for Lys-N peptides is likely to originate through a distinct, lysine-specific cleavage pathway as it cannot be achieved via the regular b_x/y_z fragmentation pathway (36).

Evidently CID of doubly charged Lys-N peptide ions is different from CID of singly charged Lys-N peptide ions. In ESI-CID MS/MS of the identical, Lys-N-generated, doubly charged peptide ions, a mixture of b- and y-ions is detected with the b-ion series being somewhat more intense (24, 28,30). During fragmentation of doubly charged peptide ions, one of the protons will be sequestered by lysine. The other, mobile proton will then be less prone to protonate the N-terminus, and thus the possibility of the formation of also C-terminal ions is increased.

To isolate and obtain a statistically significant number of Lys-N peptides with a single basic residue an initial low pH SCX chromatographic step was added (31). However, the fact that we chose to isolate peptides with a single N-terminal lysine does not mean that the remaining peptides, which contain for example more than one basic residue, are of no value. Tandem mass spectra can also be obtained from Lys-N peptides with additional basic residues where the fragmentation will follow a pattern similar to that achieved with tryptic peptides containing miscleavages. Although containing b- and y-ion series, they are not as straightforward to interpret manually as the spectra from peptides with a single N-terminal lysine; these spectra have similar appearances to those originating from tryptic or Lys-C peptides (supplemental Fig. 3).

Scrutinizing spectra for additional trends, we observed that the MALDI-CID spectra of proline-containing peptides contained intense peaks corresponding to fragment ions with an N-terminal proline, a well described phenomenon (4, 37–39). The intensity of these peaks has been explained by the extraordinary structure of proline with its side chain forming a five-membered ring with the peptide backbone. This hinders cleavage C-terminal of proline as it would involve the generation of a strained bicyclic structure (39). Cleavage on the N-terminal side causes the formation of the secondary amine group of proline that can sequester the mobile proton, causing the generation of the C-terminal fragment ion (see Fig. 4b) (38). Our data suggest that a combination of both effects prompt the emergence of intense peaks of y-ions with an N-terminal proline. First, the fact that the intensity of a b-ion with a C-terminal proline is significantly lower than other b-ions indicates that the formation of a b-ion with a bicyclic C-terminal proline structure is unfavored (39). Second, the high gas phase proton affinity of the proline amine group is reflected in the intensity of the y-ion with an N-terminal proline being substantially larger than the intensity of the b-ion that is generated N-terminal of the same proline. It appears thus that the two amino groups at the N-terminus of the potential b-ion exert less influence than the secondary amine of the proline present on the y-ion at the point of cleavage (38). On average, b-ions in MALDI-CID spectra of proline-containing peptides account for 71% of the total intensity of b- and y-ions. When the intense y-ions corresponding to fragments with an N-terminal proline were removed, this percentage of b-ions increased to 81%. This is slightly lower than the percentage for proline-free peptides. Despite the occurrence of y-ions, the remainder of such Lys-N MALDI-CID spectra is still dominated by sequence ladders of b-ions. About 56% of the peptides we identified with a single N-terminal lysine contained a proline residue. This is similar to a recent study in which the occurrence of at least one proline in a tryptic peptide was determined, both theoretically (using the IPI human database) and practically, to be around 50% (40).

Efficient and facile de novo sequencing requires good quality straightforward mass spectra. Lys-N proteolysis allows significant portions of a protein to generate simplified MALDI-CID spectra without any derivatization steps. An in silico digest of all proteins in the IPI human database revealed that approximately one-third of all theoretical Lys-N peptides contain a single basic residue. Therefore, as an example of the applicability of Lys-N proteolysis for de novo sequencing we strived to identify the protein contents of a 1D SDS-PAGE gel band of ostrich muscle tissue. Clear, simplified spectra were generated facilitating de novo sequencing. Sequenced peptides could be aligned to chicken (the phylogenetically closest species to ostrich with a genome that is sequenced) creatine kinase and actin. Of these peptides, at least two would not have been identified if a traditional database search strategy had been performed because these peptides slightly differ in sequence compared with sequences available in genomics databases. For example, in the ostrich creatine kinase peptide KLSAEALNSLEGEF, the third residue, an alanine, is a valine in chicken and human; this is probably a DNA point mutation as the translation codons of valine and alanine only differ by one nucleotide. The same is true for the actin peptide KYPIEHAIITNWDDME where a single DNA point mutation could explain the conversion of glycine in chicken and human to alanine in ostrich. Furthermore modifications to this peptide were found in other mass spectra that could easily be detected, including methionine oxidation, tryptophan hydroxylation, and decomposed carboxymethylated methionine. Although methionine oxidation is generally included as a variable modification in database searching, for tryptophan hydroxylation and decomposed carboxymethylated methionine this is generally not the case, thus underscoring the potential of our de novo sequencing approach. Similarly we expect that other modifications can be easily identified and located using the MS/MS sequence ladders by combining Lys-N digestion with MALDI-MS/MS analysis. Labile modifications such as phosphorylation might result in neutral loss-dominated spectra as fragmentation is performed by CID. However, these spectra will still be simpler than those achieved by tryptic peptides.

As discussed before, an apparent drawback of Lys-N MALDI-CID is that the clear b-ion-dominated spectra will exclusively be observed for peptides containing a single N-terminal basic residue. Depending on the position in the peptide sequence, an extra basic residue might lead to the generation of an additional (y-) ion series thereby complicating manual interpretation. Chemical derivatization that aids de novo sequencing such as CAF and SPITC can be applied, but it is necessary to have a basic residue at the C terminus; thus tryptic peptides are necessary. In theory, the number of peptides with a single basic residue in a tryptic digest is ∼3-fold higher than in a Lys-N digest. However, internal basic residues negatively affect these derivatization strategies in a similar way to Lys-N, i.e. generating additional fragment ion series (11,16, 18,19). Furthermore the sulfonyl group, which is key to CAF/SPITC, reduces peptide signal intensities due to the intrinsic negative charge. Also these chemical derivatization steps potentially lead to sample loss and further signal reduction. Finally side reactions (often) hamper the analysis by increasing the complexity of the sample with uninformative peptides. Through the use of Lys-N de novo sequenceable peptides can be attained without chemical derivatization.

In summary, here we evaluated MALDI-CID fragmentation of singly charged Lys-N-generated peptide ions. With a lysine residue on the peptide N terminus, protonation of N-terminal fragments of these peptides is favored, resulting in the detection of dominant, nearly complete series of b-ions, rendering Lys-N a useful protease aiding in the unambiguous sequencing of peptides in MALDI-MS/MS. Lys-N can be applied on a proteome scale using a low pH SCX multidimensional protein identification technology strategy where one can isolate and separate single lysine-containing peptides that represent the whole protein content. Equally Lys-N is also applicable in a 1D or two-dimensional gel strategy where a significant portion of generated peptides for each protein can be de novo sequenced in a similar vein to CAF and SPITC.