Efficient Fractionation and Improved Protein Identification by Peptide OFFGEL Electrophoresis

EXPERIMENTAL PROCEDURES

Sample Preparation—

An E. coli total protein lysate from Bio-Rad was reduced and denatured using 50% 2,2,2-trifluoroethanol with 200 mm dithiothreitol at 95 °C for 20 min. This was followed by alkylation with iodoacetamide at room temperature for 1 h. The reduced and alkylated sample was diluted 1:10. Trypsin (Promega, Madison, WI) was added at 1:20 enzyme:substrate, and then the sample was incubated overnight at 37 °C. The digest was aliquoted, dried, and stored frozen until use.

OFFGEL Electrophoresis—

For pI-based peptide separation, the 3100 OFFGEL Fractionator and the OFFGEL Kit pH 3–10 (both Agilent Technologies) with a 24-well setup was used according to the protocol of the supplier. Ten min prior to sample loading, 24-cm-long IPG gel strips with a linear pH gradient ranging from 3 to 10 were rehydrated in the assembled device with 25 μl of focusing buffer per well. Two hundred μg of the E. coli tryptic digest was diluted in focusing buffer to a final volume of 3.6 ml, and 150 μl of sample was loaded in each well. The sample was focused with a maximum current of 50 μA and typical voltages ranging from 500 to 4000 V until 50 kVh was reached after 24 h. The recovered fractions (volumes between 100 and 150 μl) were acidified with 5 μl of formic acid.

HPLC-Chip/MS Analysis—

A 0.5-μl aliquot of each OFFGEL electrophoresis fraction was injected onto an LC/MS system consisting of an 1100 Series liquid chromatograph, HPLC-Chip Cube MS interface, and 1100 Series LC/MSD Trap XCT Ultra ion trap mass spectrometer (all Agilent Technologies). The system was equipped with an HPLC-Chip (Agilent Technologies) that incorporated a 40-nl enrichment column and a 43-mm × 75-μm analytical column packed with Zorbax 300SB-C18 5-μm particles. Peptides were loaded onto the enrichment column with 97% solvent A (water with 0.1% formic acid) and 3% B (90% acetonitrile with 0.1% formic acid) at 4 μl/min. They were then eluted with a gradient from 3% B to 45% B in 30 min, followed by a steep gradient to 80% B in 5 min at a flow rate of 0.3 μl/min. The total runtime, including column reconditioning, was 45 min.

Molecular masses were recorded using data-dependent MS/MS acquisition. The MS and MS/MS conditions employed were:

• Drying gas flow: 4 liters/min, 300 °C; capillary voltage: 1800 V; skim 1: 30 V; capillary exit: 75 V; trap drive: 85; averages: 1; ion current control: on; maximum accumulation time: 150 ms; smart target: 500,000; MS scan range: 300–2000; ultra scan: on.

• MS/MS: number of parents: 5; averages: 1; fragmentation amplitude: 1.25 V; SmartFrag: on, 30–200%; active exclusion: on, 2 spectra, 1 min; prefer +2: on; exclude +1: on, MS/MS scan range: 100–2000; ultra scan: on; ion current control target: 500,000.

Database Search—

The SwissProt database was searched with the restriction to E. coli, using the Agilent Spectrum Mill Server software (Rev A.03.02.) installed on a dual Xeon 2.4-GHz computer. Peak lists were created with the Spectrum Mill Data Extractor program with the following attributed: scans with the same precursor ± 1.4 m/z were merged within a time frame of ± 15 s. Precursor ions needed to have a minimum signal to noise value of 25. Charges up to a maximum of 7 were assigned to the precursor ion, and the ¹²C peak was determined by the Data Extractor. The SwissProt database (01/12/03), with 120,961 total entries and 4,830 for E. coli proteins, was searched for tryptic peptides with a mass tolerance of ± 2.5 Da for the precursor ions and a tolerance of ± 0.7 Da for the fragment ions. Two missed cleavages were allowed. A Spectrum Mill autovalidation was performed first in the protein details mode. Minimum scores, minimum scored peak intensity (SPI), forward minus reversed score threshold, and rank 1 minus rank 2 score threshold for peptides were dependent on the assigned precursor charge (see Table I).

Then autovalidation in the peptide mode was performed using a score threshold of 13 and SPI of 70% for 1+, 3+, and 4+ and 11 and 60% for 2+ precursor ions. Forward minus reversed score threshold and rank 1 minus rank 2 score threshold were set to 2.

All protein hits found in a distinct database search by Spectrum Mill are non-redundant. To eliminate redundancy, the Protein Summary Modes groups all proteins that have at least one common peptide, and only the highest scoring member of each protein group is shown and counted in the protein list.

RESULTS

The recently developed OFFGEL electrophoresis method was used in this work to fractionate E. coli peptides in the pH range of 3–10 prior to mass analysis on a HPLC-Chip/MS system. The experimental setup employed for fractionation by OFFGEL electrophoresis is shown schematically in Fig. 1.

Protein or peptide separation takes place in a two-phase system with an upper liquid phase that is divided in compartments and a lower phase that is a conventional rehydrated IPG gel strip. Typically, the sample is diluted in the focusing buffer and loaded into all wells. Because there is no fluidic connection between the wells, proteins or peptides are forced to migrate through the IPG gel where the actual separation takes place. After IEF, the proteins or peptides are present in the liquid phase and can be recovered conveniently from the wells for further processing.

Tandem mass spectroscopic data were analyzed with the Spectrum Mill software and the SwissProt database restricted to E. coli using autovalidation criteria as described above. In total, 3454 peptides and 670 proteins were identified. A list of all protein hits that gives information about the number of unique peptides, sequence coverage, and peptide scores, etc. is available as supplemental data 1. More detailed information regarding protein assignments based on single-peptide assignments is available as supplemental data 2.

An in silico tryptic digest of the whole E. coli proteome shows that the pI values of E. coli peptides are unevenly distributed across the pH scale with gaps at pH 7–8 and around pH 5 and pH 9 and the majority of peptides clustering at pH 3.6–4.8 and pH 5.2–6.2 (9). In agreement with these theoretical considerations, most peptides were identified in OFFGEL electrophoresis fractions 2–6 (covering on the IPG gel the pH range 3.6–4.9), 9–14 (pH range 5.7–7.0) and 21–24 (pH range 8.9–9.7) (Table II, Fig. 2a).

One way to judge the fractionation quality of this technique is to look at the number of fractions in which each distinct peptide was found. As shown in Fig. 3, 74% of the identified peptides are found in only one fraction and about 90% are found in one or two fractions. In agreement with former studies (2), most peptides unique to each fraction were found in the pH range from 3.6 to 4.7 (fractions 2–5, Fig. 2a), which suggests this pH segment for a “zoom” analysis on a narrow range IPG strip (3).

The Spectrum Mill software automatically calculates the pI value for every identified peptide according to the algorithm developed by Bjellqvist et al. (10). Using these data, average pI values with standard deviations were calculated, without any data filtering, for all peptides identified in each fraction. The average pI values fit fairly well with the expected pH range calculated according to the specifications of the IPG gel strip supplier (Fig. 2b). Major deviations were observed only for fractions 13–18 (pH range 6.5–8.1). These deviations very likely resulted from the low number of peptides found in this pH range (see above). Deviations between expected and observed pI values were reported for conventional IPG IEF as well (2, 3).

Because IEF separates peptides according to their pI values, consideration of the standard deviation of the pI value distribution in each fraction should provide an alternative method to judge the resolving power of OFFGEL electrophoresis. Taking all fractions together, an average standard deviation of ± 0.42 pI units was observed. However, the standard deviations differ significantly between fractions. Two general trends were observed: (i) fractions with a low number of identified peptides had high standard deviations (compare Fig. 2, a and b), and (ii) fractions in the neutral and basic pH range (fractions 13–24) had higher standard deviations than fractions in the acidic pH range (fractions 1–12). When fractions with a low peptide content (<60 peptide identifications: fractions 7, 8, and 18–20) were excluded, the average standard deviation dropped to ± 0.33 pI units. Very low standard deviations that approach the pH interval covered by a single well (i.e. 0.26 pH units) were observed for several fractions with high peptide content only in the acidic pH range (Table I). A very similar distribution of standard deviations was reported for conventional IPG IEF as well (2).

An overall view of the deviation of the predicted pI values of every identified peptide spectrum from the average pI calculated for each fraction is shown in Fig. 4 for validated as well as for non-validated spectra. For the validated spectra, a narrow distribution around zero is observed with 51% of the spectra within the one well width of ± 0.13 pI units (Fig. 4a). In contrast to the validated spectra, the non-validated spectra show a broad distribution across the whole pI scale (Fig. 4b). However, 22% of these non-validated spectra still cluster within one well width implying that the stringent autovalidation criteria reject a substantial number of correct peptide spectra.

We tested whether the experimentally determined pI could be used as a filter to sort false negative spectra out of the pool of non-validated spectra. To this end, the quality of all non-validated peptide spectra with scores ranging from 5 to 10 from fractions 3 and 4 were manually assessed (Fig. 5). The majority of the peptide spectra within ± 0.13 pI units of the average fraction pI were good quality spectra. Over 90% of the peptide spectra with scores ranging from 8 to 9 were rated as “good,” and in the score range from 9 to 10 only high quality spectra were observed (Fig. 5a). In contrast to this, the quality of non-validated peptide spectra with pI values outside ± 0.13 pI units of the average fraction pI was much lower (Fig. 5b). These data showed that an additional pI filter allows attenuating the validation criteria without risking a higher number of false positives. However, the number of additional spectra that could be validated by this approach was limited: 3.6% or 43 spectra had in the present example the same data quality as autovalidated spectra (i.e. only good-rated spectra) (Fig. 5a).

DISCUSSION

The success of proteomics has lead to an explosion of the number of publications in the past few years and is to a great extent because of improvements in biochemical separation techniques. Most importantly, the broad application of two-dimensional gel electrophoresis was enabled by the introduction of IPG gels that brought superior resolution and reproducibility to the first-dimension IEF (11, 12). Because two-dimensional gel electrophoresis is labor intensive and difficult to automate, it would be beneficial for high throughput applications to combine IPG gel-based IEF technology with separation techniques that are more amenable to automation like liquid chromatography. However, the combination of conventional IPG gel-based IEF with liquid phase separations is hampered by the fact that proteins or peptides have to be transferred from the gel to liquid phase for further analysis. This is a tedious and error-prone procedure that includes several steps like cutting the IPG strip after IEF, sample extraction, and cleanup (2, 3). An alternative to circumvent this problem is offered by OFFGEL electrophoresis. This technique takes advantage of the impressive separation power of IPG gels but delivers sample after IEF in the liquid phase. High resolution protein and peptide separations can be obtained using this method (5–8). Here, we judged the fractionation quality obtained by OFFGEL electrophoresis by looking at the fractionwise distribution of the identified peptides and the distribution of the calculated pI values of the identified peptides in each fraction, respectively (Figs. 2–4, Table II). The most straightforward approach is the former one because it does not rely on the accuracy of pI prediction. Because about 90% of the peptides were identified only in one or two fractions, we concluded that the fractionation quality was very good (Fig. 3). However, standard deviations of the average pI values in each fraction that were lower than ± 0.20 pH units and hence close to the size of the pH segment covered by a single well in the respective setup (i.e. 0.26 pH units) were observed only in the acidic pH range. Factors that could account for the higher standard deviations of fractions in the neutral and basic pH range are on the one hand the ill defined pI values of peptides in the neutral pH range and the diffuse focusing of predominantly basic and proline rich peptides (7). On the other hand the lower consistency of pI prediction algorithms in the basic pH range implies that the accuracy of the pI prediction is less good than in the acidic pH range (2). Furthermore, amino acid-specific trends that affect the accuracy of pI prediction could account for a large portion of the standard deviation observed in each fraction as well (2).

The inclusion of OFFGEL electrophoresis into the proteomic workflow offers several advantages: (i) the pI of the peptides is determined; this is valuable information that is not delivered by approaches analyzing peptides by SCX and RP (13); (ii) proteins are available after isoelectric focusing in a liquid phase offering the possibility of an automated interfacing with an the LC system; (iii) the method works in a microscale format with fraction volumes of 150 μl or less; and (iv) commercially available IPG gel strips can be used.

The use of IEF for peptide separation offers the possibility to exploit experimentally derived pI values to increase the reliability of peptide identification procedures (3, 4). We demonstrate the feasibility of such an approach for the combination of peptide fractionation by OFFGEL electrophoresis and LC/MS data analysis with the Spectrum Mill software. The additional use of a pI filter allowed attenuating the stringency of the peptide validation criteria without increasing the false positive rate. However, this resulted in only a 3.6% increase in the number of validated spectra (Fig. 5). A further increase could be obtained by combining the pI filter with a filter relying on another experimentally determined peptide characteristic, e.g. the retention time on a RP column (7).

OFFGEL electrophoresis is a flexible technique for the fractionation of peptides or proteins that can play a role in various steps of multidimensional separation schemes. The feasibility of combining protein OFFGEL electrophoresis with peptide OFFGEL electrophoresis or protein OFFGEL electrophoresis using wide range IPG gel strips with protein OFFGEL electrophoresis using narrow range IPG gel strips has already been shown (6). As we have shown, peptide fractionation by OFFGEL electrophoresis with the Agilent 3100 OFFGEL Fractionator offers easy recovery of the sample combined with excellent resolution and higher confidence identifications because of the inclusion of the experimentally derived pI in the validation of peptide identifications.