Proteome coverage: SDS-PAGE versus HPLC

Complex proteomes

The analysis of the proteomes of complex systems by the so-called bottom-up method, which looks at the peptide fragments released from the intact proteins, depends upon good fractionation to separate those fragments for mass spectrometric analysis. It is not sufficient to put your faith in the power of mass spectrometry because the peptide mixtures are too complex. Even with the combined effort of liquid chromatography/mass spectrometry, many peptides, especially low abundant ones, can remain undetected.

The solution is to introduce prefractionation before the LC/MS step, which creates smaller bundles of proteins or the peptides that are produced by their enzymatic digestion. These are easier to analyse simply because there are fewer components present in each fraction. Two of the most common options are SDS-PAGE for intact proteins and reversed-phase HPLC at high-pH for peptides, but Pengyuan Yang and colleagues from Fudan University in Shanghai noticed that there has never been a comprehensive comparison of the two.

So, they undertook the task, using human liver carcinoma HpeG2 cells as a model system. The proteins were removed from cultured cells before being subjected to the two fractionation procedures with a number of variations in each to see what effect they had on the degree of proteome coverage.

Protein prefractionation

For SDS-PAGE, the initial protein mixture from the cells was separated and the proteins were visualised on the gel by staining with the dye known as Coomassie Brilliant Blue G-250. Then the gels were cut into ten sections and combined in pairs to give five mixtures that were treated with the enzyme trypsin to digest the proteins for analysis by ultrahigh pressure LC-tandem-MS using C₁₈ columns that were 15 or 50 cm long.

Alternatively, the gel was divided into 20 fractions that were each digested with trypsin before grouping them into five fractions for LC/MS analysis on the 50-cm column or leaving them as 20 individual fractions for analysis on the 15-cm column.

For prefractionation by HPLC, the protein mixture from the cells was digested with two enzymes in turn, Lys-C then trypsin, and the resulting peptide mixtures were separated on a C₁₈ column under high-pH conditions at pH 10.0. The eluting peptides were collected as 40 fractions before being combined into 5 or 12 fractions for LC/MS analysis on the 50-cm column or as 20 fractions for analysis on the 15-cm column.

All of the LC/MS analyses were carried out in low-pH conditions in acidic solution and the peptides were detected by high-resolution mass spectrometry. The fragment ions in the tandem mass spectra were searched against a Homo sapiens database to identify the proteins from which they originated.

Increasing the number of identified proteins

Following SDS-PAGE, the number of unique peptides and proteins that were identified from the five combined fractions increased about 30% when replacing the 15-cm column with the 50-cm one, leading to a total of almost 4000 proteins. This figure was improved to 4604 proteins when the gel was sliced into 20 pieces that were combined into six fractions.

However, the best performance in SDS-PAGE came from analysing the 20 individual fractions by LC/MS leading to 7933 proteins in total. This good figure was achieved with the 15-cm column, showing that the increased number of fractions more than compensated for the quicker and poorer separation compared with the 50-cm column.

In the high-pH RPLC separation, the set of five fractions led to the identification of nearly 5000 proteins but this was only about 20% greater than achieved with 1D RPLC, so was not regarded as a notable difference. Increasing the number of fractions to 12 and 20 each produced more than 6000 proteins but the analysis with the shorter column was preferred due to the shorter analysis time.

The proteins from SDS-PAGE and RPLC were not the same due to the different mechanisms of separation but there was about 80% overlap between them. SDS-PAGE uniquely identified 399 but high-pH RPLC identified 1951 uniquely. So, for proteome discovery experiments, the combination of high-pH RPLC with LC/MS at low pH is the recommended option.

One consequence to be aware of in both methods was the occurrence of unexpected structural modifications to the peptides brought about by the experimental conditions. There were 277 different types in total, 79 from high-PH RPLC, 57 from SDS-PAGE and 141 from both methods. They only occurred in a small proportion of the spectra but must be borne in mind during large-scale studies.

With either method, the performance improved as the number of fractions analysed was increased but the high-pH RPLC method provided the greatest number of proteins. Ideally, given the fact that each method identified some proteins that the other missed, the two procedures should be used in combination but the constraints of time and cost mean that this is unlikely to occur in most organisations.

Follow us on Twitter!