The complexity and abundance of proteins in most biological systems exceed the resolution capacity of every currently existing analytical technique. Perhaps the most striking example is the serum proteome, which features extreme differences in protein concentrations, with the proteins considered the most interesting often ten orders of magnitude less abundant than albumin or immunoglobulins.
Protein analyses are commonly used to study various disease or treatment states. To obtain statistically relevant data a number of biological and experimental replicates are needed, the result is a large number of samples. Preparing the samples for these studies is tedious work and a source of error.
The demand for highly reproducible approaches is increasing as protein analysis and proteomics mature and begin to address actual biological questions, not just produce part lists. Hence, the need for convenient and reproducible sample-preparation methodologies to reduce sample complexity is larger than ever.
There are a number of ways in which a proteome can be fractionated dependent upon which protein properties are of interest. In essence, the purpose of the sample-preparation procedure is to make the biological sample, e.g., blood, plasma, tissue, urine, cell culture, plant extract, or bacteria, manageable enough to enable an informative and, if possible, exhaustive characterization of the protein(s) of interest.
Affinity- and immuno-capture based methods are well-established to enrich subsets or individual proteins of interest. With these approaches, sample complexity can be reduced to such a large extent that a simple one-dimensional separation procedure, e.g., electrophoresis or liquid chromatography (LC), may be sufficient to resolve captured protein constituents. The typical end point of a proteomic analysis is identification by mass spectrometry (MS).
Parallel Immuno-capture Preparation
We have applied the use of a well-documented separation media, Protein G Sepharose™ High Performance, in combination with the high-throughput format of a multiwell filter plate to enable parallel immuno-capture preparation of up to 96 samples simultaneously (Figure 1).
A model system comprising human serum transferrin (hTf) spiked in a bacterial cell extract (Escherichia coli K-12) was used to determine the applicability of the approach. The amount of hTf was 7.5 µg/mL in the start material, with 5 mg/mL E.coli proteins representing background protein contaminants.
To enable rapid downstream visualization and protein quantitation, hTf was labeled with CyDye™ DIGE Fluor Cy™5 minimal dye prior to mixing. The specific purity of hTf in the start material (0.15% w/w) was chosen to mimic the approximate level of a medium-abundant protein. Rabbit polyclonal a-hTf antibodies were immobilized on 50 µL of predispensed Protein G Sepharose HP in the wells of a 96-well filter plate, and the antibody-protein G complexes were subsequently cross-linked using dimethyl pimelimidate dihydrochloride.
In each well, 200 µL (equal to 1.5-µg hTf) of the sample diluted in TBS was added followed by incubation for target protein capture and washing with TBS and 2 M urea to further enrich the level of target protein relative to background proteins. Bound proteins were eluted in three 200-µL fractions using 0.1 M glycine, pH 3.0, containing 2 M urea, and the eluted fractions were subsequently neutralized with a small volume of 1 M Tris-HCl, pH 8.5.
As an initial analysis, aliquots of the eluted fractions were separated by SDS-PAGE and the gel was stained with Deep Purple Total Protein Stain and scanned directly in the Ettan™ DIGE Imager using excitation and emission wavelengths specific for Cy5 and Deep Purple, respectively.
The analysis of three elution fractions from six replicates originating from two different multiwell plates is shown in Figure 2. The majority of the enriched hTf was eluted in the first fraction with target protein recoveries typically above 50%, with an enrichment factor of more than 100 relative to the starting material. This procedure proved to be highly reproducible with relative standard deviations well below 10% for both target protein recovery and specific purity.
For comparison, standard reversed-phase LC-MS analyses were performed using both start material and hTf-enriched material. Start material was diluted with the elution buffer to a suitable protein concentration and thereafter treated the same way as the enriched sample.
Tricarboxyethyl-phosphine was added to reduce the disulfide bonds of the proteins and cysteines were alkylated with iodoacetamide. After alkylation, the proteins were cleaved into smaller fragments by addition of stabilized porcine trypsin and incubated overnight at room temperature. 10 µL of each sample was injected on a C18 enrichment column and desalted on-line using the Ettan MDLC chromatography system.
Bound peptides were then separated on an analytical C18 reversed-phase column (0.075 x 150 mm) with a gradient from 0 to 67% (v/v) acetonitrile in 0.1% formic acid and water during 60 minutes at a flow rate of 200 nL/minute. The effluent was sprayed into the nanoflow electrospray source of an ion trap mass spectrometer. An automatic data-dependent scan method was used to acquire MS and MS/MS spectra.
An automated protein database search completed the identification of the peptide fragments as well as the overall identification of proteins. Human and E.coli databases were used in the search, allowing for the two modifications—oxidized methionines and carboxyamido methylated cysteines.
About 50 different proteins were detected and identified with confidence p<0.01 in the start material (hTf spiked in E.coli extract). These proteins were mainly high-abundant E.coli proteins, including proteins involved in the protein synthesis machinery, metabolic enzymes, and various heat shock proteins among others (Figure 3). They were generally identified by only one or two peptide fragments. hTf was not detected in the start material.
In contrast, the identification of proteins in the enriched sample showed hTf as the major protein hit with high confidence, represented by 48 unique hTf-derived peptide fragments covering 70% of the precursor sequence (Figure 3). Spectra were obtained with signal strength that permitted detailed information extraction.
The remaining proteins identified were either ribosomal proteins or proteins closely associated with the ribosomal complex. Notably, the yield in this particular experiment was high enough to allow the use of only 5% of the first eluted fraction, indicating that protein(s) with significantly lower abundance than reviewed in this study may well be analyzed by the same protocol.
The described immuno-based sample-preparation method proved to be convenient and highly reproducible, generating an enriched sample able to be analyzed with a simple SDS-PAGE or LC-MS approach. The results demonstrate the strength of reducing sample complexity prior to analysis. In addition, the yield of the enriched protein from the model system used in this study was sufficient to enable an extensive MS analysis. Finally, the multiwell format ensured fast and reliable simultaneous capture of proteins from a large number of complex samples.