A rugged, reproducible, multi-dimensional
LC-MS system was developed to identify and characterize
proteins involved in protein-protein interactions and/or
protein complexes. This system employs SCX in the first
dimension and RP in the second dimension. It is fully
automated to avoid sample handling and robust enough to
handle direct injections of samples containing 2M urea.
The data are subjected to a streamlined post analysis
results comparison, which further ads to the overall system
efficiency. In order to evaluate the performance and reproducibility
of this multi-dimensional LC-MS system, peptides obtained
from sequential yeast extracts were used as a model system.
S. cerevisiae strain BY4741 was grown to mid-log phase
(OD595 = 1.0) in YPD broth at 30oC. Five grams of cells
were solubilized in lysis buffer and proteins were extracted
in a modified three-step differential extraction protocol
without the use of detergents. The proteins were denatured,
reduced, alkylated and digested with endoproteinase lys-C
followed by trypsin. The resulting peptides were analyzed
by a fully automated 2D-LC-ESI-MS/MS system built from
Shimadzu LC-VP Series components and connected directly
to a ThermoFinnigan LCQ Classic ion trap mass spectrometer.
Protein identification was obtained by submitting the
MS/MS data to Mascot. Mascot results were then parsed
into a MYSQL relational database and compared in html
output reports using DBParser.
Initial experiments, conducted on the automated 2D-LC-MS
system using standard protein digests, demonstrated
good retention time reproducibility (1-2% peak RSD from
the reconstructed ion chromatogram) and improved resolution
compared with its 1D-LC-MS counterpart. Yeast extract
1 was used to determine the optimal loading amount needed
to obtain the best resolution and the largest number
of peptide identifications for such a complex peptide
mixture. On average 1,400 peptides, corresponding to
~450 proteins, were detected in a 10ug sample from this
extract. Combining the results of all three yeast extracts
resulted in ~800 proteins identified. Finally, the streamlined
nature of the data analysis and results comparison using
DBParser made this entire project much easier and more
efficient than hand curation. As an example, a comparison
of two yeast files (10,000 .dta files each) from Mascot
required only 5 minutes to sort into lists identifying
proteins unique to each analysis. |
