In-gel protein digestion

This protocol describes a method for the direct digestion of proteins in Coomassie- or colloidal blue-stained polyacrylamide gels. The absence of detergents and a minimal use of salt allow the direct analysis of the resulting digest by nanoLC-tandem mass spectrometry without an additional desalting step. In this study the protein digestion was performed by Trypsin, an enzyme that cleaves specifically after Lysine and Arginine except if the following amino acid is Proline.

The colloidal blue-stained gel was placed on a clean glass plate and the gel bands of interest were excised out of the gel using a clean scalpel. Each band was cut into the smallest pieces possible that were placed in an Eppendorf tube. The bands were destained first with 50%

ACN for 10 min and second with destaining solution (100 mM ABC, 30% ACN) for 10 min.

The last step was repeated if necessary until the acrylamide pieces were completely colorless.

The protein reduction/alkylation step was skipped as the sample was already reduced and alkylated prior to SDS PAGE. The bands were then dehydrated with 100% ACN for 15 min after which the bands were dried in a Speed Vac for 10 min without heating. The dried gel pieces were re-swelled in 50 μl tryptic digestion buffer (40 mM ABC containing 10 ng/μl Trypsin (Promega)) for 30 min. Additional buffer (5 mM ABC) was then added to the gel pieces if they were not completely covered with buffer. Digestion was carried out overnight at room temperature. On the following day, the supernatant was first collected and additional 5 mM ABC (50 μl) was added to the gel pieces for 15 min at 37 ^°C. After collection of the supernatant, the gel pieces were further extracted with 300 μl 100% ACN for 10 min at room temperature, after which the supernatant was again collected. The gel pieces were then extracted with 50 μl of 5% Acetic acid, 5% ACN at 37 ^°C for 15 min and the supernatant was collected. Finally, the gel pieces were extracted again with 300 μl 100% ACN at room temperature for 15 min. The above procedure has been adapted from several papers (86-88).

The pooled supernatant was then dried in a speed Vac without heating and the peptide extract was kept at -80 ^°C until further use.

3.2.7 Mass spectrometry

3.2.7.1 Packing of NanoLC columns

An analytical fused silica emitter (75/360 μm i.d./o.d., tip diameter 8±1 μm, New Objectives) was packed with 12-15 cm of 3 μm Reprosil C18 A.Q. reverse phase material (Dr. Maisch).

The resin was mixed with methanol (approximately 50 mg/ml) and the fused silica was packed at 170-200 bar using pressurized Nitrogen. The column was then rinsed with pure methanol at the same pressure and kept dried until use.

3.2.7.2 Method development for NanoLC ESI-MS/MS

Peptide samples were analyzed by nano liquid chromatography electrospray tandem mass spectrometry using an Ultimate 3000 nanoflow chromatographic system (Dionex) coupled to a LTQ-Orbitrap tandem mass spectrometer (Thermo Electron) equipped with a nanoelectrospray ion source (Proxeon Biosystems)

The peptide mixture of each sample was dissolved in 20 μl of buffer A (2% ACN, 0.5% acetic acid). 10 μl of the sample was transferred to a glass vial of 0.25 ml volume (sun – sri) of which 5 μl were injected into the system at a flow rate of 450 nl/min at 100 % buffer A for 12

min. After loading, the flow was decreased to 250 nl/min and peptides were eluted from the reverse phase column as follows: 12—14 min, 0—5% buffer B (80% ACN, 0.5% acetic acid);

14—30 min, 5—30% buffer B using the curve 4 (slightly concave) of the Chromoleon software (Dionex); 30—90 min, 30—55% Buffer B using the curve 6 (slightly convex) of the Chromeleon software. The column was then washed for 15 min with 100 % buffer B at 350 nl/min, after which it was re-equilibrated for 25 min in 100% buffer A at 350 nl/min.

Peptides were analyzed by tandem mass spectrometry using standard operating parameters as follows: the electrospray voltage was set to 2.2 kV and the ion transfer capillary temperature was at 170 ^°C. Survey scans (scanning range m/z 400-1500) were recorded in the Orbitrap mass analyzer at a resolution of 30,000 with the lock mass option (20) enabled. Data-dependent MS/MS spectra of the five most abundant ions from the survey scan were recorded in the LTQ ion trap using a normalized collision energy of 32% for MS/MS (30 ms activation, q=0.25) and a selection threshold of 500. Target ions selected for MS/MS were dynamically excluded for 30 sec.

3.2.7.3 Data Processing Method and Protein Identification

The raw data files of each LC-MS/MS run were processed using the SEQUEST search algorithm 27 (SEQUEST version 27.0, revision 12, Thermo Electron). Searches were performed against the in house-generated MouseGP database (Version January 2007, genome version mm8/NCBI36, 60862 entries). The MOUSEGP database is generated by taking the most recent version of all the non-redundant sets of Swissprot and Trembl sequences from human, mouse, rat, other vertebrates, drosophila, C. elegans and Yeast and blast the sequences against the mouse chromosomes. The putative exons are then assembled into genes.

The alignments between genomic chromosome sequence and proteins are refined with GeneWise (89, 90), a tool that finds splice sites and corrects frame shifts. Data were searched with a mass tolerance of +/-5 ppm for parent ions and +/-1.0 Da for fragment ions.

Methionines (reduced/oxidized; +15.9949 Da) were considered as differential modifications while cysteines were considered as fully carbamidomethylated (+57.0199 Da). Only fully tryptic peptides with no more than one miscleavage were considered for data analysis.

Peptides were considered as unambiguously identified if their XCorr scores (91) exceeded 1.5, 2.0 or 2.5 for singly, doubly and triply charged ions, respectively, and if the corresponding ΔCn scores (the normalized Sequest XCorr score difference between the first and the second best peptide match,(91)) were larger than 0.2. In this study, only proteins for which at least two different peptides were identified were considered as successfully identified. The False

Positive Discovery Rate at the peptide level was estimated by searching the raw data files against a shifted database. This database, which is generated from MOUSEGP, has an identical number of proteins and tryptic peptides by keeping fixed the position of the arginines and lysines while the position of all remaining amino acids were left-shifted by two positions. The raw files were searched by using the above searching criteria and the raw spectra were submitted to the scrambled database for scoring. These spectra were binned based on their score and the number of spectra in each bin was counted and stored in the database. An objective estimate of the false positive discovery rate at the peptide level is obtained by comparing the distribution and the counts of spectra for a given score for spectra that were scored against MOUSEGP and the scrambled database (52, 92, 93). Using the criteria mentioned above, the false positive discovery rate for 1+, 2+ and 3+ charged peptides was estimated to be 7.4%, 3.2%, and 0.6%, respectively.

Comparison of the LC-MS raw data files based on the total ion current of the peptides was performed by using the Genedata RefinerMS (version 4.5) software suite. Software-specific settings were as follows: baseline subtraction (20% quantile value, m/z Window 10 Da, RT window 0.1 min); chromatogram retention time (RT) alignment (prior internal peak identification on m/z window with 5 points, RT window 0.5 min, gap penalty 0.75, RT search interval 100 scans); peak identification as in RT alignment; peak shaping (multiplicity: 67%);

isotopic clustering (minimal charge 2, maximal charge 4, correlation threshold 0.5, maximal missing peaks 1, mass tolerance 0.05, ionization: protonation; mass consistency: peptides).

The resulting peak table associates the m/z and RT of a feature with its extracted ion count (XIC) and signal-to-noise ratio (S/N) measured in each sample. Features with a median S/N<5.0 were excluded from the data analysis as these outliers displayed broad ranges of nonlinear chromatographic shifts across samples.

3.2.7.4 Sequence and topology analysis

The proteins’ cellular location and topology were analyzed with a variety of tools as follows.

Prediction for signal peptide was performed using the in-house “signal_anchor” software tool (http://bioinfo.bas.roche.com:8080/sawicgi/sawi.cgi?signal_anchor) using an algorithm inspired by the web-based SignalPep tool (94) but using a support vector machine prediction model using Roche internal data rather than a neural net prediction model.

Membrane protein prediction was performed using either the “ALOM” software tool (http://bioinfo.bas.roche.com:8080/sawicgi/sawi.cgi?alom) as described by Klein P et al. (95),

or the web-based “TMHMM 2.0” (http://www.cbs.dtu.dk/services/TMHMM-2.0/), as described by Krogh A. et al. (96).

The version of the GO annotations (as described in Gene Ontology Consortium (2000)) used in this work was from June 2008. In this work, the multiple GO annotations of a protein were reduced to what was believed to be the most relevant entry according to the following rules:

a) plasma membrane > ER, GOLGI, endosome > lysosme, microsome, peroxisome>

mitochondria > cytoplasm, cytosol > nucleus > other

b) integral > anchored > peripheral, associated to > no mention of membrane interaction c) any membrane interaction > proteasome, cytoskeleton, vesicle, etc. >extracellular In doubt, the protein annotation of the SwissProt database was used to direct the selection to one or the other direction.