Robust, Quantitative, Label‐free Phosphoproteomics Platform: Automated Enrichment and 2D‐LC/MS/MS
Brenna M. Richardson, Erik J. Soderblom, J. Will Thompson, M. Arthur Moseley
Duke Proteomics Core Facility, Institute for Genome Sciences & Policy, Duke University School of Medicine, Durham, NC
Improvements in Enrichment of Phosphopeptides with Automated, Capillary Column Format
ABSTRACT
o Traditionally, quantitative phosphoproteomics strategies were limited to the use of stable‐isotope
labeled cell cultures, which is still the most commonly used sample type. Recent developments have been
made using label‐free techniques which make the analysis applicable to biological fluids and tissues,
matrices which are not amenable to SILAC‐based approaches. Here, we describe and automated capillary‐
based enrichment strategy that achieves the robust quantitative reproducibility essential for the success of
label‐free workflows.
o For the phosphopeptide enrichment column, a fused‐silica capillary was packed with titania particles to
the appropriate dimensions for the quantity of digested protein to be enriched. The column was placed
in‐line with a Waters CapLC pump, autosampler, and UV detector. Enrichment of phosphopeptides was
performed through a series of injections of sample, wash buffer, and elution buffer. Quantitative
reproducibility of the procedures was determined using area‐under‐the curve measurements after
accurate mass/retention time alignment and robust mean normalization of identified phosphopeptides
from four replicated enrichment analyzed by LC/MS/MS on a Waters Synapt G1. The average CV of the
516 identified phosphopeptides was 8.7%, which a quantitative enrichment specificity of ~80%.
o In addition to the enrichment, the described platform also includes improvements to the LC/MS/MS
analysis of enriched samples. This strategy employs the high/low pH separation of phosphopeptides that
has been recently used to increase the coverage of un‐enriched samples in a reasonable time frame. By
adjusting the fractionation to compensate for the more acidic nature of phosphopeptides, a 3‐fraction
2DLC/MS/MS method using a Waters nanoAcquity coupled to a Synapt G2 was implemented to increase
the total number of spectra assigned to phosphopeptides in a rat brain lysate from 472 to 932 over a
single dimension analysis while only increasing the total analysis time by 70%. Additionally, 97% of
phosphopeptides were identified in a single 2D fraction, highlighting the orthogonality of the separation.
Determination of MeCN percentage for Fraction Cuts from First Dimension Column
Evaluation of Enrichment Flow Rate and Identification of Phosphopeptides Throughout Each Stage of Enrichment
¾ Due to the more acidic nature of phosphopeptides, the majority are identified in the analysis of MeCN cuts less that 20%. For this reason, a gentler series of MeCN steps was used for the 3‐fraction analysis
Fraction 1: 4.7% MeCN
230 Phosphopeptides
42 Non‐phos peptides
84.6% qualitative specificity
¾ In addition to the high level of specificity found in the Elution fractions, the absence of phosphopeptide signal in the Re‐enriched Flowthrough highlights the completeness off the enrichment in a single pass.
Effect of Modifier on LC/MS Peak Shape and Reproducibility
Sample Preparation
Automated Enrichment
Time/sample
120 minutes
Hands‐on time
60 min/sample set
• Pre‐Elute and condition
• 100 µL Elution Buffer, 100 µL Wash 2 buffer
• 1 x 23 min
• Load sample
• 1 x 10 minutes
• Wash 1 and Wash 2
• 1 x 35 minutes
• Elute
• 2 x 16 minutes
Post‐
enrichment Sample Processing
•Acidify samples to pH < 3 with neat formic acid
• Formic acid added to fraction collection vials prior to enrichment
• Dry down phosphopeptides
• Reconstitute in LC loading buffer
An initial 10‐fraction 2DLC method that is optimized for unenriched cell lysates was performed to assist in identification of ideal MeCN percentages to use in a 3‐fraction analysis.
Experiment Details
• Replicate enrichments of 750 ug rat brain lysate were performed at both 5 and 10 uL/min keeping all volumes constant
• Fractions were collected throughout the enrichment
• Flow through from sample injection
• Wash with glycolic acid
• Wash without glycolic acid
• Elution
• Flowthrough fraction was concentrated and re‐enriched a second time. The elution fraction from this was analyzed as the ‘re‐enriched flowthrough’
Phosphopeptide Enrichment Workflow
• Solubilize protein from tissue or cells using protocol specific to sample type
• Digest overnight with trypsin
• Dry down peptides
• Reconstitute in Loading buffer
Improvements in LC/MS Analysis of Complex Phosphopeptide Mixtures
Intensity Ratio Plot of Phosphopeptides
GlyAcid‐enriched vs. DHB‐enriched
Difference in LC/MS peak width
of 100 phosphopeptides
Fraction 2: 9.4% MeCN
211 Phosphopeptides
54 Non‐phos peptides
79.6% qualitative specificity
Resulting Quantitative CVs of Identified Phosphopeptides
Wider in DHB‐
enriched sample
Fraction 3: 30.0% MeCN
143 Phosphopeptides
78 Non‐phos peptides
64.7% qualitative specificity
Wider in GlyAcid‐
enriched sample
Buffers:
Equilibration and Wash 2:
Loading and Wash 1:
Elution:
LC loading buffer:
80%ACN, 1%TFA
80%ACN, 1%TFA, 50mg/mL 2,5‐dihydroxybenzoic acid (DHB)
20%ACN, 5% aqueous ammonia
200mM ammonium formate, pH 10, 10mM citrate, 33% aqueous ammonia
¾ While there is a slight decrease in the number of phosphopeptides identified in fractions 2 and 3, they a more evenly distributed across fractions as opposed to what was seen in the 10‐fraction analysis.
¾ Overall, the quantitation of phosphopeptides between the two enrichments are similar as shown in the ratio plot. However, the peak widths are significantly broader in the DHB‐enriched samples, which results in a greater number of phosphopeptides with larger degrees of quantitative variability.
•
•
LC: NanoAcquity UPLC System
• Column: 75um x 150mm BEH300 C18, 1.7 um particle size
• Gradient: 3‐30%ACN in 90 minutes
MS: Waters Synapt G1 Mass Spectrometer
• Operated in DDA mode, submitting up to 3 ions from a single survey scan to MS/MS
MS: Thermo Orbitrap XL
• Operated in DDA mode, submitting up to 5 ions from a single survey scan to MS/MS
LC and MS Experimental Details for 2D‐LC of Phosphopeptide‐Enriched samples
•
•
LC: Waters 2D NanoAcquity UPLC System using
• 1st Dimension: pH 10 Ammonium Formate
• 2nd Dimension: pH 3 Formic Acid
• 3 Fractions were analyzed by 30 minute gradients in the second dimension after
eluting phosphopeptides off 1st dim. column using fraction cuts of acetonitrile at
4.7%, 9.4% and 30% MeCN.
MS: Waters Synapt G2 Mass Spectrometer
• Operation in sensitivity mode and DDA mode, submitting up to 3 ions from a single
survey scan to MS/MS
225
Experiment Details
• Spin Column
• Three separate enrichments of 200 µg zebrafish lysate, each analyzed 3 times by LC‐MS
• Capillary Column (Soderblom, et al.)
• Four separate enrichments of 1000 µg rat brain lysate, each analyzed once by LC‐MS
Median
%CV
Spin Column
19.4
17.2
25.4
Standard Peptide/Protein Identification and Relative Quantitation Workflow
Capillary Column
8.2
5.7
10.7
•
Spin Column reference: Soderblom, EJ; Philipp, M; Thompson, JW; Caron, MG; Moseley, MA. (2011) Quantitative label-free
phosphoproteomics strategy for multifaceted experimental designs. Analytical Chemistry, 83( 10), 3758-3764
•
•
Fraction 2
211 total
202
Improved Reproducibility with Automated Enrichment Setup
Average %CV
Peptide Identification was performed by searching DDA runs against a 1x reversed species‐specific database
• Zebrafish samples: NCBI Danio Rerio (for spin column comparison)
• Rat Brain samples: NCBI Rattus
Quantitation was performed within Rosetta Elucidator following Peak Teller retention time alignment and
intensity scaling where appropriate
Peptide intensities were reported as the sum of all features belonging to peptides above a Mascot Ion Score
that resulted in a 1% spectral FDR determined by hits to reversed sequences.
2
Fraction 1
230 total
LC and MS Experimental Details for Phosphopeptide Enrichment Evaluation
•
Overlap of Phosphopeptides Identified by Fraction
Format
80% peptides below
3
7
133
Fraction 3
143 total
¾ The limited overlap between phosphopeptides identified in each LC fraction highlights the orthogonality of the 2D approach utilizing pH 10 for the first dimension separation and pH 3 for the second dimension
Overlap of Phosphopeptides Identified in Single 2D LC/MS vs. Singe 1D LC/MS Analysis
¾ Overall technical variation was approximately cut by more than half by using an automated enrichment setup and Glycolic Acid to inhibit non‐specific binding.
1D LC/MS
394 total
www.genome.duke.edu/proteomics/
B02 Levine Sciences Research Center
194
200
372
2D LC/MS
572 total
¾ The somewhat limited overlap between the two analysis types is due, in part, to the serendipitous nature of data‐dependent acquisition.