Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass WorkflowMaryann S. Vogelsang1, David A. Sarracino1, Janin Schulte2, Bryan Krastins1, Michael Athanas3, Amol Prakash1, Alejandra Garces1, Gouri Vadali1,3, Shadab Ahmad1, Gaiané Demirdjian2, Jennifer N. Sutton1, Heinz Redl4, Joachim Struck2, Bruno Darbouret2, and Mary F. Lopez1

1Thermo Fisher Scientific, Cambridge, MA, USA; 2Thermo Fisher Scientific, Hennigsdorf, Germany; 3VAST Scientific, Cambridge, MA, USA; 4Ludwig Boltzmann Institute, Vienna, Austria

2 Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow

Design and Optimization

•Robust, commercially available nanoflow LC

•Commercially available columns

•Focus on stable spray•Focus on high

reproducibility of peak intensities, CV<8%

Pass 1: Quantification

•Chromatographic alignment•Uncompromised full scan

measurements •Each sample is measured once

– no need for technical replicates

•Internal peptide standards (normalization)

•Triplicate runs of peptide standards every 12 runs

(instrument QC)•“Top10” data dependent

acquisition• Stringent precursor ion

selection criteria

Pass 2: Identification

•Targeted fragmentation by inclusion list

•Relaxed precursor ion selection criteria

•Not all samples measured –subset as determined from SIEVE

analysis•Internal peptide standards

•Marker stratification using multi-marker and single ROC AUC•Export to pathway analysis

methods Inclusion list

Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow Maryann S Vogelsang1, David A Sarracino1, Janin Schulte2, Bryan Krastins1, Michael Athanas3, Amol Prakash1, Alejandra Garces1, Gouri Vadali1,3, Shadab Ahmad1,Gaiané Demirdjian2, Jennifer N Sutton1, Heinz Redl4, Joachim Struck2, Bruno Darbouret2, Mary F Lopez1

1Thermo Fisher Scientific, Cambridge, MA; USA 2Thermo Fisher Scientific, Hennigsdorf, Germany; 3VAST Scientific, Cambridge, MA, USA; 4Ludwig Boltzmann Institute, Vienna, Austria

Conclusion The Two-Pass workflow was successfully applied to multiple cohorts of both

primate and human plasma research samples.

The Two-Pass workflow accelerates the time to targeted assays and validation of potential biomarkers.

The Two-Pass workflow delivers increased confidence in protein biomarkers through increased peptide coverage of differentially expressed peptides spanning nine orders of magnitude of protein abundance.

References 1. Athanas, M., MacCoss, MJ., Prakash, A., Kall, L., Tomazella, D., Maclean, B.,

Rezai, T., Krastins, B., Sarracino, D., Garces, A., Fortune, S., and Lopez, MF. (2009) Label-free Differential Analysis: An Iterative Approach to Increased Coverage, Improved Statistics and Results. Poster presentation at ASMS.

2. Athanas, M., Sarracino, D., Rezai, T., Prakash, A., Sutton, J., Krastins, B., Ning, M., and Lopez, MF. (2010) A Two-pass Informatics-driven Label-free Workflow For Discovery Of Neurovascular Mediators In PFO-Related Stroke. Poster presentation at ASMS.

3. Lopez, MF., Kuppusamy, R., Sarracino, DA., Prakash, A., Athanas, M., Krastins, B., Rezai, T., Sutton, JN., Peterman, S., and Nicolaides, K. Mass Spectrometric Discovery and Selective Reaction Monitoring (SRM) of Putative Protein Biomarker Candidates in First Trimester Trisomy 21 Maternal Serum. J Proteome Res. 2011 10(1):133-42.

4. Lopez, MF., Sarracino DA., Prakash A., Athanas M., Krastins, B., Rezai, T., Sutton JN., Peterman S., GvozdyakO., Chou S., Lo E., Buonanno F., and Ning MM. Discrimination of ischemic and hemorrhagic strokes using a multiplexed, mass spec-based assay for serum apolipoproteins coupled to multi-marker ROC algorithm. Proteomics Clinical Applications 2012 6: 190-200.

5. Anderson, NL, and Anderson, NG. The Human Plasma Proteome: History, Character, and Diagnostic Prospects. Mol. Cell. Proteomics 2002 1, 845-867.

Overview Purpose: Sample preparation relying on fractionation to simplify the complexity and large dynamic range of plasma or serum samples does so at the potential cost of inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery using a “Two-Pass” workflow that reduces the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Methods: Two-Pass discovery workflow using high-resolution LC-MS/MS coupled to ROC and differential expression analyses from stratified human sample cohorts and non-human animal models.

Results: The Two-Pass workflow reduced the number of replicates needed. Compared to a single-pass experiment, the workflow provided 20–50% more quantitatively associated protein identifications in a shorter time period (2–10X).

Introduction Proteomic-based biomarker discovery approaches have primarily focused on direct profiling of serum or plasma to determine abundance changes that can be used to discriminate between populations. Due to the complexity and large dynamic range of serum and plasma, samples are often processed by depletion and/or fractionation in order to reduce their complexity. However, these manipulations can result in inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery the uses a Two-Pass workflow to reduce the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Previously, we introduced a two-pass workflow exploiting the mass spectrometer’s accurate mass and broad dynamic range, which investigated the uncompromised quantitative data in Pass 1, and then targeted differentially expressed MS features in Pass 2 1,2,3,4. In Pass 1, we take advantage of the speed of the hybrid mass spectrometer to measure quantitative MS1 frames while concurrently measuring and identifying the Top-10 data-dependent MS2. In Pass 2 we identify additional differentially expressed peptides from the inclusion list built from Pass-1 data analysis.

Methods Sample Preparation

This poster presents results from both primate and human plasma samples. The Two-Pass workflow was applied to 64 plasma samples of microbial-infected primates (monitoring eight subjects over a 24 hr time period, two of which were controls). Additionally, a small cohort of IRB approved human plasma samples were analyzed. The preliminary results in this report are intended to demonstrate the Two-Pass workflow, not biomarker discovery results.

Liquid Chromatography and High-Resolution Mass Spectrometry

As shown in Figure 1, the Two-Pass workflow strategy consists of the separate optimization of MS parameters and configuration for protein quantification and identification.

Pass 1. Plasma samples (300 ng of primate samples or 500 ng of human samples), were digested with trypsin and injected onto a Thermo Scientific EASY-nLC system configured with a 10 cm x 100 µm trap column and a 25 cm x 100 µm ID resolving column. The sample load was optimized for optimum quantification, (i.e. full scan data). Buffer A was 98% water, 2% methanol, and 0.2% formic acid. Buffer B was 10% water, 10% isopropanol, 80% acetonitrile, and 0.2% formic acid. Samples were loaded at 4 µL/min for 10 min, and a gradient from 0–45% B at 375 nL/min was run over 130 min, for a total run time of 150 min (including regeneration and sample loading). The Thermo Scientific LTQ Orbitrap Velos hybrid ion trap-Orbitrap mass spectrometer was run in a standard Top-10 data-dependent configuration, except that a higher trigger-threshold (20K) was used to ensure that the MS2 did not interfere with the full-scan duty cycle. This ensured optimal full-scan data for quantification. MS2 fragmentation and analysis were performed in the ion trap mass analyzer.

TABLE 1. High-throughput is an important advantage of the Two-Pass workflow. For the sample preparation and LC-MS/MS steps, the workflow in this example has a ten-fold time advantage over fractionation sample preparation approaches. In comparable timeframe we can investigate ten times the number of human samples. Thus, the Two-Pass workflow offers significantly improved sample statistics.

Results Time-Saving Two-Pass Workflow

The Two-Pass workflow was applied to both sample sets. Our previous findings have demonstrated that the Two-Pass workflow can accurately detect, quantify and identify unlabeled differentially-expressed proteins within plasma samples.1,2,3,4 Given the robust chromatography, high-resolution mass spectrometers and computing power of SIEVE 2.0 software, the workflow enables identification of differentially expressed proteins in a single-pass experiment that is two- to ten-times faster than other approaches (Table 1).

SEQUEST is a registered trademark of the University of Washington. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

FIGURE 2. Pass 1 acquisition cycle.

Figure 3 shows the gel view of the frames obtained from nine LC-MS runs In this example, we have 20,000 potentially useful frames in the gel view. A frame represents a potentially interesting feature (peptide) found in a collective data set. Using frame filters, we can separate the differentially expressed frames between the two groups (see Table 2).

SIEVE Software Analysis

The following data are representative of a proteomics differential case study with ROC analysis from an animal model for sepsis.

Table 2 shows that the Two-Pass workflow allows for confident identification of total protein as well as unlabeled differentially expressed peptides. Initial analysis steps include the generation of 20,000 frames (peptides). Using frame filters based upon ROC AUC or ratios, we easily identified frames that separate the infected from non-infected groups. Given that our Pass-1 experiments are simultaneously collecting Top-10 data-dependent MS2 spectra, we can easily identify the abundant proteins (FDR=1%). The unidentified frames can then be exported to an inclusion list for Pass-2 analysis.

Proteomics data analysis was performed using Thermo Scientific SIEVE software version 2.0 that features chromatographic alignment, framing, differential ROC and ratio analyses. Both Top-10 data-dependent scans and full-scan data were analyzed with the SIEVETM software using chromatographic alignment followed by feature extraction using unsupervised statistical techniques including isotope deconvolution. ROC curves were constructed for the top marker candidates as determined by their ROC Area-Under-the-Curve (AUC). Based upon various criteria including ROC AUC, low ratios, high ratios, high abundance, and low abundance, an inclusion list was created for the best candidates. This inclusion list was used for MS2 acquisition in Pass 2.

Data Analysis

Pass 2. Masses that represent differentially expressed MS1 features were built into an inclusion list and used in Pass 2 analysis. This ensured that the instrument only acquired MS2 data for the masses in the inclusion list (not the highest intensity masses). A larger sample load was used in the Pass 2 runs (600 ng to 800 ng), allowing for higher quality MS2 spectra. Because these full-scan spectra would not be used for quantification, peak shape and intensity reproducibility were not crucial. All fragmentation analysis was done in the Orbitrap mass analyzer, using both HCD and CID. Fragmentation scans from Pass 2 were analyzed using SEQUEST® and FDR analysis to make identifications. The fragmentation search results from Pass 2, along with the quantitative information obtained from Pass 1, were also analyzed using the SIEVE software. Fragmentation scan information was assigned to frames based upon the precursor m/z and retention time.

FIGURE 1. Two-Pass discovery workflow. With robust LC-MS/MS methods and powerful SIEVE software, we are able to identify differentially-expressed proteins in unlabeled plasma samples.

Pass 1 Acquisition Cycle

Two-Pass Workflow Classic Fractionation

n= 50 50

Depletion NO 1 hr/sample

Fractionate NO 10 fractions

Digest 50 samples 500 samples

LC-MS/MS 100 hrs 1000 hrs Condition/ Filters Proteins Peptides (frames) ALL frames 269 20,000 AUC > 0.8 127 8562 Fold Change >1.5 65 3944 Identified in Pass 1 65 112 Unidentified in Pass 1 (for Pass-2 inclusion list ) 0 3832

TABLE 2. Number of proteins and peptides that meet filtering criteria in Pass 1 within a ROC analysis experiment.

Figure 4 shows a frequency histogram of scheduled fragment events per minute for Pass-2 analysis (Pass-2 Inclusion List). After the given frame filter conditions [AUC>0.8 and (NRatio>1.5 or NRatio<0.6) and PRElement<1 and goodid<1 and Charge>1 and Charge<5], 1226 frames were exported as an inclusion list for Pass-2 analysis. NOTE: These frames represent differentially expressed peptides.

Pass-2 Results

On average, we achieved an 82% success rate in MS2 acquisitions from the inclusion list. Frame parameters pending, we have had even higher success rates in other experiments.

Pass 1Pass 2

Con

trol P

lasm

a

log 1

0 con

cent

ratio

n pg

/ml

Protein Name

FIGURE 5. Plasma protein concentrations of identified proteins.

BlankPeptide

Stds

Peptide Stds

Peptide Stds

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

Figure 2 shows the method for assessing systematic errors without technical replicates. Systematic errors are assessed from triplicate acquisitions of standardized peptide samples. Internal standards are spiked in all samples. This approach eliminates the need for sample replicates and conserves valuable specimens. All samples are acquired in high-resolution full-scan and Top-10 data dependent mode, on the LTQ Orbitrap Velos mass spectrometer. Biological variance and outliers are assessed with CV, PCA and other statistical methods.

FIGURE 3. Gel view of frames from 9 LC-MS runs.

Figure 5 shows an example of proteins identified in Pass 1 and Pass 2 that have a minimum of two peptides with FDR of 1%, from ROC analysis of human samples. The identified plasma proteins from our mass spectrometry analysis were searched against a database of known plasma protein concentrations5. In Pass 1, we identify proteins that span at least nine orders of magnitude.

FIGURE 4. Frequency histogram of scheduled fragment events per minute for Pass-2 analysis

An example whisker plot of peptide coverage of a potential biomarker is shown in Figure 6. Pass 2 often results in an increased peptide coverage of the proteins identified in Pass 1. With respect to roughly 10% of frames representing one peptide of a given protein in Pass 1, we identified a second peptide for that corresponding protein in Pass 2. This result strengthens confidence in identification and quantification.

FIGURE 6. Example whisker plot of peptide coverage of a potential biomarker.

Blank run

Peptide Calibration Standards

Human Plasma Samples – Full Scan + Top 10

3Thermo Scientific Poster Note • PN63592_E 06/12S

Design and Optimization

•Robust, commercially available nanoflow LC

•Commercially available columns

•Focus on stable spray•Focus on high

reproducibility of peak intensities, CV<8%

Pass 1: Quantification

•Chromatographic alignment•Uncompromised full scan

measurements •Each sample is measured once

– no need for technical replicates

•Internal peptide standards (normalization)

•Triplicate runs of peptide standards every 12 runs

(instrument QC)•“Top10” data dependent

acquisition• Stringent precursor ion

selection criteria

Pass 2: Identification

•Targeted fragmentation by inclusion list

•Relaxed precursor ion selection criteria

•Not all samples measured –subset as determined from SIEVE

analysis•Internal peptide standards

•Marker stratification using multi-marker and single ROC AUC•Export to pathway analysis

methods Inclusion list

Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow Maryann S Vogelsang1, David A Sarracino1, Janin Schulte2, Bryan Krastins1, Michael Athanas3, Amol Prakash1, Alejandra Garces1, Gouri Vadali1,3, Shadab Ahmad1,Gaiané Demirdjian2, Jennifer N Sutton1, Heinz Redl4, Joachim Struck2, Bruno Darbouret2, Mary F Lopez1

1Thermo Fisher Scientific, Cambridge, MA; USA 2Thermo Fisher Scientific, Hennigsdorf, Germany; 3VAST Scientific, Cambridge, MA, USA; 4Ludwig Boltzmann Institute, Vienna, Austria

Conclusion The Two-Pass workflow was successfully applied to multiple cohorts of both

primate and human plasma research samples.

The Two-Pass workflow accelerates the time to targeted assays and validation of potential biomarkers.

The Two-Pass workflow delivers increased confidence in protein biomarkers through increased peptide coverage of differentially expressed peptides spanning nine orders of magnitude of protein abundance.

References 1. Athanas, M., MacCoss, MJ., Prakash, A., Kall, L., Tomazella, D., Maclean, B.,

Rezai, T., Krastins, B., Sarracino, D., Garces, A., Fortune, S., and Lopez, MF. (2009) Label-free Differential Analysis: An Iterative Approach to Increased Coverage, Improved Statistics and Results. Poster presentation at ASMS.

2. Athanas, M., Sarracino, D., Rezai, T., Prakash, A., Sutton, J., Krastins, B., Ning, M., and Lopez, MF. (2010) A Two-pass Informatics-driven Label-free Workflow For Discovery Of Neurovascular Mediators In PFO-Related Stroke. Poster presentation at ASMS.

3. Lopez, MF., Kuppusamy, R., Sarracino, DA., Prakash, A., Athanas, M., Krastins, B., Rezai, T., Sutton, JN., Peterman, S., and Nicolaides, K. Mass Spectrometric Discovery and Selective Reaction Monitoring (SRM) of Putative Protein Biomarker Candidates in First Trimester Trisomy 21 Maternal Serum. J Proteome Res. 2011 10(1):133-42.

4. Lopez, MF., Sarracino DA., Prakash A., Athanas M., Krastins, B., Rezai, T., Sutton JN., Peterman S., GvozdyakO., Chou S., Lo E., Buonanno F., and Ning MM. Discrimination of ischemic and hemorrhagic strokes using a multiplexed, mass spec-based assay for serum apolipoproteins coupled to multi-marker ROC algorithm. Proteomics Clinical Applications 2012 6: 190-200.

5. Anderson, NL, and Anderson, NG. The Human Plasma Proteome: History, Character, and Diagnostic Prospects. Mol. Cell. Proteomics 2002 1, 845-867.

Overview Purpose: Sample preparation relying on fractionation to simplify the complexity and large dynamic range of plasma or serum samples does so at the potential cost of inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery using a “Two-Pass” workflow that reduces the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Methods: Two-Pass discovery workflow using high-resolution LC-MS/MS coupled to ROC and differential expression analyses from stratified human sample cohorts and non-human animal models.

Results: The Two-Pass workflow reduced the number of replicates needed. Compared to a single-pass experiment, the workflow provided 20–50% more quantitatively associated protein identifications in a shorter time period (2–10X).

Introduction Proteomic-based biomarker discovery approaches have primarily focused on direct profiling of serum or plasma to determine abundance changes that can be used to discriminate between populations. Due to the complexity and large dynamic range of serum and plasma, samples are often processed by depletion and/or fractionation in order to reduce their complexity. However, these manipulations can result in inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery the uses a Two-Pass workflow to reduce the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Previously, we introduced a two-pass workflow exploiting the mass spectrometer’s accurate mass and broad dynamic range, which investigated the uncompromised quantitative data in Pass 1, and then targeted differentially expressed MS features in Pass 2 1,2,3,4. In Pass 1, we take advantage of the speed of the hybrid mass spectrometer to measure quantitative MS1 frames while concurrently measuring and identifying the Top-10 data-dependent MS2. In Pass 2 we identify additional differentially expressed peptides from the inclusion list built from Pass-1 data analysis.

Methods Sample Preparation

This poster presents results from both primate and human plasma samples. The Two-Pass workflow was applied to 64 plasma samples of microbial-infected primates (monitoring eight subjects over a 24 hr time period, two of which were controls). Additionally, a small cohort of IRB approved human plasma samples were analyzed. The preliminary results in this report are intended to demonstrate the Two-Pass workflow, not biomarker discovery results.

Liquid Chromatography and High-Resolution Mass Spectrometry

As shown in Figure 1, the Two-Pass workflow strategy consists of the separate optimization of MS parameters and configuration for protein quantification and identification.

Pass 1. Plasma samples (300 ng of primate samples or 500 ng of human samples), were digested with trypsin and injected onto a Thermo Scientific EASY-nLC system configured with a 10 cm x 100 µm trap column and a 25 cm x 100 µm ID resolving column. The sample load was optimized for optimum quantification, (i.e. full scan data). Buffer A was 98% water, 2% methanol, and 0.2% formic acid. Buffer B was 10% water, 10% isopropanol, 80% acetonitrile, and 0.2% formic acid. Samples were loaded at 4 µL/min for 10 min, and a gradient from 0–45% B at 375 nL/min was run over 130 min, for a total run time of 150 min (including regeneration and sample loading). The Thermo Scientific LTQ Orbitrap Velos hybrid ion trap-Orbitrap mass spectrometer was run in a standard Top-10 data-dependent configuration, except that a higher trigger-threshold (20K) was used to ensure that the MS2 did not interfere with the full-scan duty cycle. This ensured optimal full-scan data for quantification. MS2 fragmentation and analysis were performed in the ion trap mass analyzer.

TABLE 1. High-throughput is an important advantage of the Two-Pass workflow. For the sample preparation and LC-MS/MS steps, the workflow in this example has a ten-fold time advantage over fractionation sample preparation approaches. In comparable timeframe we can investigate ten times the number of human samples. Thus, the Two-Pass workflow offers significantly improved sample statistics.

Results Time-Saving Two-Pass Workflow

The Two-Pass workflow was applied to both sample sets. Our previous findings have demonstrated that the Two-Pass workflow can accurately detect, quantify and identify unlabeled differentially-expressed proteins within plasma samples.1,2,3,4 Given the robust chromatography, high-resolution mass spectrometers and computing power of SIEVE 2.0 software, the workflow enables identification of differentially expressed proteins in a single-pass experiment that is two- to ten-times faster than other approaches (Table 1).

SEQUEST is a registered trademark of the University of Washington. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

FIGURE 2. Pass 1 acquisition cycle.

Figure 3 shows the gel view of the frames obtained from nine LC-MS runs In this example, we have 20,000 potentially useful frames in the gel view. A frame represents a potentially interesting feature (peptide) found in a collective data set. Using frame filters, we can separate the differentially expressed frames between the two groups (see Table 2).

SIEVE Software Analysis

The following data are representative of a proteomics differential case study with ROC analysis from an animal model for sepsis.

Table 2 shows that the Two-Pass workflow allows for confident identification of total protein as well as unlabeled differentially expressed peptides. Initial analysis steps include the generation of 20,000 frames (peptides). Using frame filters based upon ROC AUC or ratios, we easily identified frames that separate the infected from non-infected groups. Given that our Pass-1 experiments are simultaneously collecting Top-10 data-dependent MS2 spectra, we can easily identify the abundant proteins (FDR=1%). The unidentified frames can then be exported to an inclusion list for Pass-2 analysis.

Proteomics data analysis was performed using Thermo Scientific SIEVE software version 2.0 that features chromatographic alignment, framing, differential ROC and ratio analyses. Both Top-10 data-dependent scans and full-scan data were analyzed with the SIEVETM software using chromatographic alignment followed by feature extraction using unsupervised statistical techniques including isotope deconvolution. ROC curves were constructed for the top marker candidates as determined by their ROC Area-Under-the-Curve (AUC). Based upon various criteria including ROC AUC, low ratios, high ratios, high abundance, and low abundance, an inclusion list was created for the best candidates. This inclusion list was used for MS2 acquisition in Pass 2.

Data Analysis

Pass 2. Masses that represent differentially expressed MS1 features were built into an inclusion list and used in Pass 2 analysis. This ensured that the instrument only acquired MS2 data for the masses in the inclusion list (not the highest intensity masses). A larger sample load was used in the Pass 2 runs (600 ng to 800 ng), allowing for higher quality MS2 spectra. Because these full-scan spectra would not be used for quantification, peak shape and intensity reproducibility were not crucial. All fragmentation analysis was done in the Orbitrap mass analyzer, using both HCD and CID. Fragmentation scans from Pass 2 were analyzed using SEQUEST® and FDR analysis to make identifications. The fragmentation search results from Pass 2, along with the quantitative information obtained from Pass 1, were also analyzed using the SIEVE software. Fragmentation scan information was assigned to frames based upon the precursor m/z and retention time.

FIGURE 1. Two-Pass discovery workflow. With robust LC-MS/MS methods and powerful SIEVE software, we are able to identify differentially-expressed proteins in unlabeled plasma samples.

Pass 1 Acquisition Cycle

Two-Pass Workflow Classic Fractionation

n= 50 50

Depletion NO 1 hr/sample

Fractionate NO 10 fractions

Digest 50 samples 500 samples

LC-MS/MS 100 hrs 1000 hrs Condition/ Filters Proteins Peptides (frames) ALL frames 269 20,000 AUC > 0.8 127 8562 Fold Change >1.5 65 3944 Identified in Pass 1 65 112 Unidentified in Pass 1 (for Pass-2 inclusion list ) 0 3832

TABLE 2. Number of proteins and peptides that meet filtering criteria in Pass 1 within a ROC analysis experiment.

Figure 4 shows a frequency histogram of scheduled fragment events per minute for Pass-2 analysis (Pass-2 Inclusion List). After the given frame filter conditions [AUC>0.8 and (NRatio>1.5 or NRatio<0.6) and PRElement<1 and goodid<1 and Charge>1 and Charge<5], 1226 frames were exported as an inclusion list for Pass-2 analysis. NOTE: These frames represent differentially expressed peptides.

Pass-2 Results

On average, we achieved an 82% success rate in MS2 acquisitions from the inclusion list. Frame parameters pending, we have had even higher success rates in other experiments.

Pass 1Pass 2

Con

trol P

lasm

a

log 1

0 con

cent

ratio

n pg

/ml

Protein Name

FIGURE 5. Plasma protein concentrations of identified proteins.

BlankPeptide

Stds

Peptide Stds

Peptide Stds

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

Figure 2 shows the method for assessing systematic errors without technical replicates. Systematic errors are assessed from triplicate acquisitions of standardized peptide samples. Internal standards are spiked in all samples. This approach eliminates the need for sample replicates and conserves valuable specimens. All samples are acquired in high-resolution full-scan and Top-10 data dependent mode, on the LTQ Orbitrap Velos mass spectrometer. Biological variance and outliers are assessed with CV, PCA and other statistical methods.

FIGURE 3. Gel view of frames from 9 LC-MS runs.

Figure 5 shows an example of proteins identified in Pass 1 and Pass 2 that have a minimum of two peptides with FDR of 1%, from ROC analysis of human samples. The identified plasma proteins from our mass spectrometry analysis were searched against a database of known plasma protein concentrations5. In Pass 1, we identify proteins that span at least nine orders of magnitude.

FIGURE 4. Frequency histogram of scheduled fragment events per minute for Pass-2 analysis

An example whisker plot of peptide coverage of a potential biomarker is shown in Figure 6. Pass 2 often results in an increased peptide coverage of the proteins identified in Pass 1. With respect to roughly 10% of frames representing one peptide of a given protein in Pass 1, we identified a second peptide for that corresponding protein in Pass 2. This result strengthens confidence in identification and quantification.

FIGURE 6. Example whisker plot of peptide coverage of a potential biomarker.

Blank run

Peptide Calibration Standards

Human Plasma Samples – Full Scan + Top 10

4 Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow

Design and Optimization

•Robust, commercially available nanoflow LC

•Commercially available columns

•Focus on stable spray•Focus on high

reproducibility of peak intensities, CV<8%

Pass 1: Quantification

•Chromatographic alignment•Uncompromised full scan

measurements •Each sample is measured once

– no need for technical replicates

•Internal peptide standards (normalization)

•Triplicate runs of peptide standards every 12 runs

(instrument QC)•“Top10” data dependent

acquisition• Stringent precursor ion

selection criteria

Pass 2: Identification

•Targeted fragmentation by inclusion list

•Relaxed precursor ion selection criteria

•Not all samples measured –subset as determined from SIEVE

analysis•Internal peptide standards

•Marker stratification using multi-marker and single ROC AUC•Export to pathway analysis

methods Inclusion list

Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow Maryann S Vogelsang1, David A Sarracino1, Janin Schulte2, Bryan Krastins1, Michael Athanas3, Amol Prakash1, Alejandra Garces1, Gouri Vadali1,3, Shadab Ahmad1,Gaiané Demirdjian2, Jennifer N Sutton1, Heinz Redl4, Joachim Struck2, Bruno Darbouret2, Mary F Lopez1

1Thermo Fisher Scientific, Cambridge, MA; USA 2Thermo Fisher Scientific, Hennigsdorf, Germany; 3VAST Scientific, Cambridge, MA, USA; 4Ludwig Boltzmann Institute, Vienna, Austria

Conclusion The Two-Pass workflow was successfully applied to multiple cohorts of both

primate and human plasma research samples.

The Two-Pass workflow accelerates the time to targeted assays and validation of potential biomarkers.

The Two-Pass workflow delivers increased confidence in protein biomarkers through increased peptide coverage of differentially expressed peptides spanning nine orders of magnitude of protein abundance.

References 1. Athanas, M., MacCoss, MJ., Prakash, A., Kall, L., Tomazella, D., Maclean, B.,

Rezai, T., Krastins, B., Sarracino, D., Garces, A., Fortune, S., and Lopez, MF. (2009) Label-free Differential Analysis: An Iterative Approach to Increased Coverage, Improved Statistics and Results. Poster presentation at ASMS.

2. Athanas, M., Sarracino, D., Rezai, T., Prakash, A., Sutton, J., Krastins, B., Ning, M., and Lopez, MF. (2010) A Two-pass Informatics-driven Label-free Workflow For Discovery Of Neurovascular Mediators In PFO-Related Stroke. Poster presentation at ASMS.

3. Lopez, MF., Kuppusamy, R., Sarracino, DA., Prakash, A., Athanas, M., Krastins, B., Rezai, T., Sutton, JN., Peterman, S., and Nicolaides, K. Mass Spectrometric Discovery and Selective Reaction Monitoring (SRM) of Putative Protein Biomarker Candidates in First Trimester Trisomy 21 Maternal Serum. J Proteome Res. 2011 10(1):133-42.

4. Lopez, MF., Sarracino DA., Prakash A., Athanas M., Krastins, B., Rezai, T., Sutton JN., Peterman S., GvozdyakO., Chou S., Lo E., Buonanno F., and Ning MM. Discrimination of ischemic and hemorrhagic strokes using a multiplexed, mass spec-based assay for serum apolipoproteins coupled to multi-marker ROC algorithm. Proteomics Clinical Applications 2012 6: 190-200.

5. Anderson, NL, and Anderson, NG. The Human Plasma Proteome: History, Character, and Diagnostic Prospects. Mol. Cell. Proteomics 2002 1, 845-867.

Overview Purpose: Sample preparation relying on fractionation to simplify the complexity and large dynamic range of plasma or serum samples does so at the potential cost of inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery using a “Two-Pass” workflow that reduces the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Methods: Two-Pass discovery workflow using high-resolution LC-MS/MS coupled to ROC and differential expression analyses from stratified human sample cohorts and non-human animal models.

Results: The Two-Pass workflow reduced the number of replicates needed. Compared to a single-pass experiment, the workflow provided 20–50% more quantitatively associated protein identifications in a shorter time period (2–10X).

Introduction Proteomic-based biomarker discovery approaches have primarily focused on direct profiling of serum or plasma to determine abundance changes that can be used to discriminate between populations. Due to the complexity and large dynamic range of serum and plasma, samples are often processed by depletion and/or fractionation in order to reduce their complexity. However, these manipulations can result in inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery the uses a Two-Pass workflow to reduce the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Previously, we introduced a two-pass workflow exploiting the mass spectrometer’s accurate mass and broad dynamic range, which investigated the uncompromised quantitative data in Pass 1, and then targeted differentially expressed MS features in Pass 2 1,2,3,4. In Pass 1, we take advantage of the speed of the hybrid mass spectrometer to measure quantitative MS1 frames while concurrently measuring and identifying the Top-10 data-dependent MS2. In Pass 2 we identify additional differentially expressed peptides from the inclusion list built from Pass-1 data analysis.

Methods Sample Preparation

This poster presents results from both primate and human plasma samples. The Two-Pass workflow was applied to 64 plasma samples of microbial-infected primates (monitoring eight subjects over a 24 hr time period, two of which were controls). Additionally, a small cohort of IRB approved human plasma samples were analyzed. The preliminary results in this report are intended to demonstrate the Two-Pass workflow, not biomarker discovery results.

Liquid Chromatography and High-Resolution Mass Spectrometry

As shown in Figure 1, the Two-Pass workflow strategy consists of the separate optimization of MS parameters and configuration for protein quantification and identification.

Pass 1. Plasma samples (300 ng of primate samples or 500 ng of human samples), were digested with trypsin and injected onto a Thermo Scientific EASY-nLC system configured with a 10 cm x 100 µm trap column and a 25 cm x 100 µm ID resolving column. The sample load was optimized for optimum quantification, (i.e. full scan data). Buffer A was 98% water, 2% methanol, and 0.2% formic acid. Buffer B was 10% water, 10% isopropanol, 80% acetonitrile, and 0.2% formic acid. Samples were loaded at 4 µL/min for 10 min, and a gradient from 0–45% B at 375 nL/min was run over 130 min, for a total run time of 150 min (including regeneration and sample loading). The Thermo Scientific LTQ Orbitrap Velos hybrid ion trap-Orbitrap mass spectrometer was run in a standard Top-10 data-dependent configuration, except that a higher trigger-threshold (20K) was used to ensure that the MS2 did not interfere with the full-scan duty cycle. This ensured optimal full-scan data for quantification. MS2 fragmentation and analysis were performed in the ion trap mass analyzer.

TABLE 1. High-throughput is an important advantage of the Two-Pass workflow. For the sample preparation and LC-MS/MS steps, the workflow in this example has a ten-fold time advantage over fractionation sample preparation approaches. In comparable timeframe we can investigate ten times the number of human samples. Thus, the Two-Pass workflow offers significantly improved sample statistics.

Results Time-Saving Two-Pass Workflow

The Two-Pass workflow was applied to both sample sets. Our previous findings have demonstrated that the Two-Pass workflow can accurately detect, quantify and identify unlabeled differentially-expressed proteins within plasma samples.1,2,3,4 Given the robust chromatography, high-resolution mass spectrometers and computing power of SIEVE 2.0 software, the workflow enables identification of differentially expressed proteins in a single-pass experiment that is two- to ten-times faster than other approaches (Table 1).

SEQUEST is a registered trademark of the University of Washington. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

FIGURE 2. Pass 1 acquisition cycle.

Figure 3 shows the gel view of the frames obtained from nine LC-MS runs In this example, we have 20,000 potentially useful frames in the gel view. A frame represents a potentially interesting feature (peptide) found in a collective data set. Using frame filters, we can separate the differentially expressed frames between the two groups (see Table 2).

SIEVE Software Analysis

The following data are representative of a proteomics differential case study with ROC analysis from an animal model for sepsis.

Table 2 shows that the Two-Pass workflow allows for confident identification of total protein as well as unlabeled differentially expressed peptides. Initial analysis steps include the generation of 20,000 frames (peptides). Using frame filters based upon ROC AUC or ratios, we easily identified frames that separate the infected from non-infected groups. Given that our Pass-1 experiments are simultaneously collecting Top-10 data-dependent MS2 spectra, we can easily identify the abundant proteins (FDR=1%). The unidentified frames can then be exported to an inclusion list for Pass-2 analysis.

Proteomics data analysis was performed using Thermo Scientific SIEVE software version 2.0 that features chromatographic alignment, framing, differential ROC and ratio analyses. Both Top-10 data-dependent scans and full-scan data were analyzed with the SIEVETM software using chromatographic alignment followed by feature extraction using unsupervised statistical techniques including isotope deconvolution. ROC curves were constructed for the top marker candidates as determined by their ROC Area-Under-the-Curve (AUC). Based upon various criteria including ROC AUC, low ratios, high ratios, high abundance, and low abundance, an inclusion list was created for the best candidates. This inclusion list was used for MS2 acquisition in Pass 2.

Data Analysis

Pass 2. Masses that represent differentially expressed MS1 features were built into an inclusion list and used in Pass 2 analysis. This ensured that the instrument only acquired MS2 data for the masses in the inclusion list (not the highest intensity masses). A larger sample load was used in the Pass 2 runs (600 ng to 800 ng), allowing for higher quality MS2 spectra. Because these full-scan spectra would not be used for quantification, peak shape and intensity reproducibility were not crucial. All fragmentation analysis was done in the Orbitrap mass analyzer, using both HCD and CID. Fragmentation scans from Pass 2 were analyzed using SEQUEST® and FDR analysis to make identifications. The fragmentation search results from Pass 2, along with the quantitative information obtained from Pass 1, were also analyzed using the SIEVE software. Fragmentation scan information was assigned to frames based upon the precursor m/z and retention time.

FIGURE 1. Two-Pass discovery workflow. With robust LC-MS/MS methods and powerful SIEVE software, we are able to identify differentially-expressed proteins in unlabeled plasma samples.

Pass 1 Acquisition Cycle

Two-Pass Workflow Classic Fractionation

n= 50 50

Depletion NO 1 hr/sample

Fractionate NO 10 fractions

Digest 50 samples 500 samples

LC-MS/MS 100 hrs 1000 hrs Condition/ Filters Proteins Peptides (frames) ALL frames 269 20,000 AUC > 0.8 127 8562 Fold Change >1.5 65 3944 Identified in Pass 1 65 112 Unidentified in Pass 1 (for Pass-2 inclusion list ) 0 3832

TABLE 2. Number of proteins and peptides that meet filtering criteria in Pass 1 within a ROC analysis experiment.

Figure 4 shows a frequency histogram of scheduled fragment events per minute for Pass-2 analysis (Pass-2 Inclusion List). After the given frame filter conditions [AUC>0.8 and (NRatio>1.5 or NRatio<0.6) and PRElement<1 and goodid<1 and Charge>1 and Charge<5], 1226 frames were exported as an inclusion list for Pass-2 analysis. NOTE: These frames represent differentially expressed peptides.

Pass-2 Results

On average, we achieved an 82% success rate in MS2 acquisitions from the inclusion list. Frame parameters pending, we have had even higher success rates in other experiments.

Pass 1Pass 2

Con

trol P

lasm

a

log 1

0 con

cent

ratio

n pg

/ml

Protein Name

FIGURE 5. Plasma protein concentrations of identified proteins.

BlankPeptide

Stds

Peptide Stds

Peptide Stds

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

Figure 2 shows the method for assessing systematic errors without technical replicates. Systematic errors are assessed from triplicate acquisitions of standardized peptide samples. Internal standards are spiked in all samples. This approach eliminates the need for sample replicates and conserves valuable specimens. All samples are acquired in high-resolution full-scan and Top-10 data dependent mode, on the LTQ Orbitrap Velos mass spectrometer. Biological variance and outliers are assessed with CV, PCA and other statistical methods.

FIGURE 3. Gel view of frames from 9 LC-MS runs.

Figure 5 shows an example of proteins identified in Pass 1 and Pass 2 that have a minimum of two peptides with FDR of 1%, from ROC analysis of human samples. The identified plasma proteins from our mass spectrometry analysis were searched against a database of known plasma protein concentrations5. In Pass 1, we identify proteins that span at least nine orders of magnitude.

FIGURE 4. Frequency histogram of scheduled fragment events per minute for Pass-2 analysis

An example whisker plot of peptide coverage of a potential biomarker is shown in Figure 6. Pass 2 often results in an increased peptide coverage of the proteins identified in Pass 1. With respect to roughly 10% of frames representing one peptide of a given protein in Pass 1, we identified a second peptide for that corresponding protein in Pass 2. This result strengthens confidence in identification and quantification.

FIGURE 6. Example whisker plot of peptide coverage of a potential biomarker.

Blank run

Peptide Calibration Standards

Human Plasma Samples – Full Scan + Top 10

5Thermo Scientific Poster Note • PN63592_E 06/12S

Design and Optimization

•Robust, commercially available nanoflow LC

•Commercially available columns

•Focus on stable spray•Focus on high

reproducibility of peak intensities, CV<8%

Pass 1: Quantification

•Chromatographic alignment•Uncompromised full scan

measurements •Each sample is measured once

– no need for technical replicates

•Internal peptide standards (normalization)

•Triplicate runs of peptide standards every 12 runs

(instrument QC)•“Top10” data dependent

acquisition• Stringent precursor ion

selection criteria

Pass 2: Identification

•Targeted fragmentation by inclusion list

•Relaxed precursor ion selection criteria

•Not all samples measured –subset as determined from SIEVE

analysis•Internal peptide standards

•Marker stratification using multi-marker and single ROC AUC•Export to pathway analysis

methods Inclusion list

Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow Maryann S Vogelsang1, David A Sarracino1, Janin Schulte2, Bryan Krastins1, Michael Athanas3, Amol Prakash1, Alejandra Garces1, Gouri Vadali1,3, Shadab Ahmad1,Gaiané Demirdjian2, Jennifer N Sutton1, Heinz Redl4, Joachim Struck2, Bruno Darbouret2, Mary F Lopez1

1Thermo Fisher Scientific, Cambridge, MA; USA 2Thermo Fisher Scientific, Hennigsdorf, Germany; 3VAST Scientific, Cambridge, MA, USA; 4Ludwig Boltzmann Institute, Vienna, Austria

Conclusion The Two-Pass workflow was successfully applied to multiple cohorts of both

primate and human plasma research samples.

The Two-Pass workflow accelerates the time to targeted assays and validation of potential biomarkers.

The Two-Pass workflow delivers increased confidence in protein biomarkers through increased peptide coverage of differentially expressed peptides spanning nine orders of magnitude of protein abundance.

References 1. Athanas, M., MacCoss, MJ., Prakash, A., Kall, L., Tomazella, D., Maclean, B.,

Rezai, T., Krastins, B., Sarracino, D., Garces, A., Fortune, S., and Lopez, MF. (2009) Label-free Differential Analysis: An Iterative Approach to Increased Coverage, Improved Statistics and Results. Poster presentation at ASMS.

2. Athanas, M., Sarracino, D., Rezai, T., Prakash, A., Sutton, J., Krastins, B., Ning, M., and Lopez, MF. (2010) A Two-pass Informatics-driven Label-free Workflow For Discovery Of Neurovascular Mediators In PFO-Related Stroke. Poster presentation at ASMS.

3. Lopez, MF., Kuppusamy, R., Sarracino, DA., Prakash, A., Athanas, M., Krastins, B., Rezai, T., Sutton, JN., Peterman, S., and Nicolaides, K. Mass Spectrometric Discovery and Selective Reaction Monitoring (SRM) of Putative Protein Biomarker Candidates in First Trimester Trisomy 21 Maternal Serum. J Proteome Res. 2011 10(1):133-42.

4. Lopez, MF., Sarracino DA., Prakash A., Athanas M., Krastins, B., Rezai, T., Sutton JN., Peterman S., GvozdyakO., Chou S., Lo E., Buonanno F., and Ning MM. Discrimination of ischemic and hemorrhagic strokes using a multiplexed, mass spec-based assay for serum apolipoproteins coupled to multi-marker ROC algorithm. Proteomics Clinical Applications 2012 6: 190-200.

5. Anderson, NL, and Anderson, NG. The Human Plasma Proteome: History, Character, and Diagnostic Prospects. Mol. Cell. Proteomics 2002 1, 845-867.

Overview Purpose: Sample preparation relying on fractionation to simplify the complexity and large dynamic range of plasma or serum samples does so at the potential cost of inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery using a “Two-Pass” workflow that reduces the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Methods: Two-Pass discovery workflow using high-resolution LC-MS/MS coupled to ROC and differential expression analyses from stratified human sample cohorts and non-human animal models.

Results: The Two-Pass workflow reduced the number of replicates needed. Compared to a single-pass experiment, the workflow provided 20–50% more quantitatively associated protein identifications in a shorter time period (2–10X).

Introduction Proteomic-based biomarker discovery approaches have primarily focused on direct profiling of serum or plasma to determine abundance changes that can be used to discriminate between populations. Due to the complexity and large dynamic range of serum and plasma, samples are often processed by depletion and/or fractionation in order to reduce their complexity. However, these manipulations can result in inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery the uses a Two-Pass workflow to reduce the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Previously, we introduced a two-pass workflow exploiting the mass spectrometer’s accurate mass and broad dynamic range, which investigated the uncompromised quantitative data in Pass 1, and then targeted differentially expressed MS features in Pass 2 1,2,3,4. In Pass 1, we take advantage of the speed of the hybrid mass spectrometer to measure quantitative MS1 frames while concurrently measuring and identifying the Top-10 data-dependent MS2. In Pass 2 we identify additional differentially expressed peptides from the inclusion list built from Pass-1 data analysis.

Methods Sample Preparation

This poster presents results from both primate and human plasma samples. The Two-Pass workflow was applied to 64 plasma samples of microbial-infected primates (monitoring eight subjects over a 24 hr time period, two of which were controls). Additionally, a small cohort of IRB approved human plasma samples were analyzed. The preliminary results in this report are intended to demonstrate the Two-Pass workflow, not biomarker discovery results.

Liquid Chromatography and High-Resolution Mass Spectrometry

As shown in Figure 1, the Two-Pass workflow strategy consists of the separate optimization of MS parameters and configuration for protein quantification and identification.

Pass 1. Plasma samples (300 ng of primate samples or 500 ng of human samples), were digested with trypsin and injected onto a Thermo Scientific EASY-nLC system configured with a 10 cm x 100 µm trap column and a 25 cm x 100 µm ID resolving column. The sample load was optimized for optimum quantification, (i.e. full scan data). Buffer A was 98% water, 2% methanol, and 0.2% formic acid. Buffer B was 10% water, 10% isopropanol, 80% acetonitrile, and 0.2% formic acid. Samples were loaded at 4 µL/min for 10 min, and a gradient from 0–45% B at 375 nL/min was run over 130 min, for a total run time of 150 min (including regeneration and sample loading). The Thermo Scientific LTQ Orbitrap Velos hybrid ion trap-Orbitrap mass spectrometer was run in a standard Top-10 data-dependent configuration, except that a higher trigger-threshold (20K) was used to ensure that the MS2 did not interfere with the full-scan duty cycle. This ensured optimal full-scan data for quantification. MS2 fragmentation and analysis were performed in the ion trap mass analyzer.

TABLE 1. High-throughput is an important advantage of the Two-Pass workflow. For the sample preparation and LC-MS/MS steps, the workflow in this example has a ten-fold time advantage over fractionation sample preparation approaches. In comparable timeframe we can investigate ten times the number of human samples. Thus, the Two-Pass workflow offers significantly improved sample statistics.

Results Time-Saving Two-Pass Workflow

The Two-Pass workflow was applied to both sample sets. Our previous findings have demonstrated that the Two-Pass workflow can accurately detect, quantify and identify unlabeled differentially-expressed proteins within plasma samples.1,2,3,4 Given the robust chromatography, high-resolution mass spectrometers and computing power of SIEVE 2.0 software, the workflow enables identification of differentially expressed proteins in a single-pass experiment that is two- to ten-times faster than other approaches (Table 1).

SEQUEST is a registered trademark of the University of Washington. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

FIGURE 2. Pass 1 acquisition cycle.

Figure 3 shows the gel view of the frames obtained from nine LC-MS runs In this example, we have 20,000 potentially useful frames in the gel view. A frame represents a potentially interesting feature (peptide) found in a collective data set. Using frame filters, we can separate the differentially expressed frames between the two groups (see Table 2).

SIEVE Software Analysis

The following data are representative of a proteomics differential case study with ROC analysis from an animal model for sepsis.

Table 2 shows that the Two-Pass workflow allows for confident identification of total protein as well as unlabeled differentially expressed peptides. Initial analysis steps include the generation of 20,000 frames (peptides). Using frame filters based upon ROC AUC or ratios, we easily identified frames that separate the infected from non-infected groups. Given that our Pass-1 experiments are simultaneously collecting Top-10 data-dependent MS2 spectra, we can easily identify the abundant proteins (FDR=1%). The unidentified frames can then be exported to an inclusion list for Pass-2 analysis.

Proteomics data analysis was performed using Thermo Scientific SIEVE software version 2.0 that features chromatographic alignment, framing, differential ROC and ratio analyses. Both Top-10 data-dependent scans and full-scan data were analyzed with the SIEVETM software using chromatographic alignment followed by feature extraction using unsupervised statistical techniques including isotope deconvolution. ROC curves were constructed for the top marker candidates as determined by their ROC Area-Under-the-Curve (AUC). Based upon various criteria including ROC AUC, low ratios, high ratios, high abundance, and low abundance, an inclusion list was created for the best candidates. This inclusion list was used for MS2 acquisition in Pass 2.

Data Analysis

Pass 2. Masses that represent differentially expressed MS1 features were built into an inclusion list and used in Pass 2 analysis. This ensured that the instrument only acquired MS2 data for the masses in the inclusion list (not the highest intensity masses). A larger sample load was used in the Pass 2 runs (600 ng to 800 ng), allowing for higher quality MS2 spectra. Because these full-scan spectra would not be used for quantification, peak shape and intensity reproducibility were not crucial. All fragmentation analysis was done in the Orbitrap mass analyzer, using both HCD and CID. Fragmentation scans from Pass 2 were analyzed using SEQUEST® and FDR analysis to make identifications. The fragmentation search results from Pass 2, along with the quantitative information obtained from Pass 1, were also analyzed using the SIEVE software. Fragmentation scan information was assigned to frames based upon the precursor m/z and retention time.

FIGURE 1. Two-Pass discovery workflow. With robust LC-MS/MS methods and powerful SIEVE software, we are able to identify differentially-expressed proteins in unlabeled plasma samples.

Pass 1 Acquisition Cycle

Two-Pass Workflow Classic Fractionation

n= 50 50

Depletion NO 1 hr/sample

Fractionate NO 10 fractions

Digest 50 samples 500 samples

LC-MS/MS 100 hrs 1000 hrs Condition/ Filters Proteins Peptides (frames) ALL frames 269 20,000 AUC > 0.8 127 8562 Fold Change >1.5 65 3944 Identified in Pass 1 65 112 Unidentified in Pass 1 (for Pass-2 inclusion list ) 0 3832

TABLE 2. Number of proteins and peptides that meet filtering criteria in Pass 1 within a ROC analysis experiment.

Figure 4 shows a frequency histogram of scheduled fragment events per minute for Pass-2 analysis (Pass-2 Inclusion List). After the given frame filter conditions [AUC>0.8 and (NRatio>1.5 or NRatio<0.6) and PRElement<1 and goodid<1 and Charge>1 and Charge<5], 1226 frames were exported as an inclusion list for Pass-2 analysis. NOTE: These frames represent differentially expressed peptides.

Pass-2 Results

On average, we achieved an 82% success rate in MS2 acquisitions from the inclusion list. Frame parameters pending, we have had even higher success rates in other experiments.

Pass 1Pass 2

Con

trol P

lasm

a

log 1

0 con

cent

ratio

n pg

/ml

Protein Name

FIGURE 5. Plasma protein concentrations of identified proteins.

BlankPeptide

Stds

Peptide Stds

Peptide Stds

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

Figure 2 shows the method for assessing systematic errors without technical replicates. Systematic errors are assessed from triplicate acquisitions of standardized peptide samples. Internal standards are spiked in all samples. This approach eliminates the need for sample replicates and conserves valuable specimens. All samples are acquired in high-resolution full-scan and Top-10 data dependent mode, on the LTQ Orbitrap Velos mass spectrometer. Biological variance and outliers are assessed with CV, PCA and other statistical methods.

FIGURE 3. Gel view of frames from 9 LC-MS runs.

Figure 5 shows an example of proteins identified in Pass 1 and Pass 2 that have a minimum of two peptides with FDR of 1%, from ROC analysis of human samples. The identified plasma proteins from our mass spectrometry analysis were searched against a database of known plasma protein concentrations5. In Pass 1, we identify proteins that span at least nine orders of magnitude.

FIGURE 4. Frequency histogram of scheduled fragment events per minute for Pass-2 analysis

An example whisker plot of peptide coverage of a potential biomarker is shown in Figure 6. Pass 2 often results in an increased peptide coverage of the proteins identified in Pass 1. With respect to roughly 10% of frames representing one peptide of a given protein in Pass 1, we identified a second peptide for that corresponding protein in Pass 2. This result strengthens confidence in identification and quantification.

FIGURE 6. Example whisker plot of peptide coverage of a potential biomarker.

Blank run

Peptide Calibration Standards

Human Plasma Samples – Full Scan + Top 10

6 Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow

Design and Optimization

•Robust, commercially available nanoflow LC

•Commercially available columns

•Focus on stable spray•Focus on high

reproducibility of peak intensities, CV<8%

Pass 1: Quantification

•Chromatographic alignment•Uncompromised full scan

measurements •Each sample is measured once

– no need for technical replicates

•Internal peptide standards (normalization)

•Triplicate runs of peptide standards every 12 runs

(instrument QC)•“Top10” data dependent

acquisition• Stringent precursor ion

selection criteria

Pass 2: Identification

•Targeted fragmentation by inclusion list

•Relaxed precursor ion selection criteria

•Not all samples measured –subset as determined from SIEVE

analysis•Internal peptide standards

•Marker stratification using multi-marker and single ROC AUC•Export to pathway analysis

methods Inclusion list

Proteomics-Based Biomarker Discovery Study in Human and non-Human Plasma Using a Two-Pass Workflow Maryann S Vogelsang1, David A Sarracino1, Janin Schulte2, Bryan Krastins1, Michael Athanas3, Amol Prakash1, Alejandra Garces1, Gouri Vadali1,3, Shadab Ahmad1,Gaiané Demirdjian2, Jennifer N Sutton1, Heinz Redl4, Joachim Struck2, Bruno Darbouret2, Mary F Lopez1

1Thermo Fisher Scientific, Cambridge, MA; USA 2Thermo Fisher Scientific, Hennigsdorf, Germany; 3VAST Scientific, Cambridge, MA, USA; 4Ludwig Boltzmann Institute, Vienna, Austria

Conclusion The Two-Pass workflow was successfully applied to multiple cohorts of both

primate and human plasma research samples.

The Two-Pass workflow accelerates the time to targeted assays and validation of potential biomarkers.

The Two-Pass workflow delivers increased confidence in protein biomarkers through increased peptide coverage of differentially expressed peptides spanning nine orders of magnitude of protein abundance.

References 1. Athanas, M., MacCoss, MJ., Prakash, A., Kall, L., Tomazella, D., Maclean, B.,

Rezai, T., Krastins, B., Sarracino, D., Garces, A., Fortune, S., and Lopez, MF. (2009) Label-free Differential Analysis: An Iterative Approach to Increased Coverage, Improved Statistics and Results. Poster presentation at ASMS.

2. Athanas, M., Sarracino, D., Rezai, T., Prakash, A., Sutton, J., Krastins, B., Ning, M., and Lopez, MF. (2010) A Two-pass Informatics-driven Label-free Workflow For Discovery Of Neurovascular Mediators In PFO-Related Stroke. Poster presentation at ASMS.

3. Lopez, MF., Kuppusamy, R., Sarracino, DA., Prakash, A., Athanas, M., Krastins, B., Rezai, T., Sutton, JN., Peterman, S., and Nicolaides, K. Mass Spectrometric Discovery and Selective Reaction Monitoring (SRM) of Putative Protein Biomarker Candidates in First Trimester Trisomy 21 Maternal Serum. J Proteome Res. 2011 10(1):133-42.

4. Lopez, MF., Sarracino DA., Prakash A., Athanas M., Krastins, B., Rezai, T., Sutton JN., Peterman S., GvozdyakO., Chou S., Lo E., Buonanno F., and Ning MM. Discrimination of ischemic and hemorrhagic strokes using a multiplexed, mass spec-based assay for serum apolipoproteins coupled to multi-marker ROC algorithm. Proteomics Clinical Applications 2012 6: 190-200.

5. Anderson, NL, and Anderson, NG. The Human Plasma Proteome: History, Character, and Diagnostic Prospects. Mol. Cell. Proteomics 2002 1, 845-867.

Overview Purpose: Sample preparation relying on fractionation to simplify the complexity and large dynamic range of plasma or serum samples does so at the potential cost of inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery using a “Two-Pass” workflow that reduces the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Methods: Two-Pass discovery workflow using high-resolution LC-MS/MS coupled to ROC and differential expression analyses from stratified human sample cohorts and non-human animal models.

Results: The Two-Pass workflow reduced the number of replicates needed. Compared to a single-pass experiment, the workflow provided 20–50% more quantitatively associated protein identifications in a shorter time period (2–10X).

Introduction Proteomic-based biomarker discovery approaches have primarily focused on direct profiling of serum or plasma to determine abundance changes that can be used to discriminate between populations. Due to the complexity and large dynamic range of serum and plasma, samples are often processed by depletion and/or fractionation in order to reduce their complexity. However, these manipulations can result in inaccurate or unreliable abundance measurements. Here we describe a simplified approach to biomarker discovery the uses a Two-Pass workflow to reduce the need for physical sample fractionation. The workflow encompasses robust, reproducible sample preparation, chromatography and strong informatics-driven data analysis.

Previously, we introduced a two-pass workflow exploiting the mass spectrometer’s accurate mass and broad dynamic range, which investigated the uncompromised quantitative data in Pass 1, and then targeted differentially expressed MS features in Pass 2 1,2,3,4. In Pass 1, we take advantage of the speed of the hybrid mass spectrometer to measure quantitative MS1 frames while concurrently measuring and identifying the Top-10 data-dependent MS2. In Pass 2 we identify additional differentially expressed peptides from the inclusion list built from Pass-1 data analysis.

Methods Sample Preparation

This poster presents results from both primate and human plasma samples. The Two-Pass workflow was applied to 64 plasma samples of microbial-infected primates (monitoring eight subjects over a 24 hr time period, two of which were controls). Additionally, a small cohort of IRB approved human plasma samples were analyzed. The preliminary results in this report are intended to demonstrate the Two-Pass workflow, not biomarker discovery results.

Liquid Chromatography and High-Resolution Mass Spectrometry

As shown in Figure 1, the Two-Pass workflow strategy consists of the separate optimization of MS parameters and configuration for protein quantification and identification.

Pass 1. Plasma samples (300 ng of primate samples or 500 ng of human samples), were digested with trypsin and injected onto a Thermo Scientific EASY-nLC system configured with a 10 cm x 100 µm trap column and a 25 cm x 100 µm ID resolving column. The sample load was optimized for optimum quantification, (i.e. full scan data). Buffer A was 98% water, 2% methanol, and 0.2% formic acid. Buffer B was 10% water, 10% isopropanol, 80% acetonitrile, and 0.2% formic acid. Samples were loaded at 4 µL/min for 10 min, and a gradient from 0–45% B at 375 nL/min was run over 130 min, for a total run time of 150 min (including regeneration and sample loading). The Thermo Scientific LTQ Orbitrap Velos hybrid ion trap-Orbitrap mass spectrometer was run in a standard Top-10 data-dependent configuration, except that a higher trigger-threshold (20K) was used to ensure that the MS2 did not interfere with the full-scan duty cycle. This ensured optimal full-scan data for quantification. MS2 fragmentation and analysis were performed in the ion trap mass analyzer.

TABLE 1. High-throughput is an important advantage of the Two-Pass workflow. For the sample preparation and LC-MS/MS steps, the workflow in this example has a ten-fold time advantage over fractionation sample preparation approaches. In comparable timeframe we can investigate ten times the number of human samples. Thus, the Two-Pass workflow offers significantly improved sample statistics.

Results Time-Saving Two-Pass Workflow

The Two-Pass workflow was applied to both sample sets. Our previous findings have demonstrated that the Two-Pass workflow can accurately detect, quantify and identify unlabeled differentially-expressed proteins within plasma samples.1,2,3,4 Given the robust chromatography, high-resolution mass spectrometers and computing power of SIEVE 2.0 software, the workflow enables identification of differentially expressed proteins in a single-pass experiment that is two- to ten-times faster than other approaches (Table 1).

SEQUEST is a registered trademark of the University of Washington. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

FIGURE 2. Pass 1 acquisition cycle.

Figure 3 shows the gel view of the frames obtained from nine LC-MS runs In this example, we have 20,000 potentially useful frames in the gel view. A frame represents a potentially interesting feature (peptide) found in a collective data set. Using frame filters, we can separate the differentially expressed frames between the two groups (see Table 2).

SIEVE Software Analysis

The following data are representative of a proteomics differential case study with ROC analysis from an animal model for sepsis.

Table 2 shows that the Two-Pass workflow allows for confident identification of total protein as well as unlabeled differentially expressed peptides. Initial analysis steps include the generation of 20,000 frames (peptides). Using frame filters based upon ROC AUC or ratios, we easily identified frames that separate the infected from non-infected groups. Given that our Pass-1 experiments are simultaneously collecting Top-10 data-dependent MS2 spectra, we can easily identify the abundant proteins (FDR=1%). The unidentified frames can then be exported to an inclusion list for Pass-2 analysis.

Proteomics data analysis was performed using Thermo Scientific SIEVE software version 2.0 that features chromatographic alignment, framing, differential ROC and ratio analyses. Both Top-10 data-dependent scans and full-scan data were analyzed with the SIEVETM software using chromatographic alignment followed by feature extraction using unsupervised statistical techniques including isotope deconvolution. ROC curves were constructed for the top marker candidates as determined by their ROC Area-Under-the-Curve (AUC). Based upon various criteria including ROC AUC, low ratios, high ratios, high abundance, and low abundance, an inclusion list was created for the best candidates. This inclusion list was used for MS2 acquisition in Pass 2.

Data Analysis

Pass 2. Masses that represent differentially expressed MS1 features were built into an inclusion list and used in Pass 2 analysis. This ensured that the instrument only acquired MS2 data for the masses in the inclusion list (not the highest intensity masses). A larger sample load was used in the Pass 2 runs (600 ng to 800 ng), allowing for higher quality MS2 spectra. Because these full-scan spectra would not be used for quantification, peak shape and intensity reproducibility were not crucial. All fragmentation analysis was done in the Orbitrap mass analyzer, using both HCD and CID. Fragmentation scans from Pass 2 were analyzed using SEQUEST® and FDR analysis to make identifications. The fragmentation search results from Pass 2, along with the quantitative information obtained from Pass 1, were also analyzed using the SIEVE software. Fragmentation scan information was assigned to frames based upon the precursor m/z and retention time.

FIGURE 1. Two-Pass discovery workflow. With robust LC-MS/MS methods and powerful SIEVE software, we are able to identify differentially-expressed proteins in unlabeled plasma samples.

Pass 1 Acquisition Cycle

Two-Pass Workflow Classic Fractionation

n= 50 50

Depletion NO 1 hr/sample

Fractionate NO 10 fractions

Digest 50 samples 500 samples

LC-MS/MS 100 hrs 1000 hrs Condition/ Filters Proteins Peptides (frames) ALL frames 269 20,000 AUC > 0.8 127 8562 Fold Change >1.5 65 3944 Identified in Pass 1 65 112 Unidentified in Pass 1 (for Pass-2 inclusion list ) 0 3832

TABLE 2. Number of proteins and peptides that meet filtering criteria in Pass 1 within a ROC analysis experiment.

Figure 4 shows a frequency histogram of scheduled fragment events per minute for Pass-2 analysis (Pass-2 Inclusion List). After the given frame filter conditions [AUC>0.8 and (NRatio>1.5 or NRatio<0.6) and PRElement<1 and goodid<1 and Charge>1 and Charge<5], 1226 frames were exported as an inclusion list for Pass-2 analysis. NOTE: These frames represent differentially expressed peptides.

Pass-2 Results

On average, we achieved an 82% success rate in MS2 acquisitions from the inclusion list. Frame parameters pending, we have had even higher success rates in other experiments.

Pass 1Pass 2

Con

trol P

lasm

a

log 1

0 con

cent

ratio

n pg

/ml

Protein Name

FIGURE 5. Plasma protein concentrations of identified proteins.

BlankPeptide

Stds

Peptide Stds

Peptide Stds

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

SampleFull Scan

Figure 2 shows the method for assessing systematic errors without technical replicates. Systematic errors are assessed from triplicate acquisitions of standardized peptide samples. Internal standards are spiked in all samples. This approach eliminates the need for sample replicates and conserves valuable specimens. All samples are acquired in high-resolution full-scan and Top-10 data dependent mode, on the LTQ Orbitrap Velos mass spectrometer. Biological variance and outliers are assessed with CV, PCA and other statistical methods.

FIGURE 3. Gel view of frames from 9 LC-MS runs.

Figure 5 shows an example of proteins identified in Pass 1 and Pass 2 that have a minimum of two peptides with FDR of 1%, from ROC analysis of human samples. The identified plasma proteins from our mass spectrometry analysis were searched against a database of known plasma protein concentrations5. In Pass 1, we identify proteins that span at least nine orders of magnitude.

FIGURE 4. Frequency histogram of scheduled fragment events per minute for Pass-2 analysis

An example whisker plot of peptide coverage of a potential biomarker is shown in Figure 6. Pass 2 often results in an increased peptide coverage of the proteins identified in Pass 1. With respect to roughly 10% of frames representing one peptide of a given protein in Pass 1, we identified a second peptide for that corresponding protein in Pass 2. This result strengthens confidence in identification and quantification.

FIGURE 6. Example whisker plot of peptide coverage of a potential biomarker.

Blank run

Peptide Calibration Standards

Human Plasma Samples – Full Scan + Top 10

Thermo Fisher Scientific, San Jose, CA USA is ISO Certified.

PN63592_E 06/12S

Africa-Other +27 11 570 1840Australia +61 3 9757 4300Austria +43 1 333 50 34 0Belgium +32 53 73 42 41Canada +1 800 530 8447China +86 10 8419 3588Denmark +45 70 23 62 60

Europe-Other +43 1 333 50 34 0Finland/Norway/Sweden +46 8 556 468 00France +33 1 60 92 48 00Germany +49 6103 408 1014India +91 22 6742 9434Italy +39 02 950 591

Japan +81 45 453 9100Latin America +1 561 688 8700Middle East +43 1 333 50 34 0Netherlands +31 76 579 55 55New Zealand +64 9 980 6700Russia/CIS +43 1 333 50 34 0South Africa +27 11 570 1840

Spain +34 914 845 965Switzerland +41 61 716 77 00UK +44 1442 233555USA +1 800 532 4752

www.thermoscientific.com©2012 Thermo Fisher Scientific Inc. All rights reserved. ISO is a trademark of the International Standards Organization. All other trademarks are the property of Thermo Fisher Scientific Inc. and its subsidiaries. This information is presented as an example of the capabilities of Thermo Fisher Scientific Inc. products. It is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others. Specifications, terms and pricing are subject to change. Not all products are available in all countries. Please consult your local sales representative for details.