Ms. Christie Hunter: Hello. My name is Christie Hunter, and today I would like to talk to you about a new exciting workflow for targeted protein quantitation called MS/MSALL with SWATH Acquisition

The goal of quantitative proteomics is to identify and quantify a broad range of proteins and peptides in samples that are often complex and very precious

The biology we are studying is often also complex and very dynamic, making experimental design critical to success

Current analytical techniques often require tradeoffs in specific key analytical aspects. For example, data-dependent discovery techniques provide depth of coverage, but often, the reproducibility of peptide detection across samples is insufficient to be used across a broader sample set

Often, sensitivity and selectivity come at the sacrifice of speed of acquisition

Targeted quantitation using MRM analysis has been increasing in importance over the last five years because of the excellent reproducibility and selectivity that it affords in these complex proteomic samples

But, often, the degree of multiplexing possible with current technology is exceeded by the biological questions we are asking

So, what would the attributes be of a better way to do quantitation of proteins and peptides? Obviously, a technique would be accessible, easy to get up and running in a lab

It would have to provide the reproducibility of MRM analysis, but with much higher multiplexing, be much more comprehensive in its quantitative coverage. And ideally, the selectivity and dynamic range would be similar to that of MRM quantitation

So, today, I would like to introduce you to the new MS/MSALL Workflow with SWATH Acquisition for targeted protein quantitation

This workflow has been developed on the TripleTOF 5600 system and brings together a data-independent acquisition strategy for comprehensive MS/MS acquisition with a targeted quantitative data-processing strategy to obtained high-quality protein and peptide quantitation data with the highest multiplexing

So, to begin to explain the details of the workflow, I would like to use the example of the MRMhr workflow to explain targeted quantitation on a TripleTOF

During a LCMS analysis, there are many peptides eluting off the column across a broad dynamic range of about 400 to 1,000

In this MRMHr analysis, a single peptide is selected by Q1, the first mass-analyzing quadrupole, fragmented in the collision cell, and then analyzed in the TOF MS to generate a full-scan high-resolution MS/MS

This is done in a looped fashion across the gradient, and a small number of peptides can be targeted using select Q1 windows

After acquisition is complete, quantitative data can be obtained on each peptide of interest by generating high-resolution extracted ion chromatograms or XICs on the sequence-specific ions observed in the full scan MS/MS spectra

This provides MRM-like data, and the quantitation is obtained by integrating the area under the LC peak of each fragment ion

The peptides of interest must be defined at the point of building the acquisition method, and 30 to 40 peptides can be targeted per run. Recently, time scheduling of the MRMhr workflow has been developed, which extends the numbers of peptides that can be targeted by this strategy

This strategy still provides relatively limited coverage of the sample, however. So, how could we get this quality of data, but in a comprehensive, highly multiplexed way

Well, what if we expanded the width of that Q1 window to be wider, say, 25 Daltons wide, to pass a number of precursors through Q1

These would all be fragmented in the collision cell, and high resolution composite MS/MS spectra would be acquired by the TOF MS analyzer

Now, instead of extracting one peptide from that experiment, we could generate quantitative XICs on a number of peptides that had precursor masses in that window

The specificity would still good because high-resolution MS/MS is acquired, and high-resolution XICs can be generated.

Now, if we take that wider Q1 window and walk it across the whole peptide M-over-Z range, collecting composite high-resolution MS/MS at each step, stepping across the M-over-Z range in each cycle and doing this over and over again across the whole LC run will generate these composite MS/MS spectra on all precursors eluting off the column

Now, we have a data-independent workflow where we don't have to choose our targets before acquisition. We can just acquire all the data with a general acquisition method

After acquisition, we can go back into our dataset in which we've acquired MS/MS on virtually every peptide coming off the column and generate high-resolution quantitative XICs on peptides and proteins of interest

There are a number of key features specific to the TripleTOF 5600 and 5600-plus systems that make them uniquely suited to this type of data acquisition

The combined speed, sensitivity, and resolution of MS/MS obtained on these platforms is currently unrivaled in the market

The MS/MS sensitivity is such that you can acquire spectra as fast as 10 milliseconds per spectrum or 100 MS/MS per second and still get usable data

And at these fast acquisition rates, you can run either in high-sensitivity mode with resolution of 15,000 or greater, or in high-resolution mode with resolution of 30,000 and greater in MS/MS

Finally, as with every quantitative technique, good dynamic range is important, and three to four orders are typically observed

So, what does the data actually look like? Here is a single SWATH acquisition window, 25 Daltons wide on Q1, collected on a 1D nanoLC run on depleted plasma

You can see there are lots of peptides eluting off the column

The bottom pane shows the 3D view of the same data

The horizontal orange box shows the ions passing through Q1 from mass 550 to 575, and is the remaining unfragmented precursors predominantly

The vertical orange bar and the visible strips that we see vertically show the fragment ions generated for those precursors in that 25-Dalton mass window at that specific elution time

So, if we zoom in on the strip highlighted at 18 minutes, we can see the resulting MS/MS spectra shown in blue

Now, the post-acquisition analysis of the data is very targeted. In this case, we want to quantify this ESD peptide from a C-reactive protein

If we have a previously acquired MS/MS spectrum for the peptide, which is shown in pink, we can use that to guide our extraction

We can choose the five most intense fragment ions from our library spectrum and generate fragment ion XICs

Here, area under the LC peak is our quantitative data, and we can generate multiple measurements per peptide

The other key point to mention here as well is the cycle time for the SWATH Acquisition is easily compatible with the LC timeframe

Just like when developing a quantitative MRM assay, the number of data points across the LC peak is important for quantitation, and one can see the good sampling here of this peptide peak

The very interesting thing about this type of data that sets it apart from our standard MRM analysis is that we can go back and re-interrogate the data

Remember full-scan high-resolution MS/MS is being acquired all the time

In this case, we see a potential interference in the y5 ion, shown in orange

It is co-eluting in this case, but if our LC change or we compressed our gradient, it could interfere in some samples

Because we have full-scan MS/MS, we can just re-extract a different ion, the y8 in this case, and improve the dataset

So, for MRM analysis, the assay development is done upfront before you run your sample set. With MS/MSALL with SWATH Acquisition, the sample set is acquired using a generic acquisition method, and then the specific proteins and peptides and fragments are targeted during data processing

So, in the current data processing with the SWATH Acquisition datasets, we are working from predetermined fragment ion libraries

When you know what peptides and proteins you're interested in, you can specifically extract quantitative data

The example here highlights the advantage of the full-scan MS/MS acquired in this data-independent fashion

This phosphopeptide has a two phosphoserine sites, one on y7 and one on y11

We know the Y ions that specifically distinguish these two peptide forms, and we can get specific quantitation on each phosphoform

Interestingly, in a data-dependent experiment, it is possible that the y11 form of the peptide would not have been sent for MS/MS because of typical dynamic exclusion settings

So, obviously, for highest quality quantitation data, you need very high quality and reproducible LC separations. Coupling the Eksigent nanoLC Ultra system and the cHIPLC-nanoflex provide an excellent integrated system for the MS/MSALL workflow with SWATH Acquisition.

Shown here is a SWATH Acquisition dataset, 10 replicate injections of depleted plasma run with a one-dimensional 45-minute gradient

The total ion chromatograms of the acquisitions are overlaid to highlight the excellent LC MS reproducibility

One strategy we have used to measure the quantitative quality of this acquisition strategy is to assess the reproducibility of replicate injections on complex proteomes

Here, the 10 replicates of depleted human plasma were analyzed using the MS/MSALL with SWATH Acquisition

The high-sensitivity MS/MS mode was used such that an average fragment ion resolution of 15,000 was obtained

The ion library that was used contained 150 proteins, 756 peptides, and over 4,500 fragment ions

Using the automated data processing software, these 4,500 XICs were generated using and extraction width of 0.05 Daltons from each data file, basically the equivalent of 4,500 high-resolution MRMs in a single analysis.

The intensity distribution of the resulting fragment ion peak areas is shown on the right-hand side

Next, the cumulative frequency plot for each intensity regime is shown on the left color coded

There's a small number of low-intensity XICs that had higher variance, shown in red, but as expected, the higher abundant peaks had the highest reproducibility

So, even as an aggregate, even including the low-level peaks, the reproducibility on this dataset was very good, with greater than 85 percent of fragment ion XICs having reproducibility less than 20 percent CV, shown by the orange line

Using the 10 percent CV cutoff, we still had about 75 percent of the data that had variance lower than 10 percent CVs. And this is on a targeted assay of 4,500 MRM-like data points

Next, we tried increasingly complex proteomes to understand the multiplexing capabilities of the technique as well as the reproducibility in more complex matrices

Here, three LCMS replicate analysis using MS/MSALL with SWATH Acquisition was performed on more complex proteomes

The statistics of the ion libraries used for the data extractions are shown in the top table

The cumulative reproducibility curves for the three additional proteomes were generated and plotted with the depleted plasma results

Again, using the 20 percent CV point as a point of comparison, the impact of complexity on the data quality can be observed

The most simple sample has the highest reproducibility, and there is a small expected degradation in data quality as we move to increasingly complex proteomes

However, looking back at the ion library statistics, for, say, the human cell lysate sample, you can see that, from an ion library consisting of over 31,000 fragment ions, high-resolution XICs were generated that provide a dataset were 80 percent of the data had quantitative reproducibilities of better than 20 percent CV

This illustrates the tremendous extent of the multiplexing possible with this technique and the quality of the data that can be obtained

Again, because of the high-resolution full scan MS/MS that is being acquired all the time, the specificity of the technique remains high

Another strategy that was used to assess the power of this quantitative technique was to compare the dynamic range of the SWATH Acquisition technique with that obtained from an MS1 experiment

Here, a set of standard peptides were dosed into an N15 labeled yeast digest across a range of concentrations

The limits of detection for the two techniques were then compared

A large set of peptides was analyzed. The data shown here is for one peptide but was pretty typical across the dataset

The SWATH Acquisition strategy provided good quantitation at lower amounts on column than the TOF MS strategy because of the extra specificity provided by the Q1 isolation over the MS1 quant strategy

So, where do we see this workflow fitting into the bigger protein research picture

The current discovery strategies will still play a role in providing an initial picture of the biology and a view of the possible proteins of interest that we need to study in our system

And the end goal in many cases will be to develop a high sensitivity targeted MRM assay for use in large-scale studies

We see the MS/MSALL with SWATH Acquisition strategy as providing a powerful new tool in the early verification phase, where a highly multiplexed assay on a large protein panel needs to be applied to a larger numbers of samples

This acquisition strategy should streamline the progression of studies from discovery to validation, as the SWATH Acquisition technique will provide an extremely straightforward transition to an MRM assay

But, obviously, this is early days in understanding all the power of this strategy and the potential of this workflow, providing benefit in targeted discovery or being used more broadly downstream is possible

The last thing I would like to cover is a little bit of detail around the MS/MSALL Workflow with SWATH Acquisition micro-application in PeakView Software

This slide shows the general differential protein expression workflow

For the first generation of this workflow, we recommend that people collect both an IDA file on a sample or some of the samples in parallel with collected data using SWATH Acquisition across all the samples

The IDA file will be processed in ProteinPilot software and will serve as the fragment ion library to drive XIC generation

Next, this ProteinPilot group file, which is the ion library, and the SWATH Acquisition data files will be loaded together into the MS/MSALL with SWATH Acquisition microApp in PeakView software

After all the XICs have been generated, all the data will then get packed into MarkerView software for results interpretation

For more information on MarkerView software, please refer to the AB SCIEX Website. I don't have time to discuss in detail today

So, this is the main data analysis pipeline. However, we realize that not all researchers will require this exact workflow

So, we also designed the software to act as an XIC generation engine, where users could provide their own inputs and extract data in a generic form for downstream processing

So, in terms of input, there were three main ways to get ion libraries or fragmentation data information into the software

Through ProteinPilot is one way. Through the industry-standard mzIdentML format is another way if other search engines are used to process the IDA data. And finally, there is a text format input that can be used to bring in the ion Library

After data processing, as driven by the ion library provided, there are a number of options for results export

Processing by MarkerView as discussed, the peak areas can also be output as a text export for use in other downstream processing tools, like analyzing in Excel. And finally, false discovery rate analysis can be performed by exporting the data in mProphet format

Finally, the ion library used for the processing can be exported as well for storing or refining and loading back in

With these generic interfaces, many additional workflows leveraging the MS/MSALL with SWATH Acquisition workflow will be possible

The data processing interface is shown here. As you can see, all of the protein information and peptide information from the input format is shown on the right-hand side

Proteins and peptides to process can be selected, and assay parameters like numbers of peptides per protein and numbers of fragments per peptide can be defined

The data files are also loaded in and shown on the right-hand side, overlaid so the LCMS reproducibility can be assessed

The user can then set extraction parameters, which will depend on the retention time reproducibility obtained and MS/MS resolution obtained

Fragment ions for each peptide are extracted, and the correct retention time for peak integration is determined using the embedded algorithm

Manual review of the data is possible by clicking on a peptide and viewing the XIC data and the full scan MS/MS data as shown here

On the left is the XIC data with all the fragment ions overlapping. On the right is the both the ion library spectrum or experiment MS/MS spectrum, shown in pink, and the full-scan MS/MS from the SWATH Acquisition shown in blue

Once all the parameters are set according to the experimental requirements, XICs for the entire dataset are generated automatically

In conclusion, the MS/MSALL workflow using SWATH Acquisition is an exciting new workflow that has the potential to really change the way we approach our quantitative proteomics projects

It is now possible to collect in a single injection a dataset that has both comprehensive ID and quant of all components detectable in the sample

Using a generic acquisition strategy, one can obtain high-quality MS/MS quantitation on all species. And this MS/MS quantitation will typically have better dynamic range and reduced chance of interferences than the more traditional MS1 strategies for global profiling

What is most exciting is that the quality of the quantitation is actually more comparable to that obtained by MRM analysis on the triple quad or QTRAP instrument, but with much higher multiplexing possible

With MS/MSALL with SWATH Acquisition, one is effectively creating a permanent record, a digital archive of everything in the sample that can be re-interrogated over and over again with different proteins and peptides as new hypothesis or new algorithms arise

Finally, I'd like to acknowledge that the team that was involved in the development of this workflow, the R&D team at AB SCIEX involved in this project, and also the scientists at ETH working in Ruedi Aebersold's lab that we've been collaborating with during the development of the SWATH Acquisition workflow

Thank you