Thread by @Sci_j_my

Extremely excited to share our work published in Nature Methods!

The first example of shotgun proteomics without liquid chromatography, enabled by ion mobility.

Wonderful collaboration with @nieminm @Pagliarini_Lab @Coon_Labs

@naturemethods nature.com/articles/s4159…

Thanks to my collaborators for making this possible and making it awesome.

Thanks to the reviewers for their reasonable comments that made the paper better.

This technique forms a central pillar of my labs research

SEEKING POSTDOCS to work in on all aspects of this eco system (applications, data collection, data analysis, and what to do with thousands of proteome profiles we produce?!)

After much deliberation, the name of the method changed several times over review.

We went with:

Direct Infusion Shotgun Proteome Analysis (DISPA)

Please omit the hyphen and pronounce "dis-puh"

The central idea is laid out in Figure 1a: we replace liquid chromatography with ion mobility, and we scan with a DIA strategy to get fragments (never any precursor measurement). We identify peptides with the projected spectrum concept (Figure 1b/c)

This idea is based on a computation analysis I did showing complexity reduction in human peptide mass distributions after fractionation by FAIMS and quadrupole selection (Extended data figure 1)

Extended data figure 2 may give you a better idea of what is going on by showing infusion traces and a lack of chromatography

What orbitrap resolution, Q1 isolation width, and ion injection time should I use? I performed parameter scouting experiments testing a grid of these conditions, the concept of those experiments are showing in Extended data figure 3

the result from these parameter scouting experiments are shown in figure 2A-2C. Small Q1, longer injection time, and higher resolution are all best. Note in 2b that addition of FAIMS tripled the peptide identifications.

Although we only find about 500 proteins in this first iteration, as shown in Extended Data Figure 4, many of those proteins are very important for metabolism, cell structure, and protein synthesis/degradation

We got reviewer comments about 10 days before I left to start my lab. In total, reviewers comments could be addressed by 4 new experiments. One reviewer asked if it was reproducible. I knew it was but we hadn't shown it. I went to lab Sunday and setup 100 runs extended data Fig 5

Another reviewer asked if we could analyze one of the most difficult sample matrixes - human plasma. I ordered some from sigma and compared to frozen plasma we already had. Yes, we can identify over 100 proteins from human plasma, and in theory this can be done in about 30 sec 🤯

Extended data figure 7 details a standard mixture experiment I used to demonstrate quantitative qualities, and supplemental figure 3 shows generally how that works by co-isolating heavy and light precursors to get quant from fragments

We generated a targeted quantification strategy for only the most intense peptide from the 500 or so proteins we identify (concept shown in Extended Data figure 8)

the quant from DISPA is not perfect (Figure 3b/c), as it shows ratio compression at extremes, but ratios do generally reflect mixing ratio

We devised some use case experiments: @nieminm isolated mitochondria from 293T cells and we showed that we can quantify many proteins from this simplified system, and that the quant agrees very well with values from LC-MS (Extended data figure 9)

Many mitochondrial proteins appeared to be lower in cells with PPTC7 knocked out, some of them statistically significant (extended data figure 10)

This is the most exciting part: we showed that DISPA can screen cellular responses to toxins very quickly. Over 45k protein quant (redundant) in about 4.4 hours. #UMAP of proteotypes segregates toxin treatments in a logical way (fig 4)

This project was tons of fun, and it was a pleasure working with Natalie Niemi, Dave Pagliarini, and Josh Coon to get it done!

Fun fact, all four authors are professors

Want to try DISPA?

I have someone full time right now working to make the data analysis integrated and easy. We should be done in about a month. Watch for a preprint early next year.

Link to the article on readcube if you need access:
rdcu.be/cbcdC

Recent or future PhD graduates: go here to learn more about the lab and apply for a postdoc role:

jessemeyerlab.com

@threadreaderapp please unroll