Pain and joy of quantifying phosphorylation sites | Proteomics |

The mission was to relatively quantify phosphorylation sites on a protein after in vitro treatment with a panel of kinases. I had lots of spectra from previous work on the Orbitrap which had identified about 18 sites. (Some sites were considered to be ambiguous; I was sure that some peptides as a whole were phosphorylated, but unsure of the exact position as any of some neighbouring residues could be phosphorylated.) Luckily for me, an extremely talented post-doc had made lots of protein and done an excellent experiment: I had tones of material to work with.


The methods: Of course I used Skyline (MacCoss labs) to generate MRM methods of the Xevo from the Orbitrap data. I also wanted to test out positive/negative switching to see if I could identify new sites by the loss of -79 from phosphorylated peptides in nESI. The Xevo comes with a fancy predesigned template for this type of method; it combines polarity switching, with a quick precursor scan, then some MRM, then a product ion scan, before cycling back. Exciting stuff! I could not wait to get started.


The Pain: The pre-designed ‘phosphopeptide’ method appears to consistently* crash the mass spec such that it requires a full electronics re-start. Of course this happened at about 6 pm on a Friday night and to do so one needs to go round to the back of the instrument and put the vacuum pumps into ‘override’ so that you don’t vent the mass spec and add two more hours to your evening’s work. Then the electronics may be power cycled and the system restarted. OK, this is not a hard thing to do, but it is daunting with a new instrument and really not helped by small, hard to access switches (in our lab I have to wedge myself between two benches, try not to tread on the roughing pumps, nor crush or pull out any gas lines or cables and not knock over a gas cylinder). These elusive little sliver switches are shown in the picture above and the critical information – which leaver is the override and which turns off the electronics? – that information is finely engraved into the sliver panel by the green lights. Nice design job. Sure this is not something a user should do every day. But still occasionally we have to and believe me the adrenaline pumped as I made my choice. Getting is wrong would not be a flat-out disaster, but is something one should try to avoid.

*To be fair I only did this twice, but with two full system crashes on a Friday night with a recovery time of at least an hour, I called it a night. I have shared all this data with Waters, who advise ‘make the method simpler’. Sure enough, methods using ‘survey scans’ triggered from either -79 or the neutral loss of 98 do work.


The Joy: Once I abandoned the new fancy method and returned to MRMs the work went well. Out of my 18+ candidate sites, 10 could be reliably observed by MRM. I also monitored many unmodified ‘proteotypic’ peptides, and other peptides with and without oxidation of methionine and with various miscleavages. Why bother with these last two things? Well, phosphorylation sites are very inconsiderate and will often occur with additional modifications (oxMet being a common one) and their peptides frequently have ragged ends (sites where trypsin has two close choices RK and the like) or phosphorylation itself can influence how a peptide is cleaved. Many quantitative proteomic experiments avoid such modified or miscleaved peptides, but there the objective is generally to quantify the protein and use the peptides as surrogates. Here I was reasonably sure I had similar amounts of protein and was after relative change in phosphorylation status.


Well it worked, three replicated injections of each sample across the panel of 8 kinases and 10 phosphorylation sites showed that yes – even with a generous in vitro incubation of kinases and substrates - some sites are specifically phosphorylated.

This work was tremendously exciting for me, not just getting nice data from the new instrument, but seeing theory (that certain proteins may act as ‘hubs’ to integrate signals from many different pathways) come alive in my hands. Well that’s the sort of feeling that makes science, and late Friday nights, worthwhile. Of course we need to do more biological replicates (two down) and pull together a few more things, but the essence is there and the ‘oh gosh it actually worked!’ is enjoyable. Next challenge will be to move from in vitro to vivo…

Via alex