Wetenschapsforum

Beste Chemieforummers,

Ik ben bezig met een vergelijkend onderzoek van peptidenexpressie in de hersenen van honingbijen. Het principe is dat ik peptiden label met een label dat ofwel 128 D zwaar is, ofwel 137 D (met het verschil dat in het tweede negen deuteriums zitten).

Nu, het labelen gebeurt bij hoge zoutconcentratie, maar op de dingen daarna in een massaspectrometer te analyseren, moet ik de boel wel ontzouten.

Ik ben al enkele weken bezig met pogingen tot ontzouting, maar er loopt vanalles mis. Gisteren stuurde ik een mail naar mijn begeleiders, maar niemand wist raad. Ten einde raad zou ik de mail hier graag posten. Hij is in het Engels, en hij is lang, maar het kan een uitdaging zijn voor iemand die iets kent van analytische technieken.

Alvast veel dank voor het neme van de moeite om het ding te lezen!

Groetjes

Rik

About desalting the labelled peptides.

In the course of the last weeks I have encountered a lot of problems with the desalting of the samples. I want to give a comprehensive overview of everything I did so far, so maybe together we can get to answers.

The aim of the study is to compare the brain peptides of honey bees under two conditions. Therefore I extract peptides from the brain, I label them with TMAB, a label that covalently binds to the peptides, and I analyse them using mass spectrometry. As the labels differ in isotopic composition, I should be able to compare peptides derived from two samples, within one mass spectrum.

Peptide extraction

During the entire procedure of peptide extraction, samples were kept on ice as much as possible. First, 100 µl acid methanol (90% methanol, 9% H20, 1% acetic acid) containing 75 picomol of the peptide standard was added to the samples. Tissues were then homogenised by sonication at 0° during 15 seconds. After centrifugation, (14,860 rpm for 10 minutes, 3°C), the supernatans was brought over to a fresh tube. The pellet was resuspended in 100 µl of acid methanol and sonicated and centrifuged again. After adding the second supernatans to the first, they were centrifuge-filtered on a filter membrane (0.5µm pore) for 4 minutes, 12000 rpm at 3°C. Finally, samples were vacuum centrifuged until they were dry and stored at -20°C.

Labelling

I labelled the peptides with trimethyl-ammonium butyric acid chloride (TMAB) that binds covalently to amino groups of the peptides (i.e. the N-terminus and the ε-amino group of lysine). If two samples are to be compared in one nano-LC MS run, one sample is labelled with a light label (128 Da) and the other with a label containing 9 deuterium atoms (137 Da). The labels were produced by the research group of professor Fricker (Yeshiva University, New York) and stored under dry conditions at 3°C. As TMAB is unstable in the presence of water, contact with air is to be avoided as much as possible and TMAB should always be dissolved in water free DMSO.

The labelling procedure was executed at room temperature. In a first step, the dried peptides were brought back into solution using 100µl 5% acetonitryle, 0.1% TFA. Samples were vortexed for 10 seconds and bath-sonicated for 5 minutes. Next, 200 µl of aqueous disodium hydrogen phosphate (0.5M Na2HPO4) solution was added, and the pH was adjusted to 9.5 with 1M NaOH. Samples were again vortexed and sonicated to prevent the peptides from precipitating due to the elevated pH.

Next, 4 µl of 1.5 M TMAB in water free dimethyl sulfoxide (DMSO) was added to the extract. After 12 minutes, the pH was adjusted to 9.5 with 1M NaOH, followed by 12 minutes of incubation. This procedure was repeated another 6 times.

After 7 cycles of adding label and adjusting the pH, the samples were incubated for another 2 hours at room temperature. After two hours, 30 µl of 2.5 M glycine was added in order to quench the excess of TMAB. The pH was adjusted to 9,5 and samples were incubated for 40 minutes.

After quenching, 3 µl 2M hydroxylamine (NH2OH) in DMSO was added in order to remove labels bound to tyrosine, serine or threonine residues. Again, NaOH was added to adjust the pH to 9.5. This procedure was repeated after 10 and 20 minutes. At this stage, the samples were stored at -80°.

Now the problem is to get rid of the salt...

For mass spectrometric analysis, the samples should be concentrated (by the end of the labelling procedure, they are dissolved in about half a ml of aqueous salt/glycine solution containing DMSO and a small amount of acetonitril) and desalted.

In a first experiment, we used Pierce PepClean™ C18 spin columns to desalt the samples. Spin columns are just put into a centrifuge tube, and are loaded with volumes of 150 - 200 µl. The columns are spun for one minute at 1,500 x g. Everything that flows through the spin column is collected in the tube and stored in the fridge for later investigation if something goes wrong.

First, the column is equilibrated 3 times with 200µl 70% acetonitril, 0.5% TFA. Next, it is rinsed 3 times with 200µl 5% acetonitril, 0.5% TFA. Now the sample is loaded onto the column in portions of 150µl. The flowthrough is collected and reloaded an additional time to maximise the peptide retention. After all the sample has been loaded, the column is rinsed again 3 times with 200 µl of 5% acetonitril, 0.5% TFA.

Finally, the sample is eluted 3 times in 50µl 70% acetonitril, 0.5% TFA. I know in the protocol it is stated that two times with 20µl is sufficient, but I want a maximal recovery, and I don"t think that an extra bit of eluens can do anything wrong.

The first times we used this technique with labelled peptides, we never got a signal on Nano-LC Q-TOF mass spectrometry. Then we realised that it might be due to the high pH of the samples. We decided to analyse the waste of the desalting. Of course we still had to desalt and concentrate the waste before we would be able to analyse it. Therefore I used an Oasis™ HLB cartridge. This is a non-silica reverse phase desalting bed in a plastic syringe. It has the advantage to be able to cope with larger volumes, and to be applicable in a wider range of pH.

First, the sorbent is equilibrated with 1 ml of methanol. Next, it is rinsed with 1 ml H20. Now, the sample is loaded and the flowthrough is collected. After all the sample has been loaded (in the case of the flowthrough of the first experiment, it was a lot), it is rinsed one last time with 1 ml of 5% methanol in water. Finally the sample is eluted, first in 750µl methanol, then 750µl acetonitril. I used two different eluents to make sure that peptides with a wide range of solvent affinities would all elute well.

When we analysed this "waste"with nano-LC Q-TOF, to our surprise we found perfectly labelled honey bee peptides. We also found a rather severe polyethylene glycol (PEG) contamination. Anyway, the labelling works, but there is something screwy going on with the desalting.

We thought we had the answer: just make sure the pH of the samples is not too high and use Pierce PepClean™ spin columns. Nevertheless, we had lost some confidence in the Pierce columns and decided that HLB was maybe just a better bet. I made a mimic of the labelling conditions (i.e. all the salts, glycine, DMSO, but no TMAB) on 4 leftover brain peptide extracts and desalted them with HLB cartridges, as described above. For two samples, I did not adjust the pH (so it was around 9.5), and for the two others I adjusted it to 4. When after a while the 1,5 ml of eluate was vacuum dried in the speedvac, I put resuspended the samples into 5µl 70%ACN, 0.5% TFA and put 1µl of each sample on MALDI. I got a signal, although not a very strong one (actually I was rather disappointed). Also I saw some contamination but I cannot remember its mass now (I"ll check it tomorrow). Finally, I did not see any difference between the high and low pH samples. It is important to consider the fact that these peptide extracts were leftovers of earlier experiments, so they were old and did not contain the equivalent of 10 brains. For fresh extracts of 20 brains we would expect a stronger signal.

Nevertheless we decided to use Pierce Spin Columns anyway.

We started all the labelling procedures and labelled 9 times 2 samples. As we are cautious and suspicious people, we decided to first try one sample, control it on MALDI, and then desalt all the other samples.

So, I thawed 2 samples, sonicated them for a while, pooled them in a fresh tube, and adjusted the pH to 4 with about 20µl of TFA. I followed the Pierce PepClean™ Spin column procedure as described above, and vacuum dried the samples in the speedvac. I resuspended them into 5 µl of 70% acetonitril, 0.5% TFA and put one µl on MALDI. The result was interesting: I had no signal at all.

This is puzzling: we thought it was the high pH, but all of a sudden it appeared to be something else. I decided to analyse the waste flowthrough again on an HLB cartridge, assuming there was just something seriously wrong with the spin columns. I followed the procedure as described above, and vacuum dried the 1,5 ml of eluate. Later that night, the sample was finally dry (it took a while indeed), I resuspended again in 5µl 70% acetonitril, 0.5% TFA and put one µl on the MALDI plate. Again, the result was interesting. Quite a lot of contamination, no peptide at all. I went home and had a cup of tea.

This morning, I took a last chance on HLB. Maybe I did something wrong during the pooling of the samples. I thought I maybe lost the peptides because they were not well suspended in the high pH solution. I now adjusted the pH to 3,5 in the two samples apart, before I pooled them. I sonicated them for several minutes, pooled them and started the HLB desalting. I also diluted the sample with one extra ml of H20 (you never know the salt concentration was just too high for the column to cope with).

Something interesting to mention: when I eluted the samples (into 750µl of methanol followed by 750µl of acetonitril) I observed a white band-like contraption in the test tube. I don"t know what it was, but it seemed to be peptide or something like that (well, it could have been peptide I think). I don"t know how it ended up there in a small disc, it was really bizar. It was also not in the middle, but almost at the bottom of the tube (near the 0.1ml line). I now regret I did not take a small amount of the band to analyse separately.

Anyway, I vortexed the eluate (so the band was gone) and I vacuum dried it in the speedvac. I resuspended it as usual and put in on MALDI. Surprising result: no signal at all.

This is where I am now. I have little idea what to do now. I honestly believe that the peptide extractions were successful, and I trust the labelling (it worked in the past). So the question is: where are the peptides?

Possible answers:

1. Sticking to some test tube: seems implausible because we have already seen peptides in earlier experiments and we used the same tubes.

2. Still in the waste water: for one or another reason they just always flush through columns. Also implausible because it worked one time with the HLB.

3. Lost during vacuum drying?

4. They are really present in the eluate, but I cannot see them on MALDI.

5. The burglar who stole Nele"s bag also stole my peptides.

The fourth possibility might seem to be as ridiculous as the fifth, but I have a reason to consider it plausible. If nothing would be present in the eluate, one would expect nothing to be present in the matrix on the MALDI spot either (except for some TFA maybe). But, if I put just pure alpha matrix on a spot, and I shoot on it, I see a lot of matrix peaks, whereas in the "samples" I don"t get any signal. No signal at all indeed. Yesterday I witnessed this complete absence of signal in the spin column sample. Today I observed it in the HLB sample.

If you now assume that for one or another unknown reason, the TMAB labelled peptides interfere with MALDI (in the end we did never ever see labelled peptides on MALDI before), so that no signal at all is generated, but the energy is absorbed by something, resulting in the absence of matrix peaks. That something could be the label, and the absence of signal could be due to the fact that only the label (or a fragment of the label) is accelerated, but it is too small to be detectable for MALDI.

That would also explain why we could observe contaminations in the HLB sample yesterday, but not today. If we assume that pierce as well as HLB work, then yesterday no labelled peptides were present in the HLB sample (because they were present in the Pierce sample), but today labelled peptides were present, resulting in no signal at all.

I don"t state that this hypothesis is true. I just cannot think of any other explanation for the things we observe.

Ik heb zelf ook een hoop quantitative proteomics gedaan. Ik ben geen fan van de pierce producten en spin cardridges. Ik zou ze gewoon helemaal niet gebriuken en meteen voor Oasis HLB gaan. Hoe minder stappen hoe beter het werkt.

Verder geeft nano LC ook altijd een hoop problemen. Dus zorg er altijd voor dat je eerst checked dat je systeem werkt. Wat je ook kan doen is gewoon je sample met zout injecteren op een kolom. Als je echt bang bent voor de zouten in je MS, dan laat je gewoon die kolom een half uurtje flowen als je hem niet in de massa spectrometer hebt zitten op 1% organisch. Dan de kolom terug monteren en je gradient starten.

Verder als je je maldi matrix niet eens ziet, dan is er natuurlijk iets mis met het MS systeem. Maak de juiste controles. Spot ook meteen ff een BSA digest ofzo.

Hoeveel samples wil je analyseren?

@ Napoleon: dankjewel voor de suggesties

Ondertussen ben ik een pak opgeschoten.

Eerst over de MALDI: ik heb een pak controles gedaan hoor. Standaard peptiden zijn perfect zichtbaar, MALDI alfa cyano 4 hydroxycinnamic acid matrix geeft matrix-peaks. Mijn staal niet.

Conclusie: het probleem zit niet in de MALDI, maar in mijn staal.

Zoals ik dus zei, zijn we ondertussen wat opgeschoten. Ik heb de stalen geanalyseerd op Nano-LC-ESI-Q-TOF en de gelabelde peptiden zijn zichtbaar. Grote opluchting hoor. De vraag blijft nog een beetje hoe het komt dat ze niet op MALDI zichtbaar waren.

Mogelijkheden:

1. Het label bevat een groep die laser-energie absorbeert, in-source decay ondergaat, waardoor het peptide niet versneld wordt. De groep word wel versneld, maar is zo klein dat hij buiten de detectie range van de TOF valt.

Vandaar ook geen matrix peaks (energie wordt geabsorbeerd).

2. Er zat nog te veel zout / andere rommel in het staal (vetten uit de hersenen?) die er voor zorgden dat er geen correcte kristallisatie plaatsvindt.

Vandaar geen matrix peaks (matrix is slecht uitgekristalliseerd).

Het volgende plan is om de negen stalen die ik nu nog heb, op te zuiveren op HLB, af te dampen in speedvac (tot bijna droog), eventjes aan te lengen met genoeg aangezuurd water, en dan te ZipTippen.

Sounds like a plan, denk ik.

Prettig weekend.

Rik

DAt is goed nieuws. Waarom nog ziptippen?

Om er zeker van te zijn dat het geen onzuiverheden zijn (zoals vetten) die de HLB overleefd hebben. HLB heeft een goede recovery, maar is niet heel erg specifiek in wat er uit het staal verwijderd wordt (zout meestal wel, maar andere contaminanten durven al wel eens achterblijven). ziptip is de ideale manier om van die laatste rommel af te raken.

Dat ben ik niet met je eens. HLB is inderdaad weinig selectief, maar dat is je C18 ook niet. Ik mag aannemen dat je de methode niet hebt geoptimaliseerd specifiek voor je ene peptide, dus je elueert met hoog organisch, en dus heb je ook hier geen selectiviteit. Verder is de ziptip gewoon goedkope silica en krijg je al die nare silanol interacties die analiet verlies veroorzaken.

Je nano-LC kolom is ook C18 of C8 waarschijnlijk dus je doet in weze dezelfde scheiding 2x. Ik zou die ziptip stap gewoon overslaan.

Hmm, interessant.

In ieder geval zijn ze er niet blij mee als ik een 'vuil' staal door de nano-LC stuur.

Ondertussen heeft iemand anders ook een beetje met mijn stalen zitten spelen en de spectra zien er effectief minder rommelig uit nadat ze nog eens geziptipt zijn.

Ik begrijp ook niet heel goed waarin HLB zo hard verschilt van C-18. Misschien is het -ongeacht welk sorbent je gebruikt- onvoldoende om één keer te ontzouten?

Groetjes

rik

Nee dat is genoeg om te ontzouten, anders moet je gewoon even meer was oplossing er door gooien. Maargoed iedere stap die je doet zal je minder interfering peaks zien, maar je zal ook minder peptides zien.

Ziptip is een silica gebaseerde C18 kolom.

HLB is een pyrollidone styrene polymeer. Het heeft dus niet de negatieve eigenschappen van een silica gebaseerde kolom en mijn ervaring is dat het veel beter werkt voor voornamelijk polaire peptiden. Waarschijnlijk ziet je nano-LC chromatogram er voornamelijk veel schoner uit in het begin, wat gewoon analiet verlies betekend.