Question about HILIC column again

[squall]

Hi All,

Last time, I posted a question about phosphopeptide identifications. At that time, I am using a cell digest, so it's hard to know what peptides are really there, if I don't get good LC-MS/MS data.

Recently I got some "phosphopeptide standard", which I know what is in my sample. First, I tried C18 (LC-MS) and I saw most of the peaks. Then I switch to HILIC column (Amide 80 HILIC), surprisingly I almost lost most of the peak. The problem is not that I got bad peak shape, etc. It's "the peaks are gone".

Actually I got similar problems before which I really can't understand. Currenly I suspect that whether my sample is really dissolved or not. I have 20pmol per phosphopeptide sample, for C18 test I dissolve them in 40ul water with 0.1% formic acid, then I inject 1ul sample for LC-MS analysis. For HILIC, I first add 4ul water(with 0.1% formic acid) to the sample tube, using pipet to mix it. Then I add 36ul ACN (with 0.1% formic acid) inside the tube, sonicate and votex, then also inject 1ul for LC-MS (on-line HILIC MS).

Then I didn't see any peaks. Ideally the HILIC should work better than C18, right? Even the Amide 80 might not be as good as ERLIC for phosphopeptide, how it can lose peaks even from C18?

Is my sample preparation method in HILIC not proper? I really suspect whether the sample got dissolved in this 90% ACN solution. But I saw most papers use 90% or even higher ACN for sample solution, in order to avoid bad peak shape.

Before this sample, I have another single peptide tested on this column using HILIC condition. That works in 90% ACN. But since that sample I have a lot, so I use almost 1ml water to dissolve first, then dilute.
Does this give any information for the above one?

The other question I have is that I always see references adding ammonium salt in HILIC condition. But for the test I did before using simple peptide and small molecules mix, I really can't see any difference adding or not having ammonium salt inside. Is that salt really useful? If yes, why my data shows no difference with regular buffer (water or ACN with 0.1% formic acid)?

Sorry for the long question, I have struggled in this for a long time. Hope I can find some help here.

Thanks a lot

[Andy Alpert]

Squall:

A lot of tryptic peptides do have trouble with solubility in 90% ACN. My colleagues and I are exploring this solvent for ERLIC of complete tryptic digests, and we find that we have to use a much larger volume of the solvent to get all of the digest into solution than is the case with 85% ACN.

Here are a few more suggestions:
1) Presumably the phosphate groups have negative charge under your conditions. That means that they will have a cation as a counterion. If you only use unbuffered formic acid as an electrolyte in your solvent, then the counterion will be whatever your phosphopeptide was exposed to during your process. Give it a counterion that will promote solubility in 90% ACN: triethylammonium ion. Specifically, instead of 0.1% formic acid, prepare a stock solution containing triethylamine formate at pH 3.0 (by adding triethylamine to a solution of formic acid in water) and then use an aliquot of that to prepare a solvent that's 90% ACN and contains 15 mM formate ion overall (not just in the 10% of the solvent that represents the aqueous portion).
2) It's possible that the particular Amide-80 column you have has "hot" frits that are adsorbing low concentrations of phosphopeptides. Passivate the frits (and the rest of the metal parts in the column) by eluting it overnight at a low flow rate with 40 mM EDTA.2Na. Check the pH of this solution to make sure that it's not higher than 7.2.

[squall]

Thanks so much Andy! I can try the triethylamine formate later. But does this one supress the ionization efficiency for mass spec?

[Andy Alpert]

I expect it will, to some extent. However, are you running directly into the mass spec from the HILIC column? Anyway, you need to get the phosphopeptides off the column before you worry about the ease of their detection.

Incidentally, why are you starting with 90% ACN at all? The phosphate group makes a peptide unusually hydrophilic, and phosphopeptides usually elute around 70-40% ACN during a decreasing ACN gradient from a HILIC column with a good, thick hydrophilic coating. You can load them nicely in 85% ACN. Solubility of tryptic peptides is appreciably better in 85% ACN than in 90%.

[carls]

Are you detecting in ESI+ or ESI-? If you are using ESI I would try ammonium formate pH 3.2, 10mM (actual concentration in mobile phase, e.g. 10% of 100mM buffer solution). As Andy mentioned you should be able to go to 80-85 v/v% ACN as your initial mobile phase composition since these compounds show unusually strong retention in HILIC. You also may need to run the column at 50-60C to sharpen up the peaks. The best column I have seen for phosphate compounds is Merck/Sequant ZIC-pHILIC which is a polymeric (not silica based) HILIC column.

[squall]

Thanks again Andy and Carls. I will try 85% or 80% soon.

Carls, the ammonium formate you mentioned is for sample solublization or the mobile phase?

Back to the ammonium formate mobile phase question. I know almost everybody use it, but the real strange thing for me is when I tested a few simple std mix (several peptide mix or small molecule mix), why I didn't see obvious difference of adding or not adding ammonium salt? Or it will help for certain compounds, but might just not have great effect on the molecules I tried?

Thanks

[Andy Alpert]

If your standard compounds aren't electrolytes, then inclusion of an electrolyte in the HILIC mobile phase and sample solvent won't make a big difference. If they are electrolytes, then it does make a difference. Specifically, nearly all HILIC materials have some electrostatic charge. You need about 20 mM salt in the mobile phase to titrate the charged groups on the surface of the stationary phase and get retention per an analyte's polarity without having significant electrostatic attraction or repulsion superimposed as a mixed-mode effect. Read my paper on the subject:
http://pubs.acs.org/doi/pdf/10.1021/ac070997p

While this paper deals mainly with repulsion effects, it also documents attraction. So do a good number of papers by other authors.

Another reason to include salt in a sample solvent is to make sure that all the molecules have the same counterions for their charged groups. Combinations of different counterions differ in polarity. In the HILIC mode, this will result in tailing peaks or even multiple peaks for a pure solute (counterion exchange with the mobile phase is surprisingly slow under HILIC conditions). Again, see the above paper for an extreme example.

If you want a more detailed response to your question, then supply all the details about what your analytes were, what the sample solvent was, and what column and mobile phases you were using.

[carls]

Thanks again Andy and Carls. I will try 85% or 80% soon.

Carls, the ammonium formate you mentioned is for sample solublization or the mobile phase?

Back to the ammonium formate mobile phase question. I know almost everybody use it, but the real strange thing for me is when I tested a few simple std mix (several peptide mix or small molecule mix), why I didn't see obvious difference of adding or not adding ammonium salt? Or it will help for certain compounds, but might just not have great effect on the molecules I tried?

Thanks

Try adding 10mM ammonium formate pH 3.2 to your mobile phase. Ensure the concentration in mobile phase is 10mM, for example add 10 v/v% of a 100mM buffer solution to ACN. Actually its a better practice to add the ACN to the buffer to ensure the water remains at least at 10%.

As Andy mentioend and I stated in a different recent post on HILIC, the effect of salts in HILIC mobile phase depends on the ionic character of your analyte in HILIC mobile phase. Different salts can for example give different selectivity. Unfortunately the salts amenable to MS detection are limited and thus likely will not provide major differences in selectivity.

[squall]

Hi Andy and Carls, I tried again my sample using 85% and 80% ACN to dissolve the sample, actually I even use extreme conditions like 50% and 100% water (these definitely dissolve the sample because I got signal on C18). I also prepared the mobile phase having 10mM and 20mM ammonium formate inside.

I tested first with some simple peptide mix and BSA digest (all dissolved in 80% ACN), they all show peptide peaks. But when I start using the phosphopeptide mix, then almost no signal is observed. I am really frustrated about it.

Here is some example phosphopeptides in my sample: pTKLIpTQLRDA[K], EVQAEQPSSpSSP[R], ADEPpSSEESDLEID[K], ADEPSpSEEpSDLEID[K], FEDEGAGFEESpSETGDYEE[K]. They all have decent signal in C18 column, but not appear in Amide 80 HILIC or just mix together with noise so I can't distinguish them.

The column I used is Amide80 HILIC and it is fine to identify BSA peptides and other standard peptides such as Bradykinin.

I tried several mobile phase: such as just 0.1% formic acid in ACN and water. Also getting 10mM and 20mM ammonium formate in both mobile phase, I also prepared the mobile phase such as B: 90%ACN+20mM ammonium formate+formic acid(adjusting pH to 2-3), A is high water percentage with ammonium formate and formic acid.

Based on these discription, do you think I did something wrong that making the phosphopeptide not showing up? Or my conditions should work, there is something else might cause the problem?


Thanks

[squall]

There is one thing I forgot to mention that I am not using trap column, so everything should be injected into the separation HILIC column.

[Andy Alpert]

Hello again, Squall,

I hope that you have as much ACN in the sample solvent as in the starting mobile phase. If you dissolve the sample in a 100% aqueous solvent, then you must add 4.5 vol. ACN to the sample after that in order to match a mobile phase that starts at 85% ACN. If you have much less ACN in the sample solvent than in the starting mobile phase, then that is equivalent to injecting a sample on a C-18 column when dissolved in 70% ACN. Chances are that it will elute in the void volume. At best, the peak shape will be terrible.

At this point it's worthwhile ascertaining if your Amide-80 column is eating phosphopeptides. Elute the column overnight, at a very slow flow rate, with 40 mM EDTA.2Na. Next morning, flush out the EDTA solution, equilbrate with solvent A, and run a quick gradient to solvent B. Reequilibrate well with solvent A and then try a run with a phosphopeptide standard.

You shouldn't need more than 85% ACN to get good retention of any of these phosphopeptides, and even 80% would suffice.

[squall]

Hi Andy,

I actually tried using 90%ACN, 80%ACN to dissolve my sample. The starting mobile phase I used is around 90%ACN. Do you think that could be problem? I guess not.

As for the columns. Besides Amide80, I also used a bare silica one, both of them are giving similar data. Maybe they all have "hot" frits inside? Although I feel the chance could be low, but I may give a try later since I can't think of any other possible reasons.

Thanks again for all your help

[Andy Alpert]

Don't be so quick to dismiss the solubility issue. We've been playing around with 90% ACN in connection with ERLIC of tryptic digests. You need about 10-20x more of it to dissolve such digests than 80 or 85% ACN. Maybe you just never got your phosphopeptides into solution. Try my suggestion of dissolving the digest or your standards in a purely aqueous solution and then adding 4 vol. ACN, slowly, with mixing.

Uncoated silica is fine for analysis of a lot of simple pharmaceuticals by HILIC. It is not fine for tryptic peptides and larger polypeptides. There are good reasons why people bother to put thick, hydrophilic coatings on silica for HILIC applications.

Once again: YOU DO NOT NEED 90% ACN TO GET GOOD RETENTION OF PHOSPHOPEPTIDES IN HILIC! Use 80 or 85%.