I want to analyse plant free amino acid content, probably by a pre-column derivitisation method such as AccQ Tag, or OPA.

I expect a new peak: a methionine homologue which is not commercially available. I will need to verify its identity, or find it if it is co-eluting. Therefore I figure that MS identification is the best approach (I have access to an LC-MS, and an LC-(MS/MS)).

I want to find out what kind of solvents I should use, and what kind of MS conditions would be good for finding my amino acid (bearing in mind there's going to be the derivative tag attached).

Any help would be very gratefully received. I have little experience in this field!

Ben;

OPA derivatives of amino acids are normally separated in reversed-phase columns using NaOAC or KH2PO4 buffers at nearly neutral pH. Consult with Agilent for more details about columns and conditions.

You will need to change to a buffer suitable for MS, perhaps NH4OAC or similar.

Good Luck

Benjamin

Hi Ben,

I agree with you that the LC-MS is the best approach. However, you do not need to derivatize the amino acids as they are readily detectable by MS (ESI ionisation). Derivatization is needed only for conventional detectors (UV, fluorescence etc) as most of the amino acids do not contain strong chromophores or fluorophores. They are though nowadays several commercial detectors that can do the job without any previous derivatization (see Petritis et al. J. Chromatogr A 961 (2002) 9-21).

Back to your question:

The best way to separate your underivatized amino acids (in respect with the mobile phase additive MS compatibility- see LC GC Eur. 15 (2002) 98-102) is with ion-pairing chromatography or hydrophilic interaction chromatography (HILIC). For ion-pairing chromatography you should use volatile perfluorocarboxylic acids (from heptafluorobutyric acid -chromatographic limit- up to pentadecafluorooctanoic acid -volatility limit)and silica or carbon based hydrophobic stationary phases. With HILIC you can use low water/high ACN/formic acid and columns like the polyhydroxyethyl aspartamide from PolyLC (I think they are others but this is the one I have use). I have achieve separation of these amino acids with chiral columns like Teicoplanin T(AG) from Astec but this is a very expensive solution if you are not interesting in chiral separations.

Most of these results have been published... see Petritis et al.
J. Chromatogr. A 833 (1999) 147-155
J. Chromatogr. A 855 (1999) 191-202
J. Chromatogr. A 870 (2000) 245-254
J. Chromatogr. A 913 (2001) 331-340.

The above deals mainly with LC development and use ELSD or MS as detector.
About the use of MS in simple or tandem modem for the underivatized amino acid analysis you may see:
Petritis et al.
J. Chromatogr. A 896 (2000) 253-263
J. Chromatogr. A 896 (2000) 335-341
J. Chromatogr. A 855 (1999) 191-202

On November we sumbitted a manuscript which deals with an improved method for LC-MS-MS analysis of underivatized amino acids. This method can analyse 76 amino acids in less than 20 minutes and uses 1 mm C18 column. Among the amino acids we analyse Selenomethionine as well, if ever that is the modification you search.

Furthermore, if you want at all means a derivatization step before LC-MS (personally I do not see any reason), I recall having seen the AccQ Tag amino acid derivatizes to be analysed in that way but never the OPA.

Well that was long but I hope it helps,

Kostas

Thank you very much for your advice. Kostas- I was planning derivitisation as a way to quantify my aas. That way I can do several experiments in one go: identification of the novel peaks (if any) then quantification of them, and of the standard amino acids. I dont think this is so easy with an underivatised method.

Ben

Ben,

I do not quite agree with you. What could derivatization offer you from the moment that you can already detect your analytes? In general only problems (especially for the OPA).

There is though something that you gain if you derivatize amino acids and analyse with MS and that is a little bit more sensitivity. But from my personal experience I had enough sensitivity and was able to quantify amino acids in several matrices like plants, biological fluids (plasma etc) and agricultural products (wine, beer, tobacco etc). Some of the above have already been submitted for publication or they are already in press.

Furthermore, the MS/MS instrument I used was not the most sensitive of the market (API-300 from Sciex). You may find now MS/MS instruments that are about 100 fold more sensitive (i.e. API-4000, Micromass... etc.).

Shortly my point is that quantification is always easier with an underivatised method, from the moment you can detect your analytes.

Kostas

Kostas, Having access to a LC/MS and having it available all the time may be two different things

Anonymous,

I agree but:

1) He already spoke for identification of an unknown compound so he will need to use the MS for that,
2) He didn't speak about the need of doing routine analysis and though use the MS for extenting periods of time and
3) I just expressed my opinion, he has the last word on this after all.

Maybe someone else will have a better approach to face Ben's analytical problem.

Regards,

Kostas

Thanks to everyone who contributed to this thread. I'll leave my conclusions for anyone else who is attempting something similar.

I have decided to use the Waters AccQ Tag derivitized AAA system. Probably using an Agilent 1100 LC, and HP 1100 MSD. After a lot of searching I came full circle, and found what I needed buried on the Waters website: http://www.waters.com/WatersDivision/waters_website/Applications/Lcms/amd35.pdf

Waters have adapted their aaa procedure for use in LC-MS (it doesnt appear perfect- for example they need to use ESI in both positive and negative mode to rule out molecular ion CIDs).

This suits me very well because:

1) AccQ Tag derivitisation gives every aa the same response factor. Therefore you can quantify an aa for which you have no standard. (this would be the main drawback of any non-derivitisation method)
2) MS will allow me to perform diagnostics on any new peaks, and the fluorophore tag is robust enough for clean molecular ions.
3) The method should not require development- someone else has done all the hard work.
4) Last and not least, a colleague has the accQ tag column and kit on site ;)

Wish me luck!

Ben, could you elaborate on your point 1)? How could the response factor be the same? Do you mean in fluorescence?
Still, in my experience the only reason for derivatization, if detection did not require it, was to stabilize a compound (for instance, ascorbic acid). Usually, though, derivatization ensues a destabilization, etc. etc.

Hans

Yes, I mean the fuorescence response.

Each aa has the same tag, so the UV response peak for each aa can be directly compared. This is easier for quantification than relying on some inherent UV property (whose strength may vary between different aas) as you rely on the tag (whose fluorescence response is the same- we assume- irrespective of the aa it is attached to).

I have done some analysis and quantification of a class of small molecules based on their inherent UV property. This is great, but one compound might give a much higher response than another compound at the same concentration (for structural reasons).

That means I use an internal standard of known concentration in each run. The IS has been calibrated against standards of each compound I expect to see in a run, so then I can work out the concentrations of everything in a run relative to my IS (whose absolute conc. is known).

Derivitization can introduce problems, but as I want to quantify an aa for which no standard is obtainable the derivitisation method is ideal.

Ben

Ben,

I'm glad you found something that suites your needs. I am only going to comment your statement on February 1, 2003:

"Therefore you can quantify an aa for which you have no standard. (this would be the main drawback of any non-derivitisation method)".

In fact that is not correct. There is one detector that can do very well this job (quantify without having standards) for all nitrogen containing compounds. This is the chemiluminescence nitrogen detector (for liquid chromatography) and it has been shown to be an equimolar detector (the only one in liquid chromatography)in respect with the nitrogen concentration in your molecule.

The reason this detector has an equimolar response has to do with its operation mode. Briefly, the HPLC column effluent is nebulized with oxygen and argon or helium and itnroduced into a pyrolysis tube heated to 1050 C. In an oxygen-rich envoronment the mobile phase and the solutes are oxidized into oxides. All the nitrogen-containing compounds except from diatomic nitrogen are converted to nitric oxide (NO). Nitric oxide is then mixed with ozone to form nitogendioxide in the excited state (NO2*), which reapidly decays to the graound state, releasing a photon (hv). A photmultiplier tube tehn detects the photon emission. The detected signal is dirctly proportional to the amount of nitrogen in the original sample.

You may find the use of this detector for the analysis of underivatized amino acids in Petritis et al. LC GC Europe July 2001 389-396.
I had observed an equimolar response for the amino acids with an RSD of 6.3 % (a part of the RSD should be due to weighting/standard solution making/dilutions etc).

This not only means that you can quantify your amino acids with single calibration curves; your standard do not even have to be an amino acid but just a nitrogen containing compound (e.g. caffeine). Co-eluted compounds that do not contain nitrogen do not interfere with the quantification of amino acids. You can even directly calculate the stoichiometry of peptides (after hydrolysis) of known MW without the need of any standard (see above article).

The evaporative light scattering detector can do also pretty decently this kind of job see Chaves-das Neves and Braga-Morais:
Anal. Quim. (1997) 93, 98-101.
J. High Resol. Chromatogr. (1997) 20, 115-118.

Good luck again with your research and keep us posted how your method worked.

Kostas

UPDATE on AAA LC-MS

I have now used the modified waters protocol for MS coupled amino acid analysis. Its worked very well. I used a ACCQ derivitisation of acid dissolved MeOH extracts, a methanol/water gradient, a C18 column, and ES+ in a Finnegan ion-trap MS.

I did not find what I expected to find. But I did find some very interesting changes in my mutant plant ;-)

There were several 'unknown' UV peaks. These could be id'ed using the ion-trap and turned out to be non-protein amino acids, and also the secondary metabolite dirhamno keimpferol. Proof of principle.

I have not quantified using the DAD signal as it wasnt strong enough for the low abundance amino acids (changes in the main aas were very easy to quantify). This means I'm going to transfer to straight HPLC with FLD(X10 more sensitive)for quantification using the ACCQ Tag. So, the main reason I used the AccQ-Tag system (quantification of unknowns)is redundant. But if I'd had LC/FLD/MS-MS....!!!

We will be experimenting with Kostas Ps direct AAA method to look for other metabolites, when we get the carbon column. This will be better- faster, cheaper, provide more information. Unfortunately we dont have the facilities to quantify changes we see using it, for example by using kostas' chemiluminecent N detection (other than by relative ion abundances between two samples).

Ben

Hi Ben...
All of this aside, have you looked into Phenomenex's EZFaast Amino Acid Analysis Kit?
It looks pretty interesting (15 minute prep and run time!!) and I wanted to see if anyone had any first hand experience.