Recovery problem with SER, LYS, ARG, with AccQ-Tag/Waters

[agmatinase]

(former Amino acid analysis of complex culture media)

we're doing since a while HPLC-AAA with the AccQ-Tag method of Waters (detection with UV@248nm).
Now it seems that there's a methodic problem:
When we sent our samples to another lab where they quantify the AAs with the Ninhydrin-method and they get reproducibly very different values for some AAs.
Here the amounts we get in [%] of the Ninhydrin method:

ASP (Asparaginic acid): 96%
GLU (Glutamic Acid): 106%
LYS (Lysine): 135%
ARG (Arginine): 144%
SER (Serine): 162%

So incredibily too high results with the AccQ-tag method for some AAs!
This is totally reproducible.

I have three possible explanations:

1) the first two AAs above have more or less the same value with both methods, the last three are inconsistent. LYS, SER and ARG have additional to the amid group "chemically active" groups (LYS: Amin, SER: Hydroxy, ARG: Guanidino). So it might be possible that these react as well with the reagent (a carbamate). But why should this happen only with the sample and not with the AA standard solution (pH correction of sample doesn't change the result)?!

2) the complex media contains lots of peptides. When the sample gets mixed with the borate buffer (pH=8.6), this may generates AA out of di-, tri-,..peptides which then would increase the amount of the specific AA.
With the Ninhydrin method as a post column reaction this couldn't happen.

3) The Ninhydrin method has problems with the quantification of these AA


Thanks for any comments!

achim

[Bryan Evans]

This method is gaining popularity for amino acid analysis of complex mixtures:
http://www.ncbi.nlm.nih.gov/pubmed/15386506

A fully automated amino acid analyzer using NBD-F as a fluorescent derivatization reagent
Biomed Chromatogr. 2004 Nov;18(9):630-6

[Uwe Neue]

If you send me an e-mail with your telephone number, I have somebody contact you to troubleshoot the issue.

Uwe_Neue@Waters.com or Uwe.Neue@Prodigy.net