A couple of ideas - one is that CHCA matrix, even commercially available matrix, can "die". We have had many cases where individual gel spots spotted from investigators labs using commercial CHCA gave very poor spectra - we overlay with our CHCA matrix (which we re-crystallize ourselves once a year and store as powder at -20°), and voila, much better spectra. For automated spotting like what you are doing, we make the CHCA solution up fresh everyday (7 mg/ml recrystallized CHCA, 2 mg/ml ammonium phosphate, 0.1% trifluoroacetic acid, 80% acetonitrile), and C18 separate each SCX fraction into 370 spots.

2nd, it is important to calibrate your instrument minimally the same day that you are running your spots - we automatically run a plate and MS/MS calibration at the start of each new plate using 13 cal spots we put on the sample plate. It only adds a few minutes per plate, but improves the database matching step considerably.

Not sure what instrument you are using, but 0.5 Da mass tolerance for the MS Parent peaks would be very high for any of the MALDI TOF instruments we use (particularly when using the just-in-time calibration steps above), and using such a wide mass tolerance means that your search space in Mascot is expanded greatly. Similarly to trying to search for too many PTMS, anything that greatly expands your search space (the number of different possible mass matches in your database for any particular spectrum) will negatively affect Mascot's (or most search engines) performance, as a higher score (higher quality spectrum) is required for a "confident match". This is because there are so many more possibilities for random matches with a larger database search space. Again, this will depend on your exact instrument, but we use a mass tolerance of 50-150 PPM for the MS spectra, and 0.4-0.6 Da for the MS/MS peaks (the tightness of the tolerance band is much less important for the MS/MS spectra, assuming you have used a stringent parent mass tolerance).

Using a 2D-LC separations and MALDI as described above, we can typically confidently ID 500 - 3500 proteins per crude trypsin-digested sample (FDR 1%).

Try INSPECT http://proteomics.ucsd.edu/LiveSearch/ with the target protein database. It worked for me for few spectra which were not identified using MASCOT.

The default MS tolerance we are using is 0.5 Da. Upon another look of the spectra, we found that the S/N is not very good - well at least not as good as the literature. Some reports were doing MS/MS for peaks with S/N > 100, but in my case most peaks were well below this threshold. I know the publications are not to be trusted completely...still my guessing is that the quality of our MALDI-TOF/TOF spectra is not ideal.

I have read articles regarding the use of ammonium salt to suppress matrix cluster, but not many LC-MALDI articles are using this method. Any thought on this?

Hi,

for the MSMS I would check the calibration of your MSMSs. I am using the UltrafleXtreme from Bruker and I can say that every time I do not get great MSMS results is due to the LIFT not well calibrated (indeed, if I increase the fragment tolerance up to 1.2Da I get results!)

Mascot calculates the MOWSE score (=“probability of matching N peaks by random chance”) but is useful for scoring tandem mass spectra with high-accuracy (< 10 ppm) fragment mass data, where the significance of each matched peak is high. If your signals are not significant, they are probably discarded.

If I was you, I will try to play with Mascot parameters and in case it does not work, my ideas are :your starting material is too low, or the recovery from HPLC is low or your instrument set up is not so sensitive.

Finally, if you wonder about matrix, I recommend to add ammonium citrate for avoiding matrix clusters or just switch matrix (DHAP>DHB are more salt tolerant than CCA).

Hope it helps

Stella

Hi, Did you consider your peptide might be modified ? What kind of modifications do you allow in the search ? If your modification is not selected for, Mascot will fail.

Greetings

Bart

Re Stella: it happens that our instrument is also from Bruker (Autoflex Speed). When you said "LIFT not well calibrated" do you mean the mass accuracy or the laser power? After talking with the engineer he also agree that mass accuracy is the key, but still we are struggling with the S/N and spectral quality (at this moment). The major peaks are there, but MASCOT just ignore them, probably with good reason, I don't know. The matrix we are using is saturated CHCA in 30% ACN, diluted ten-fold before mixing with LC eluate. Most literature use CHCA in 70% ACN. Probably that is the direction we are going.

Starting material and LC recovery should be OK, judging from the chromatogram and MS1 spectra.

Re Bart: at this moment we only considered standard modification like IAA derivatization and Met oxidation. We also included carbamylation of Lys just in case for the presence of urea.

A couple of ideas - one is that CHCA matrix, even commercially available matrix, can "die". We have had many cases where individual gel spots spotted from investigators labs using commercial CHCA gave very poor spectra - we overlay with our CHCA matrix (which we re-crystallize ourselves once a year and store as powder at -20°), and voila, much better spectra. For automated spotting like what you are doing, we make the CHCA solution up fresh everyday (7 mg/ml recrystallized CHCA, 2 mg/ml ammonium phosphate, 0.1% trifluoroacetic acid, 80% acetonitrile), and C18 separate each SCX fraction into 370 spots.

2nd, it is important to calibrate your instrument minimally the same day that you are running your spots - we automatically run a plate and MS/MS calibration at the start of each new plate using 13 cal spots we put on the sample plate. It only adds a few minutes per plate, but improves the database matching step considerably.

Not sure what instrument you are using, but 0.5 Da mass tolerance for the MS Parent peaks would be very high for any of the MALDI TOF instruments we use (particularly when using the just-in-time calibration steps above), and using such a wide mass tolerance means that your search space in Mascot is expanded greatly. Similarly to trying to search for too many PTMS, anything that greatly expands your search space (the number of different possible mass matches in your database for any particular spectrum) will negatively affect Mascot's (or most search engines) performance, as a higher score (higher quality spectrum) is required for a "confident match". This is because there are so many more possibilities for random matches with a larger database search space. Again, this will depend on your exact instrument, but we use a mass tolerance of 50-150 PPM for the MS spectra, and 0.4-0.6 Da for the MS/MS peaks (the tightness of the tolerance band is much less important for the MS/MS spectra, assuming you have used a stringent parent mass tolerance).

Using a 2D-LC separations and MALDI as described above, we can typically confidently ID 500 - 3500 proteins per crude trypsin-digested sample (FDR 1%).

I have a question for you, have you got some sequences that don't match with any peptides in databases by manual assignment?

Re Bruce: we have adjusted the formulation of our matrix (from Bruker) using higher organic solvent (70% ACN). When working with BSA we can achieve approx. 40 ppm mass accuracy with parent mass, and below 0.4 Da with MS/MS. It also gave me some nice MS/MS signal that can be matched in MASCOT. But whenever we try to do the sample, again MS/MS is not working. Sometimes the S/N is too low, sometimes the peptide does not fragment. Since the BSA result told me that our mass spec and matrix is working properly, I guess it has something to do with my sample prep.

The engineer told me that peptide fragmentation is compound dependent. It sounds right, but I don't think it is that simple.

Re Ahmed: all the manual assignment result failed in database search. I guess this is not the right approach....a mass difference of any amino acid does not necessary mean that there is "really" an amino acid there.

Yes I understand you, but I work in some proteins that are absent form databases (the protein existe physically but there is no evidence on the genetic level. Yes, it is crazy, but it is true!!), and using Maldi TOF and the public databases we were not able to identifie any of them. However, by using de novo sequencing we obtained peptides of these proteins, I don't know if you have got some petides, but I can give you the sequence of one of my proteins (see the link bellow) and try to check manually if there is some of your peptides that can be aligned on it.

http://www.uniprot.org/uniprot/P85173.fasta