Validation of peptide map

[SSGG]

How do I validate a peptide-mapping assay, without an MS?

The chromatogram supplied with the reference standard is slighltly different from mine, as I am using a similar, but not identical column.

Will the chromatogram of my product, if it is exactly the same as the reference standard, be acceptable to the FDA?

[tom jupille]

I would think yes.

[SSGG]

thanks Tom. but how can I be sure that there are no peak co-elutions?

[tom jupille]

You can't. But presumably you have the same co-elutions in the reference standard and in your sample. If the chromatograms are truly identical (same number of peaks, same retention times, and same area% values, then you can argue that the proteins are identical.

I'm assuming that the chromatogram that came with the reference standard had at least the major fragments identified, and that your separation conditions are similar enough that you can identify the same fragments (by area% and retention order, if nothing else).

If you're really concerned about it, get hold of the column used by the vendor to run the reference standard.

[SSGG]

Another question : why do people specify such a high concentration of TFA in the mobile phase in the separation of proteins?

Inevitably, I cannot match the concentration, because when I check the pH, it is always ~1.0 to 1.5, which is too low for my columns.

Anyway, here is the reference chromatogram from Eur Pharm (I've arbitrarily numbered the peaks 1-18. There should theoretically be 21 peptide fragments):



And here is my sample (the number of peaks has increased, and the labelling is arbitrary):



Since I am using a different mobile phase (lower TFA), and a different column, can I just stop my method development here, and not try and improve the separation further? (As long as it matches the reference standard, as Tom said)

Thanks

[HW Mueller]

One does peptide mapping to get at the amino acid sequence of a protein? And then the FDA accepts something like this reference chromatogram as being sufficient to do this? Obviously there are unidentified substances there, how does the FDA, or anybody else, put this puzzle together?? With "libraries" where the blind lead the lame?
Also, I always wondered, even when MS is used, how analysts know for sure that their starting protein is homogeneous (all molecules of the protein ensemble have the same amino acid sequence)?

[Daren]

SSGG,

When you run the reference sample on your method do you get the exact same chromatographic profile as your sample? That is what you need to show. In addition I would continue the development of HPLC method to achive baseline separation of all your peaks (if possible). The point of the test is to accurately ID your protein, if there are varying peeks or lack of resolution to discern this then I would question if you are accurately identifying your protein.

[SSGG]

Thank you Daren and Mr/Ms Mueller.

1. Baseline separation - I don't think I can. After all, much better aanlysts have probably worked on this, and the reference peptide map was the best that they could achieve.

2. Blind leading the blind - don't know about the FDA, but I am certainly pretty clueless. For example, if there are only supposed to be 21 fragments, how come there are more than 21 peaks (counting all the unresolved peaks with the naked eye) in the reference chromatogram?

3. Homogeneity - if most of the protein molecules in the sample have the same amino acid sequence, won't the peptide map be quite consistent? Whereas, if there was a genetic shift in some of the producer cells, and 10% of the proteins have a mutated sequence, presumably I would see different peak ratios, or even an extra peak or two. Can anyone help with this?

TQ

[Kostas Petritis]

Although I can not reply what the FDA needs in this case, I will provide some information on peptide mapping by MS.

First of all, the tryptic digestion is often lead to partial tryptic and sevaral miss-cleavages (i.e. several argininines or lysines in the peptide). In general it is not uncommon to see up to 3 miss-cleavages. I would add that in proteomics, about 40% of the peptide identifications are from peptides with at least one miss-cleavage. Anyway, that of course would give additional peaks in the chromatogram.

Now about the assingement of the peptide sequence by LC-MS, in general you need to know what you are analyzing (i.e. the sequence of the parent peptide/protein). That is why in proteomics, the DNA of an organism needs to be sequenced before any serious work start. So if you know your protein's sequence, and there are not variants, that is great, you would be able to assign almost all the peaks by LC-MS-MS and the adequate software for data intepretation (i.e. Sequest, Mascot, Spectrum Mill, X Tandem etc...).

Now if you have modifications in your protein and you do not know what these are, there is still hope as you can try to do de novo sequencing. In this case and LTQ-FT would be perfect as you can use the high mass accuracy information and narrow down the possibilities.

For example, I noticed that in one one of the commercial proteins from Sigma, one of the peptides has a deamidated asparagine (mass shift of 1 Da)....

[HW Mueller]

But narrowing down the possibilities is not knowing exactly. It´s difficult to do chemistry if you assume the structure of the molecule with which you are dealing. Thus monoclonal antibodies (Mab) were considered to be one type of molecule until RP-HPLC was done on them, showing many peaks. The result is what we saw (only one of many examples) when we tried to locate intestinal tumors, in vivo, with an iodinated Mab: The tumors/metastases were barely visible while the gonads showed very strongly. (Probably this is not only due to the molecular inhomogeneity of the Mab, but olso to that of the antigen (molecule that is supposed to function as marker of tumor cells in this case).

[SSGG]

Kostas,

Thanks for your reply. Is mis-cleavage still possible with DPCC- or TPCK-treated trypsin (which is better, in your opinion)? And if so, will the mis-cleavages be consistant?

If the mis-cleavages are not consistant, won't I get a slightly different chromatogram each time?

By the way, it's amazing what you guys are capable of; detecting a difference of 1Da...

Fortunately, I am not in R&D, just QC. Hopefullly, if there are any problems with the primary sequence, I can just return it to manufacturing and let them deal with it.

So my main concerns are how to go about validating the peptide-mapping assay, what acceptance limits to set, and so on. Does anybody have any experience with this?

Mr/Ms Mueller

Was the Mab inhomogeneity caused by different polypeptide sequences? Or different glycoforms?

[HW Mueller]

It doesn´t look like anybody checked into this with these "humanized" Mab.

(Hans is a he)

[Kostas Petritis]

Hans,

You narrow down the possibilities with the accurate mass and then identify it with the information from MS/MS. Also with de novo you do not need to have any information about the protein...

Although there are some weakness (i.e. difficult to deal with post-translational modifications). that is the way that most of the people are doing proteomics (shot-gun proteomics). That is the reason that there is some interest in intact protein analysis, but there are some more problems associated with it.

SSGG,

We use Promega trypsin which is TPCK treated. For a single protein the results would be more or less reproducible but you will have partial tryptic and miss-cleavages...

Differences of 1 Da is not that impressive but it all boils down to your mass accuracy (i.e. it would be impressive if you could differenciate it in the protein level but not that impressive in the peptide level...). With an LTQ-FT, in the data direction mode you can achieve sub-ppm mass accuracy which really narrow down the possibilities.

In your case, just find out what the FDA approves and go with it..

[Mark Tracy]

For protein drugs, the FDA ultimately defines the product by its manufacturing process, not by chemical tests of identity. All these chemical tests (and there are many) are used to demonstrate that the manufacture has not changed.

This is good for consumer safety. It is also powerful market protection for an established manufacturer, and a high barrier for generic drug makers.

[HW Mueller]

Now, it seems to me that if your protein is not sequence homogeneous to start with you have lost out with any method to get at its (their) structure.

If one works with such inhomogeneous molecules it should not be surprising to get the low specificities mentioned in my example above. (In my opinion, all the materials I have seen being used in nuclear medicine are insufficient in this respect, with the possible exception of iodine which is highly concentrated in the thyroid. Note that I-131 might not be totally homogeneous either in the sense that other isotopes may be present, but at least chemically it´s homogeneous).