Protein 'purity' by HPLC

[SSGG]

How does one quantitate the 'purity' of a protein solution by HPLC?


With a reference standard, I can calculate the amount of protein present in terms of mass/vol (e.g. mg/ml).


But what I don't understand is how I can come up with a results in terms of '% purity', if I do not know what other impurities (nucleic acids, lipids, other proteins, etc..) might be in my sample.


Can I sum up the peaks? What wavelength do I use? Don't the impurities have different detector response factors?


Thanks

[tom jupille]

You can use the HPLC results to tell you the concentration of protein in your sample in mass/vol units. You prepared the sample, so you know the total concentration in mass/vol units. The ratio (expressed as a percentage) is the purity of your sample (% protein).

[SSGG]

Tom,

If I understood your reply correctly, I can calculate the purity in percentage terms if I actuallly weighed the (solid) sample. However, the sample is in the form of a purified liquid from our bioreactor.

I think there are indirect ways, such as finding out the total protein concentration using a colorimetric method, and using an ELISA to find out how much of the total protein is actually the protein of interest.

I was just wondering if there was a more direct way using HPLC. I have been told that I can just sum the peaks from GPC, but I don't understand the principle behind this.

How can I be sure that, for example, 90% of total peak area in the chromatogram is equivalent to 90% purity? What about the different kinds of impurities and their dfifferent response factors?

[tom jupille]

You can't be sure, as evidenced from the parallel thread on mass balance for small molecules:
http://www.sepsci.com/chromforum/viewtopic.php?t=2744

You could, as an obvious example, have a huge percentage of something like NaCl, which you would never see with UV detection.

However, my original suggestion still stands. Take an aliquot of that original reactor broth and lyophilize it to get the dry weight.

[SSGG]

Thanks a lot, Tom. That's a v. good thread.