Hello

I'm not sure how quantitatively accurate deconvolved CD data is - some people say that it is accurate whereas others say that it's either a (very) vague estimation of structural content or not accurate at all. I have always been told not to overinterpret my data, hence, I have done my deconvolutions only for high quality SRCD data (also for a series of mutants) using the CDSSTR algorithm at DichroWeb and I have always compared the "magnitude" of secondary structure elements to results from other methods (like crystal structures). To compare whether there are differences between mutants I have always just overlaid my spectra and checked for differences by eye. Sort of kept things qualitative.

Perhaps this article could be helpful http://www.ncbi.nlm.nih.gov/pubmed/17406547

I would like to suggest that you convert the CD signal into the MRE for each of your mutant protein and then compare the spectra. I anticipate single amino acid mutants should have overlapping spectra. If they are overlapping then you don't have to do all this unnecessary work. However, interpretation about the structural content from the CD spectra is not very accurate and should be used as gross signature.

I hope this will help you.

I would prefer using the method proposed by Wu et al. 1981 : http://www.ncbi.nlm.nih.gov/pubmed/6111337 ,

it calculates helical contet based on the spectrum , but it is sufficient for structural comparison. I would suggest that > 10% differences in the content suggesting the conformational changes between WT and mutant.

Why dont u confirm the near-UV for your WT and its mutant? depending on the degree of mutation, if it is only single mutation the far UV tends to be similar to that of WT. Better to compare both far- and near-UV , as the mutation may changes only local conformation arround the mutation site (in case you are lucky enough to have some aromatic residues located nearby the site that can be detected by near-UV).

As Jörg pointed out, accurate concentrations are essential. I wouldn't put too much faith in the absolute values of secondary structure estimates from CD deconvolution programs, but comparing a wt to mutant should be okay. How much difference are you seeing? And how big is your protein? The size of the protein is important - if you are seeing, for example, 50% helix in the mutant vs. 55% in the wt protein, that's likely not significant for a small (say ~100 residue) protein, but may be significant for a large (e.g. 600 residue) protein. Again, there is always the caveat of errors in the concentration measurements.

3-5% sounds approximately right provided:

1. The concentration is accurately measured as suggested

2. It is not a membrane protein

3. It is well-folded without large disordered regions. Largely disordered regions can adopt unusual secondary structure elements such as very short helices, 310 helices, and PII helices, that are not accurately accounted for in the program.

To make an accurate comparison of structure upon mutation, you can try another method besides far-UV CD that is more sensitive to tertiary structure. One is near UV CD as suggested. Other possibilities include radius of gyration measurements or fluorescent dye binding.

https://www.researchgate.net/post/In_the_absence_of_a_complete_high_resolution_structure_what_are_the_best_experimental_approaches_to_validate_my_3D_model_of_a_large_protein

I agree - comparing the shape and amplitude of the CD spectra on the absolute scale (molar residue ellipticity) is, in my experience, much more informative that spectral deconvolution (which introduces errors of its own). Computing a difference spectrum between WT and mutant protein also helps to see what specific structure (if any) has been lost upon mutation. More importantly, protein stability is much more sensitive to mutation-induced changes than the structure at ambient conditions. Often you don't see much difference in far-UV CD but the melting data show significant difference in Tm or cooperativity upon strategic mutation. I suggest you record such melting data (220 nm is usually a good choice) to compare your WT and mutant proteins. Finally, near-UV CD can be informative to probe for changes in the aromatic packing which often report on the overall protein fold. Hope this helps!

PS Generally, 5% difference obtained by quantitative data analysis is within the error margin and can be ignored, but larger differences can be significant.

Excellent points above, and to what Olga said, measuring the melting temperature is a nice way of uncovering mutations that alter stability. Melting data acquisition can be done using CD but also using simpler methods like ThermoFluor, which in my experience has always worked and provided repeatable results.

Moreover, you could test the stability of your proteins by titrating samples with a chaotrope and following denaturation via CD.

I agree with the points made above. I also tend to not overinterpret single spectra. It is a powerful technique for comparative studies e.g. to get an indication if certain mutations or buffer conditions have a major effect on the structure of your protein. To compare different samples it is important to be as accurate as possible with your concentrations and experimental conditions. I also recommend to calculate the mean residual ellipticity to normalise your data and to allow a comparison of different samples. I always found the Kelly et al., 2005 paper very helpful (http://www.ncbi.nlm.nih.gov/pubmed/16027053).

Dear Trevor Creamer

I have made 3 single mutations and 1 double mutation from an alpha/beta hydrolase. secondary structure prediction by program like PSIPRED showed that two of the sigle mutations have small changes in their structures. The protein have 292 amino acids. I am seeing 17.1% helix in WT vs. 22.7%, 22.6% and 19.3% helix in mutants. This differences could be significant?

As you said, I know it is not an accurate estimation.

Thank you from all of you because of your helpful suggestions. Some of the additional experiments that mentioned by you, is not executable for me at present time. I just want to know my CD data can be helpful for me? I attached the result of my CD spectra both from 190-250 and 200-250 (to remove noise). And it was based on molar ellipticity. If you have any more suggestions about them, please inform me.

Thank you again