Hi Prem

the question is wrong, but the answer is NO; it is generally not (I am very close to say "never") possible to determine the (small part of) a gene sequence of an "unknown", purified protein.

Here are my thoughts:

Let us assume: an antibody recognizes an epitope on a protein; there are many ways an antibody can recognize an epitope on a protein - a) this does not need at all to be a consecutive sequence of amino acids, i.e. the epitope could be part of two or more different sites of the 3-dimensional sequence of amino acids just forming together the recognizable part b) the antibody could theoretically detect an epitope of a well definded and known sequence of amino acids, but due to the 3-dimentional structure of proteins, amino acids from different parts of the sequence in the protein could also form such a recognizable epitope - and then lack any relation to the original sequence c) the antibody could react with several, but slightly differering epitopes (amino acid sequences) - how to tell the differences if a reaction is strong d) even if the amino acid sequence could be detected, one could only guess several possible DNA sequences (might be useful for PCR primers).

So, could you mention the example you might think of?

This question isn't clear. But perhaps you are referring somehow to phage-display technology? You can screen for phage encoding a cDNA that reacts with an antibody of interest. This will enable you to recover nucleotide sequence associated with an antigen of interest, without determining information about protein sequence first.

I agree with daniel, the question is not clear. Do you mean interaction in WB or other assay? Is it polyclonal or monoclonal antibody? If antibody in monoclonal, and you know against which epitope antibodies has been design, you can theoretically get a protein sequnce. Goog luck!

Hi Prem,

I too don't understand well your question. Anti DNA antibodies do exist but the epitope that an antibody can accommodate in its binding site is quite small. I am not sure what do they recognize. Even if you manage to isolate sequence specific antibodies they will not tell you about the linear nucleotide sequence. This is not similar to Mass spectrometry where you can identify protein from peptides mass.

Hi Prem

the question is wrong, but the answer is NO; it is generally not (I am very close to say "never") possible to determine the (small part of) a gene sequence of an "unknown", purified protein.

Here are my thoughts:

Let us assume: an antibody recognizes an epitope on a protein; there are many ways an antibody can recognize an epitope on a protein - a) this does not need at all to be a consecutive sequence of amino acids, i.e. the epitope could be part of two or more different sites of the 3-dimensional sequence of amino acids just forming together the recognizable part b) the antibody could theoretically detect an epitope of a well definded and known sequence of amino acids, but due to the 3-dimentional structure of proteins, amino acids from different parts of the sequence in the protein could also form such a recognizable epitope - and then lack any relation to the original sequence c) the antibody could react with several, but slightly differering epitopes (amino acid sequences) - how to tell the differences if a reaction is strong d) even if the amino acid sequence could be detected, one could only guess several possible DNA sequences (might be useful for PCR primers).

So, could you mention the example you might think of?

Dear All, as this question was asked in an interview in a university entrance. That is why I put this question here. Exactly ! for me too it is not clear. But thanks to everybody for your valuable views. Thanks once again.

Thanks, Prem, for clarifying. It might be a good idea for a university to ask such questions to see how a candidate may react with a (what I called) "wrong" question; although I don't know the real intentions of the originator of the question (the university), this type of question might be used to test not only the knowledge of a student, but also could stress the diplomatic and politeness potential, or show how focused or distractable a candidate is.

Robert- the technology may be old, but it doesn't mean it isn't true. Please see

"Bacteriophage lambda display of complex cDNA libraries: a new approach to functional genomics" PMID: 10669604

cDNA expression libraries used to screened routinely with antibodies to find a gene of interest. In many cases, the nucleotide sequence was not previously known.

The technology never mapped the precise location of the epitope. But this aim is not part of the original question.

So while the question is worded awkwardly, I doubt it is a wrong-headed, improbable, or trick question.

Daniel- thanks for your point.

I also thought in exactly the same direction with several rounds of lambda phage libraries... and that was basically the reason why I mentioned "it is generally not (I am very close to say "never") possible to determine the (small part of) a gene sequence of an "unknown", purified protein."

But at the end this is different to the question. Even if there was an antibody specific for a (rather small) amino acid sequence, then the specific binding of that antibody to an "unknown" protein could be due to either that amino acid sequence, or to the tertial structure just aligning these or similar amino acids. Even in the unlikely first case the redundancy of the genetic code could give just several estimates.

There are (very few) antibodies which can detect specific versions of a protein resulting from single point mutations which lead to different conformations. In such a case an antibody could determine the (or estimate) the nucleotide sequence of a "known" protein (but the question was on "unknown protein").

I think there is just some semantic confusion here. An unknown gene in this context simply means that the nucleotide sequence has not been determined. The question pre-supposes a purified protein against which a specific antibody has been raised.

Robert's answer is spot-on in terms of requiring that the cDNA encode sufficient sequence to enable the epitope to fold properly (assuming it is a non-linear epitope), and that you cannot deduce nucleotide seqeunce without either knowing the protein sequence OR having a cDNA expression library as part of your toolkit.