Let us go back in time to around 1975. It is my understanding that at that point (or at least by the end of the decade), both genomic and cDNA libraries had been created for a few organisms, e.g. rabbits. I'm interested in the logic of techniques that first used these libraries to isolate genes encoding for particular proteins. Note: I'm not interested in how we'd isolate a gene today; I want to understand the history.

First, consider the following thought experiment. Say it's around 1975. The one assignment for my Cal Tech biology class is a doozy: it goes as follows. The instructor (Tom Maniastis?) hands me a protein along with the cells from which the protein was isolated as well as genomic and cDNA libraries of the organism that makes the protein; the challenge is to isolate the gene encoding for the protein. I do not know anything about the protein except that the product I've been given is pure.

Of course, the basic idea is simple: isolate the gene by screening the libraries with a radioactive nucleic acid probe and using a hybridization assay. But alas, I do not have a complementary probe: all I have is the protein! What do I do?

Here is one solution I can imagine. (I don't know whether it was possible around 1975, much less whether it actually happened.) I could try to isolate the mRNA that codes for my protein (but how to do that?), then radiolabel it and use the radiolabelled mRNA as a probe in a hybridization assay on the cDNA library (not the genomic library, since the unprocessed DNA could have noncoding regions).

  1. Would this have worked?
  2. Did anyone actually do this?
  3. If yes to either (1) or (2), how could one figure out which mRNA coded for the protein?

The question, in essence, is how to get to DNA, given only a protein, circa 1975.

  • $\begingroup$ I am curious as to why you posted here. If you really are at Caltech and just wanted to know the answer you should be smart enough to do a literature search and find the answer. However, again if you are at Caltech you presumably are bright enough to want to work it out by yourself, so you don't want those of us (me) who were around in 1975 gave you the answer (which I won't). As you, yourself admit, your proposal is flawed ("how to isolate the mRNA?"). Do you want a hint or discussion? That's not what SE Biology is for. Tell me why I shouldn't vote to close. $\endgroup$ – David Apr 25 '17 at 10:56
  • 1
    $\begingroup$ @David: I'm not at Cal Tech. I'm a math-trained grad student in the history and philosophy of science. I said explicitly it's a thought experiment. I don't know how to find the result in the literature. If you want a different way of asking the question: tell me how Maniatis et al 76 and 78 got mRNA rabbit insulin probes. I'm not trying to start a discussion: I want the answer. I find it cynical and unnecessary for you to assume I didn't want the answer. $\endgroup$ – symplectomorphic Apr 25 '17 at 13:48
  • $\begingroup$ Another possible solution: get the protein's amino acid sequence (using Sanger's technique), then synthesize a lot of RNA or DNA probes compatible with it (there will be a huge number, thanks to degeneracy of the genetic code). But was synthesis possible then? I don't know enough of the history to know which of these methods could've been done in the late 70s. $\endgroup$ – symplectomorphic Apr 25 '17 at 13:57
  • $\begingroup$ I wasn't being synical, just trying to find the context to your question (which nobody else had responded to). Your latest suggestion is a more logical use of the pure protein idea. You might think what else you can do with a pure protein, and also — as you mentioned cells — why proteins like insulin and globin were "low-hanging fruit". I'm a bit busy at the moment, but will respond in due course if nobody else does. $\endgroup$ – David Apr 25 '17 at 14:39
  • $\begingroup$ Thanks. I assume no one answered because others don't know the answer: it's a non-trivial problem, not Googleable or capable of being looked up easily. I know insulin and globin were overproduced by insulinomas and red blood cells, so their mRNAs were abundant. But how to isolate the mRNA responsible for insulin, say, in particular? $\endgroup$ – symplectomorphic Apr 25 '17 at 14:51

The following is based on a chapter on Cloning DNA in the 10th edition of The Biochemistry of the Nucleic Acids (RLP Adams et al.) published by Chapman & Hall in 1986. (This is probably more pertinent than the 11th and final edition, which was published in 1992 and had a more contemporary treatment.) Apologies for the fact that it will be difficult to find. It should be available from some university libraries and through inter-library loan systems in certain countries. Section A.7.3 gives a taste of the methodology used at that time.

Protein properties that could be used as the basis of a cloning strategy

In attempting to clone DNA for a particular protein for which there were no pre-existing clones to use as hybridization probes, one had to consider the properties of the protein one is able to exploit. These might include:

  • Amino acid sequence
  • Antibodies to the protein
  • Biological activity
  • Particular physical properties, such as size

(Clearly if one had no such ‘handle’ to the protein one was not in a position to try to clone its cDNA or gene.)

Three general approaches

There seemed to be three general approaches:

  • No expression
  • Indirect expression
  • Direct expression

The approach that one might take was also governed by whether the protein is particularly abundant in a specific tissue, when one might expect less sophisticated more labour-intensive approaches to stand a chance of success.

Another point to emphasize is that, at least with eukaryotes, one would look for a cDNA clone first, and only try for a genomic clone after one had this to employ as a hybridization probe.

1. ‘No expression’ approaches

In the no expression approaches one relied on the amino acid sequence to identify the cloned DNA. One such approach was to use hybridization probes against stretches of amino acids, the codons of which had limited redundancy (Met and Trp have a single codon, and some other amino acids only two) using low stringency hybridization conditions. This required luck. Alternatively one could sequence random clones and hope to find one that correlated with the amino acid sequence. At that time DNA sequencing was slow and labour intensive, so that this was only applicable to proteins that were relatively abundant in the tissue from which the libarary was prepared.

2. Direct expression approaches

Direct expression generally entailed using a bacterial expression vector which produced a fusion of a bacterial protein with a part of the eukaryotic protein. (Statistically only one in six of clones would have the correct reading frame in a fusion protein.). This approach was favoured if one had an antibody, which did not require a complete protein.

3. Indirect expression approaches

By indirect expression was meant using immobilized cloned cDNA to select corresponding mRNAs (‘hybrid selection’) which could be expressed in a cell-free system. This was appropriate if the distinguishing characteristic required a full-length protein, as in the case of biological activity or size.


Specific to the insulin gene, the following paper is, as far as I can tell, the first to isolate it:

Ullrich A, Shine J, Chirgwin J, Pictet R, Tischer E, Rutter WJ, Goodman HM. 1977. Rat insulin genes: construction of plasmids containing the coding sequences. Science 196:1313-1319.

The took advantage of the fact that insulin is produced by β cells in the pancreas and so are enriched in insulin mRNA. After poly(A) RNA was isolated and reverse transcribed, the major band observed upon electrophoresis of the resulting cDNA was expected to belong to the insulin gene:

It seemed likely that the [major] fragments were derived from insulin cDNA because of their prominence in the total cDNA preparation. Therefore, these fragments as well as the complete cDNA preparations were used in the cloning experiments.

Thus they cloned this cDNA, sequenced it and discovered that it contained the entire proinsulin coding sequence:

Since the mRNA contains a terminal poly(A) sequence, the [cloned] DNA strand containing 3' terminal poly(dA) is of the same sense. This strand determines an amino acid sequence which exactly corresponds to the entire coding region for rat proinsulin I and 13 out of 23 amino acids of the prepeptide sequence.

With this sequence determined and because of its high degree of conservation, the homologous human gene could be isolated by hybridizing with rat insulin cDNA:

Bell GI, Swain WF, Pictet R, Cordell B, Goodman HM, Rutter WJ. 1979. Nucleotide sequence of a cDNA clone encoding human preproinsulin. Nature 282:525-527.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.