Unexpressable protein

Peter Ellis via methods%40net.bio.net (by pjie2 from cam.ac.uk)
Sat Sep 25 18:12:42 EST 2010

On 25/09/2010 23:25, DK wrote:
 > Thanks for the details. All seem reasonable. One nitpick, perhaps
 > will be helpful in the future: don't bother with anti-peptide Ab unless
 > functional inhibition studies are planned. It's alwasy easier to make
 > small fragment in E.coli, immunize with denatured protein and get
 > super pure polyclonals for Western by purifying against blotted
 > protein.

A consideration I hadn't explained is that I'm distinguishing between an 
X gene and an autosomal relative, so a peptide antibody was necessary to 
direct it to regions of greatest divergence between the two.  As for the 
ease or otherwise of expressing proteins in E. coli, that's kind of the 
problem here!

 > OK, things I can think of off hand:
 > - The gene produces super-unstable mRNA.
 > - The gene, somewhere near the start, has termination signal
 > (either transcription or translation; the former occasionally
 > happens for termination).
 > So my first impulse would be to see if the message is even made.

As I said in the previous post, for the most recent construct (in 
pT7CFE1 vector) we did indeed isolate/sequence the mRNA and confirm it's 
expressed correctly in one of our in vitro systems (T7 transcription 
coupled with rabbit reticulocyte lysate). For this vector, the RBS, 
transcription initiation site and first half dozen amino acids are 
supplied by the vector, so we can also rule out a problem with 
transcriptional initiation.

 > If with a totally deifferent gene sequence you continue to observe all
 > the same, get all kinds of more preliminary data and start get
 > writing a grant. :-)

Current working hypothesis is that there's something near the start of 
the ORF that causes the problem. Could be a non-standard translation 
termination signal, could be mRNA structure, could be a run of 
non-preferred codons for which the tRNA is more abundant in spermatids, 
could be some problem in the protein domain structure.... or any number 
of other options I've not thought of yet!

Interestingly, the IHC signal alters its localisation as germ cells 
mature - it's cytoplasmic in round spermatids and much more localised in 
elongating spermatids.  In whole testis Western the 55 kDa band comes up 
in both the RIPA-soluble and insoluble fractions.  I suspect that the 
cytoplasmic signal is the soluble form and the localised signal is the 
insoluble form.  Westerns next week should help resolve this :-)

If I'm right about the soluble/insoluble forms, we can hypothesise that 
there is a specific chaperone protein in round spermatids that keeps the 
monomers in a soluble form until they're ready for use.  In the absence 
of that factor, maybe it starts clumping up and dropping out of solution 
in our in vitro / culture systems.  The E.coli expression experiments 
did give a load of insoluble crud - absolutely *huge* pellet after lysis 
with BugBuster - whatever it was didn't however have a His-tag on it 
(C-terminal) and didn't react with the peptide antibody (also C-terminal).

I think what I'm going to do is try cloning just the second half of the 
protein - only requires ordering 1 more forward primer and a few days' 
PCR/cloning work.  I'd really like to do a yeast 2-hybrid screen further 
down the line though, so expressing as much of the protein as possible 
would be very helpful.  On the other hand, if the interaction domain(s?) 
mean it can't be expressed in yeast, we're a bit stuffed on that front...

There's also the Hail Mary pass option of spiking testis extract into 
the in vitro translation reaction and hoping it contributes whatever 
factor is necessary. I guess I could homogenise testis tissue in a 
buffer of choice (RIPA buffer or PBS?  Which one won't inhibit the in 
vitro reactions?) and just stick some in.  Or I could do a collagenase 
digestion of the tissue, spin down the separated cells and actually make 
a testis lysate to use in place of the reticulocyte lysate. The latter 
especially seems like something that would take a lot of time and money 
to optimise though :-(

Any further suggestions?


More information about the Methods mailing list