Sufficient for pub?

Robert Horton horton at cis.umn.edu
Wed Apr 19 11:40:54 EST 1995


GORDON BETTS (betts at ORION.ETSU.EDU) wrote:
: Being new in molecular biology work, I often ask rather simple questions,
: but I hope others will gain from my ignorance.

: I am asking whether I have sufficient evidence for the existance of a gene
: in a species and if this evidence is sufficient for publication.  
: 1). we stimulated cells to produce mRNA.
: 2). We performed RT reverse transcriptase.
: 3). We performed PCR using a 20 nucleotide probe of a sequence same as mouse.
: 4). We ran on a gel and found-
:   low amounts of signal for 0 hour stimulation,
:   lots at 4 hour,
:   moderate amounts at 8 hour,
:   this closely approximates what is in the literature for mouse.

: Is this significant evidence? Is this publishable? It couldn't be an
: artifact could it?  Do we need to back up and do Northern or Southern
: blotting using biotinylated probe?

You'll probably need to confirm that the amplified product is really the
gene of interest. You could do this by hybridization or, better yet, by
sequenceing the puppy. Sequencing would give you something else to put in
your paper. That would probably satisfy me, but some people might want to
see (in order of increasing anal-retentiveness): results with low numbers
of cycles; "quantitative" PCR using a known amount of added control
sequence; Northern blots. 

The most important thing to control for is contamination. Be sure to 
include appropriate no-template controls.  In the case of RT PCR, try
using just RNA (without reverse transcribing it) as a control. If you
get signal without reverse transcriptase, you've got a problem (papers
about RT activity of Taq notwithstanding).

An internal control with some "housekeeping" or even exogenously added 
sequence is probably a good idea. Then you can call the whole mess 
"semi-quantitative". At least it controls for things like bad, 
PCR-inhibiting juju in your reactions.

The other thing people worry about is "illegitimate transcription". This
argument is based on the idea that many tissues express genes that are
irrelevant to that tissue, but at levels too low to matter. How you decide
what level is too low to matter is a philosophical question. The usual
not-so-philosophical answer is that if you can see the message on a 
Northern blot (or RNase protection), then it matters. Whatever.

I think that if you can show authentic, properly spliced message by PCR, 
then you know the message is there. The question of "illegitimacy" is 
more pressing if you had to do multiple rounds with nested primers to see 
anything.  Still, you can only address functionality with a functional 
assay anyway. Northern, Schmorthern.

If you have independant data (function, antibody binding, or whatever),
that will help a lot. 

Good luck.

Bob Horton (Ph.D.!)   /\ "Crash programs fail because of the theory that
U. of Minnesota, CBS  || with nine women pregnant you get a baby a month" 
1479 Gortner Ave.    /||\   -Werner von Braun.  Disclaimer:"Bob who?"
St. Paul, MN 55108    ^^   horton at biosci.cbs.umn.edu/(612) 624-3790



More information about the Methods mailing list