Immunohistochemistry - Leakage of protein from nucleus?

D.K. dk at
Thu Oct 4 23:38:47 EST 2001

gliftor at (Melissa Greeve) wrote:
>Hi everyone again,
>Thanks for your replies to my last posting of immunoprecipitation
>woes. I seem to have managed to get IPs to work, resorting to the use
>of a more dense SDS-PAGE gel to avoid the problem of the protein G
>band - I have, however, yet to get consistent results. But it's only a
>matter of time, more sleep deprivation and some more tweaking... :)
>Anyway, onto my next curious problem. One of the important experiments
>that will (hopefully) nicely supplement the IP results will be
>immunohistochemistry of cells grown on slides. 

Oh, boy, yet another notoriously artifact-prone and misleading method. 
Believe it or not, it is possible to get ANY result with 

>The protocol itself
>appears to be fine, but one of the proteins I have been staining for
>is meant to be (published for other cell types) nuclear - but I have
>only been able to show that it is cytoplasmic and perinuclear. So far,
>I have tried two different antibodies to the protein. In contrast, I
>have shown that another protein, also published as nuclear, is indeed
>expressed in the nucleus and cytoplasm in a cell-cycle dependent
>Someone has suggested to me that one of the potential reasons for this
>discrepancy could be leakage of the protein from the nucleus into the
>surrounding cytoplasm during insufficient fixing of the slide. So, I
>tried chilled acetone, with nasty results - the cells, understandably,
>Is this a logical explanation? 


>Why would one protein act like this
>when fixed in 95% ethanol and not the other (both have similar
>cellular functions)? 

On itself, there is nothing illogical here. Different proteins have
different properties. Some need more fixing than others. What 
appears to be extremely improbable is the suggestion that a nuclear
protein is not being fixed in nucleus, freely diffuses into cytoplasm and
is being fixed there. 

If your results are solid (e.g: positive controls work, negative controls 
work, results are qualitatively independent on antibody used dilution 
and fixation/permeabilization method, the signal is inhibited by an 
excess of purified antigene and the staining can be reproduced 
by someone else in the lab), then believe in what you see, not in 
what someone else has published. They may have very well screwed 

>Can a PhD get any more complicated??

Welcome to the contemporary cell biology. :-) Combine hasty performed 
immunocytochemistry, immunoprecipitation, yeast double hybrid and 
GST fusion pull-outs in any combination with a "novel" protein of unknown or
speculative function and the confusion will never cease. 

>I will try a different cell type as a control, to confirm that the
>apparent localisation of the protein in cytoplasm results from
>insufficient fixing, otherwise, maybe my cell type is a strange
>exception! It's important to demonstrate the true localisation of this
>protein before I obtain any results (of course, analysis of this
>protein is crucial to my project!)

It sounds like it is totally crucial for you to get that "other" cell line
and see if you can reproduce nuclear localization on it. You can claim 
"strange exception" only of you can observe it in an experiment
performed in parallel with "your" cell line, back to back. 

More information about the Methods mailing list