Correlation between mRNA and protein

Thumb default avatar
Mar 01, 2017


We have been studying breast cancer cell lines and trying to determine how well the PIP (Prolactin Induced Protein) gene and its translated protein detects cancerous cells. mRNA levels for the PIP gene seem to be down regulated significantly (100 fold) within cancerous breast cells. However, the subsequent protein levels seem to not show the same trend. Protein levels of PIP seem to be similarly expressed within both cancerous and non-cancerous breast tissue. What would cause this dissociation between mRNA levels and its translated protein?

It appears that PIP is needed for these cancerous breast cells due to these consistent protein levels. Therefore, why would the mRNA levels be down regulated?

One hypothesis we have is that there may be a change in post transcriptional modification within cancerous cells. Alterations to mRNA made post transcriptionally might make the PIP gene not show up through use of PIP primers.


Thumb default avatar
Roberto Rosati 10 months ago

I'm not sure I have a biological explanation for discrepancies between mRNA and protein levels, that can hold true for a 100-fold decrease in mRNA that doesn't affect protein levels at all. (But I hope others will...)
How are you testing protein levels, by western blotting? Elisa? How much do you trust your methods, and your own technique if I might ask?

Thumb default avatar
morganac94 10 months ago

We used both western blotting and immunohistochemistry; however not ELISA. Both cancerous and normal cells and tissue show up as having similar protein levels across multiple cell lines.

Thumb default avatar
Roberto Rosati 10 months ago

A few ideas are:
1) what you are detecting from Western Blot and Immuno might actually be protein fusions that include the part of PIP your antibody was raised against; while your PCR is amplifying a part that is lost in the gene fusions.
2) the tumors are expressing some alternatively spliced transcript.

You're posting in bioinformatics though - does this mean you've done a transcriptome analysis to determine the mRNA levels? If so, did you already discard these hypotheses?

Other possibilities are that either the mRNA or the protein are being stabilized so that either the mRNA can be translated for longer, or the protein has a longer turnaround. Of the two, I'd be inclined to think about the protein having a longer turnaround, because a feedback loop might then be responsible for the decrease of mRNA levels. But hundredfold? That's a lot.

Good luck! I'm sure others will post their ideas too.

Edit: I'm re-reading your post and you write that normal tissue also expresses as much PIP as tumor tissue. Do you mean normal breast tissue, from nonlactating women? Having the same amount of protein and 100 times more mRNA for PIP than cancerous cells? Did you expect this finding?

Thumb default avatar
mchlbrmn 10 months ago

Is there a possibility that the antibody cross reacts with another protein? Perhaps confirm with another antibody of specific for a different epitope?
How was mRNA level determined? If specific to a certain local sequence, perhaps different primers /probes, possibly specific for any alternate splices, might find different results.

What about the article below. Maybe it's old news, and has been cleared up? Are these alternate names for the same gene, different splices, or post translaitonal modifications (which you've discussed as a possibility above)? :

Differential antibody reactivity and CD4 binding of the mammary tumor marker protein GCDFP-15 from breast cyst and its counterparts from exocrine epithelia.

Caputo E, Autiero M, Mani JC, Basmociogullari S, Piatier-Tonneau D, Guardiola J
International Institute of Genetics and Biophysics, CNR, Naples, Italy.

Analysis of biopsies from breast cancer patients demonstrated that GCDFP-15 (gross cystic disease fluid protein-15) is a specific immunocytochemical marker of primary and secondary apocrine breast tumors. The protein has an amino acid sequence identical to SABP (secretory actin-binding protein), to PIP (prolactin-inducible protein) and to gp17, a protein isolated from human seminal plasma. The latter was found to bind to CD4, a T-cell [?] co-receptor involved in antigen recognition, thereby inhibiting the ability of the receptor to interact with the HIV-1 envelope protein gp120. We compare here the ability of independently purified GCDFP-15, SABP and gp17 and of recombinant PIP both to cross-react with a panel of monoclonal antibodies (MAbs) raised against GCDFP-15 or gp17, respectively, and to bind to CD4. We show that, although the various factors share the ability to bind to the panel of antibodies used, differences in the pattern of MAb recognition can be demonstrated. By comparing the kinetic constants for binding of GCDFP-5 and gp17 to CD4 by biosensor technology, significant differences in binding affinities were observed between the 2 factors, thus reflecting structural differences. Surface plasmon resonance analysis also showed that anti-GCDFP-15 and anti-gp17 antibodies inhibit the binding of CD4 to GCDFP-15 and gp17, respectively, to different extents. Our data thus indicate that, while the various forms of the protein are encoded by the same cDNA, tissue specificities due to post-translational modifications exist. This information may be relevant for developing more sensitive and accurate tests for the use of GCDFP-15 as a diagnostic mammary tumor marker and, most importantly, raises the possibility that GCDFP-15 may constitute a breast tumor-specific antigen.

Int. J. Cancer (1998)