Wednesday, April 01, 2009

mRNA structures translationally prefered by cancer


Translation initiation is often over-activated in cancer cells, however, this over-activation does not favour mRNAs equally. In fact it seems translation over-activation favours mRNAs that promote tumourgenesis and tumour progression. Santhanam et al. in PLoS ONE recently took a look at what general mRNA structural characteristics determine translational activity in cancer cells with over-activated translation initiation. Previously lots of focus has been on the influence of secondary structure and length of the 5'UTR as this has been demonstrated to influence 'translatability'. With new information about microRNAs binding to the 3'UTR to influence translation, an unbiased approach to evaluate their relative influence seems like a very good idea. The methods in this work essentially involved comparing microarray analysis of cell lines with over-activated translation initiation and those without. In each cell line the translational activity of each message was determined by quantification of the percentage of the message that was being actively translated. Then they looked for commonalities in the sequence of the mRNAs that are specifically translationally activated by translational over-activation. (probably using computers) :) Massive data set and unbiased approach = pretty convincing coorelations (we're talking p values of 10E-6)

Here is what cancer cells look for in a potentially translatable mRNA:
1. GC rich 3'UTR
2. secondary structure involving sequences JUST before the start codon and JUST after the stop codon. These results are more striking for the 3'UTR.
3. Short 3'UTRs
4. No microRNA target sites. Indeed the presence of predicted microRNA target sites was negatively coorelated with translational over-activation. (with some exceptions)

The take home message is, somewhat surprising, that the 5'UTR is less correlated with effects on translation over-activation relevant to cancer than the 3'UTR.


5 comments:

Bayman said...

I like the idea.

I have to say that I doubt this study is picking of mRNAs that are preferred by cancer cells. Rather, it is picking up signatures of difficult to translate mRNAs (ie secondary structure, GC content) which are relatively much more abundant in polysomes when the translational machinery becomes hyperactivated and therefore less picky about substrates.

Also the fact that cancer cells show a "preference" for miRT-free mRNAs is bizarre. The only explanation I can think of is that miRNAs are generally overexpressed in cancer.

rob said...

I have to say that I doubt this study is picking of mRNAs that are preferred by cancer cells. Rather, it is picking up signatures of difficult to translate mRNAs (ie secondary structure, GC content) which are relatively much more abundant in polysomes when the translational machinery becomes hyperactivated and therefore less picky about substrates.
I think there is evidence that indeed cancer cells are just less picky about substrates and therefore preferentially translate these messages which are enriched in messages for tumourgenesis and tumour progression.
Also the fact that cancer cells show a "preference" for miRT-free mRNAs is bizarre. The only explanation I can think of is that miRNAs are generally overexpressed in cancer.
Or.. this form of translational dysregulation is not able to overcome the microRNA mediated translational repression.

Bayman said...

Or.. this form of translational dysregulation is not able to overcome the microRNA mediated translational repression.

Yeah I think that makes sense and that seems to be more the argument presented in the paper. (now that that I read it). But the thing is that the "form of translational dysregulation" they are talking about is artificial overexpression of eIF4E. I find it confusing to think about what this means in real life.

I think what they're trying to say, as you alluded to, is that 4E dependent translation is uniquely sensitive to miRNA regulation. But compared to what? In these experiments, compared to un-molested cells. But isn't almost all translation 4E (cap) dependent, even in unmolested cells? This is where I get confused.

Maybe when 4E is overexpressed, some other aspect of translation becomes rate limiting, other than cap-recognition. For example, elongation. So they're picking up signatures of mRNAs that are well translated when elongation, or whatever step X, is limiting, when cap-binding proteins are in excess.

rob said...

I don't know all the evidence but I think that eIF4E is supposed to be a reasonable model for the different types of translational dysregulation that occurs in cancer. AKT signalling, ect.
Also I interpret their results as saying that microRNA transcripts are specifically less translationally upregulated in the eIF4E overexpressors. Not that the overexpressors have less expression of microRNA regulated transcripts, just less upregulated.
I agree with your last paragraph, I think. If translation initiation is no longer rate limiting it is likely something else is. I don't know if that effects any of their interpretation but it's interesting. Elongation might actually make sense if you look at the repeating pattern in the ORF in the GC content data. But apparently that was already known.

Bayman said...

microRNA transcripts are specifically less translationally upregulated in the eIF4E overexpressors.

That makes sense.

Question is, do tumor cells have other ways of over-stimulating translation that are miRNA insensitive? I suspect so.

But at the least this data definitely supports the idea that miRNA-mediated translational repression is via a 4E/cap-dependent related mechanism. That's interesting.

Are picornaviral IRES-driven transcripts insensitive to miRNA-mediated translational repression?