mRNA in general have 5'UTR (untranslated region) followed by ORF (open reading frame) and then comes 3'UTR. Both UTRs regulate mRNAs stability, localization, translational efficiency. In eukaryotes cytozolic mRNAs have 3'UTR polyA, which regulates mRNA stability and translational efficiency, and 5'UTR which regulates efficiency of translation.
Mitochondrial mRNAs... it's a mess.
1) 3'UTRs are regulating localization (i.e. transport into the mitochondria, (Pattini 2003 PIMD 15376913)) and in mRNA stability (Gagliardi 2004 PIMD 15145579). Mechanisms of mRNA stabilization seem to be very, very different in plants (polyA destabilizes, like it does in bacteria!), yeast (no polyA at all, just like it is in yeast cytozolic mRNAs!) and mammals (polyA stabilizes, just like it does in eukaryotic cytozolic mRNAs!). 3 different approaches?! It is about as insane as it gets, really.
In mammals 3'UTRs are short (Ojala 1981 PIMD 7219536), unlike in plants and yeast, where 3' UTRs are long.
2) 5'UTRs are regulating translation (mRNA-specific enhancers of translation bind to them and thus regulate translation initiation) and localization within the mitochondria, thus coupling translation and insertion of the mRNA into the membrane. Most of the proteins that are translated in yeast mitochondria are intermembrane proteins involved in respiration and ATP production (review Towpik 2005 PIMD 16341268), thus coupling translation and insertion is a must. And indeed, when multisubunit complexes are assembled, translation of the individual subunits is geometrically coordinated (Naithani 2003 PIMD 12529447).
All this is well and good, but there are issues. Somehow investigation of 3' and 5'UTRs is a big thing in plant mitochondria, and not much is done nowadays with yeast or mammals. Or at least it is not easy to find. Second, no one tried systematically comparing UTRs from different organisms.
What I have dug out by now is this: in plants 5'UTRs are long, and there is a lot of experimental material here using 5'-RACE (Froner 2007 PIMD 17488843, Kuhn 2005 PIMD 15653634). In yeast - long 5'UTRs as well (review Costanzo 1990 PIMD 2088182), though much less studied experimentally. In mammals 5'UTRs are claimed to be short, at least in humans (Montoya 1981 PIMD 7219535). Here signals for mRNA-specific initiation enhancers are located within the ORF.
Is plant and yeast mitochondria translation radically different? Why did mammalian mitochondrial mRNA loose 5'UTR regulation? If yes, where is the watershed? What are the differences in the yeast+plants mitochondirial machinery vs mammalian?
Mammalian mitochondrial genome is super-streamlined, cutting corners where possible (review Attardi 1985 PIMD 3891661), so that could be a reason for the loss of 3' and 5'UTR. But what drives this minimization? Why trying THAT hard?
Experiments on mitochondrial translation seem to be done on different systems in different areas of research: yeast are used for identifying initiation enhancers and studying genetics and molecular biology of translation regulation, in plants 3' and 5' UTRs are extensively mapped, and in mammals using very, very simplified translational system (tRNA, IF2, IF3, EF-Tu and EF-G) some rudimentary biochemistry is done. This does scitsofrenic devision of labor is in keeping with the mitochondrial spirit indeed.
PS: mitochondrial ribosomes are also very, very strange.
Mitochondrial evolution: Karlberg 2003 PIMD 12728281
Mitochondrial and chloroplast translation: Gillham 1994 PIMD 7893142
Mitochondrial translation and desease: Perez-Martinez 2008 18991722
Plant mitochondrial translation: Binder 2003 PIMD 12594926, Hoffmann 2001 PIMD 11642360
Yeast mitochondrial translation: Costanzo 1990 PIMD 2088182, Dieckmann 1994 PIMD 8206703
Mammalian mitochondrial translation: Spremulli 2004 PIMD 15196894