Well the latest paper from the Reed lab (squeeking into Cell on its last issue of 2006) demonstrates that the cap is indeed promoting nuclear export of mRNA in vertebrate cells. (For more on mRNA export, click here.)
This idea that each step of mRNA metabolism is "coupled" to the following step of mRNA metabolism is an idea that has been going around quite a bit in the RNA world.
The cap, is a structure found at the start (or 5' end) of mRNAs, and consists of an inverted methyl-Guanosine linked by a tri-phosphate to the first nucleic acid, invariably guanosine. The role of the cap is to prevent degradation of the mRNA by enzymes, RNAses, that chew up the 5' end of RNAs and to promote splicing. Initially it was reported that the cap did promote mRNA export, but then subsequent publications found that the cap is dispensable for export of intronless mRNAs and that mutations of cap binding proteins in yeast do not affect mRNA export.
It turns out that that the cap, through the cap binding proteins CBP20 and CBP80 collaborate with the splicing machinery at the very first splice site to recruit the TREX complex (for translation export) to the transcript. This deposits the export machinery at the 5' end of the mRNA and suggests that nuclear transcripts are transported "head first" through the nuclear pore. Indeed EM micrographs from balbiani ring cells have suggested this head first mode of export for mRNAs and thus allowing for translation initiation of the mRNA before the its export from the nucleus is complete.
Now having said that, there may exists other mRNA export machines that act in conjunction with this cap/splicing regulated machine to facilitate mRNA transport. Indeed other groups have seen that certain mRNAs do not require splicing for efficient export and as I mentioned above these transcripts do not require a cap for their export. How many of these alternate export machines exist is unknown. What proportion of mRNAs use the main mRNA export machine categorized so far is also unknown. What we do know is that the whole process is likely to be much more complicated than we currently think. Remember that the separation of mRN A processing and mRNA translation has been proposed to the main reason that the eukaryotic nucleus developed and thus the mRNA export machines are likely to be quite elaborate and diverse ... after all eukaryotic cells have had plenty of time to evolve these mechanisms and play around with their function.
There's been reports about the translation factor eIF4E being able to transport a subst of mRNAs and is cap dependent, however, i don't believe this has been incorporated into the TREX model.
Yup, I wrote about that paper a couple of months back. I bet that quite a few other mRNA export pathways are yet to be described.