In addition to asking an interesting question, using new or novel techniques, having GREAT data, and following the journal's format and Instructions for Authors to a tee, there is one more procedure that some people use to get a manuscript reviewed in a high-impact journal. If the paper is really outstanding, some authors will contact well-known scientists (Nobelists, National Academy members, etc.) in their field and ask them to send a personal letter to the journal editor in which they strongly endorse the work (of course it helps if you know these people personally). And by receiving these letters ahead of time, the editor is then "pre-adapted" and sometimes this may result in the manuscript being sent out for review. Of course, there is never any guarantee that it will be accepted.

PS- I have never tried this myself, but some of my colleagues have.

Muneeb Faiq, I suggest you perform an exact replication of an already published Nature Neuroscience article. The paper you would be replicating should have all the raw data as well clearly described methods available so that one can compare your and the original results without readers having to resort to uninformative comparisons of p-values and post hoc effect sizes. Obviously, your data should likewise be available.

You should then submit that paper to the journal, regardless of whether you feel your experimental results indicate a positive replication or falsification. If Nat. Neur. declines your submission, then it is rather scandalous, in my opinion, and you should submit somewhere else, clearly stating that your replication is of an original paper in that high profile journal. It would be the most valuable contribution you could make as a scientist and would probably get you a good amount of press and reputation.