Elife & PloS both charge fairly high submission fees. They employ their own teams of editors. PloS for example has close to 500 employees. Fees for most of their research areas are > $2000 per article submission. Their latest addition of Computational Biology has 80 articles, or $160,000 in revenue. I believe PloS offers fee assistance in certain circumstances, so the total amount will probably be a little less. Either way, this is more than enough to cover the sorts of activities I mentioned.

As an example from Elife, in addition to the peer review feedback process, Elife describes their editorial output as providing: "detailed feedback on the manuscript for the authors, including requests for revisions and suggestions for improvement." detailed feedback on the manuscript for the authors, including requests for revisions and suggestions for improvement"

In terms of the mechanics of publishing, it's all very little different than a traditional publisher, except for the funding model. In terms of public access, it is far superior: Publication costs are covered without exorbitant subscription fees and the public gets the benefits of the research for no cost of their own.

This is why I would like to see a move to grant models where the grant includes the cost of publication, and a requirement for the portion to be used to get knowledge into the public even if the results did not confirm the experimental hypothesis.