In one word, don't.  E.coli is not efficient at that.  Your yield will go down by orders of magnitude compared with intracellular expression.  If you must secrete, you should change to an organism that does it efficiently, like yeast or mammalian cells.  (unless you absolutely don't care for yield). 

Type II secretion system will helps.  Cloning and expression of your target gene in expression vector (controlled by chemical induction - IPTG, Arabinose) can produce your protein inside the cells.  Basically signal sequence will help your protein to secrete out of the cell.  So, addition of various signal sequence before your target gene can help your protein to secrete out.    Below attached paper will helps you.  Good luck.

All gram negative microbes, including E.Coli secrete proteins and allied microbial signal molecules as packed nanoscale outer membrane vesicles. Here is an example http://www.labome.org/research/Membrane-vesicle-trafficking-in-prokaryotes-molecular-biomechanics-of-biogenesis-of-outer-membrane-v.html

You could try using commercially available pET22b+ with PELB signal sequence. This is  IPTG inducible expression vector. The article attached in previous message by sivachandiran is a good start. All the best.

There are generally 8 secretion pathways in gram negative bacterials, that you can try all of them. Type  II secretion pathway is the most basic one. First of all, a signal sequence such as pelB or ompA is needed which leads the protein secreting into the periplasm space. It depends wether it can secrete from the periplasm  to the environment. Bye the way, I have read a paper about YebF. Maybe you can fuse your protein to YebF. Good luck.

Article Extracellular accumulation of recombinant proteins fused to ...

why do you want to make it in two step (first production, second secretion)?