Which signal sequence to use for periplasmic expression in E. coli?

It is difficult to predict which signal peptide will be the best. 

The cleavage efficiency will depend of the first residue of the mature protein: some residues with positive charges, in particular lysines into the 10 first positions, may be a real problem for the cleavage by the signal peptidase. Consider also that a too large production (driven by a strong promoter) may produce inclusion bodies even in the periplasmic space. Some E. coli strains are able to form the disulfide bridges in the cytoplasm (eg, Rosetta-Gami 2 host strains). 

There are a number of variables that are important to test:

1. Signal sequence: you may test the native signal sequence (where appropriate), pelB, OmpA, DsbA, TolB and MalE. All are E. coli derived except PelB which is from E. carotovora. You may find out that target protein dependent preference for different signal sequences.

2. Strain: different strains may have different effects. We have had a couple of proteins that have failed/largely failed in B834(DE3) but have worked nicely in Rosetta2(DE3).

3. Media: we routinely always compare an IPTG induction in a rich media against autoinduction media carried out at 20C and 25C respectively. To fine tune expression it can be helpful to add glucose to the former and vary the induction temperature for both. It should also be noted that rich media isn't always the best for high levels of expression and we have also seen better expression of some proteins in more minimal media.

Also be aware of that about 50-60% of human proteins never express solubly in E. coli, no matter how hard you try. Extracellular proteins are usually worse than intracellular ones, as they are almost all glycosylated and otherwise modified. Be prepared to try Pichia or refolding.

Representative signal peptide sequences used for the secretory production of recombinant proteins in Escherichia coli.

Souce: NovoPro    2018-02-01