The best approach depends a great deal on the size of your proteins and whether stoichiometry of expression matters. T2A or P2A sequences work really well for maintaining comparable level of expression of your two sequences, but may not be optimal if the cDNA size gets too large (3kb or larger).
Other approaches are to use a directional promoter (e.g. PGK or EF1alpha on the sense strand coupled to minimal CMV on the antisense strand), or completely independent promoter-gene cassettes. The latter approach is best exemplified by the MultiLabel system from Philipp Berger's lab: http://www.ncbi.nlm.nih.gov/pubmed/21081918
The best approach depends a great deal on the size of your proteins and whether stoichiometry of expression matters. T2A or P2A sequences work really well for maintaining comparable level of expression of your two sequences, but may not be optimal if the cDNA size gets too large (3kb or larger).
Other approaches are to use a directional promoter (e.g. PGK or EF1alpha on the sense strand coupled to minimal CMV on the antisense strand), or completely independent promoter-gene cassettes. The latter approach is best exemplified by the MultiLabel system from Philipp Berger's lab: http://www.ncbi.nlm.nih.gov/pubmed/21081918
For E. coli expression, the pETDuet plasmid family is suited to express up to 8 proteins (you can combine different vectors). Each vector can encode for 2 proteins carries the T7/lac promotor.
since you didn't write which organism you want to use I will only recomend a system for E.coli.
I have been working with the pQLink system. It worked very well concerning 1:1 stoichiometry of and high expression levels of co-expressed proteins. For more information see: http://www.ncbi.nlm.nih.gov/pubmed/?term=17311810
The plasmids provided are freely combinable by LIC (ligation independent cloning) which is very cheap and highly reliable.
With this system you can express ether GST-tagged, His6-tagged or untagged proteins.