If you need an antibiotic free system many have developed approaches to plasmid selection without antibiotics. For example complementing a mutation in the host strain (say a Met- mutant and you have the complementing Met gene on the plasmid), this can work for any "essential" gene in the E. coli host. You could even go smaller in size by having a suppressor tRNA on the plasmid that "corrects" a nonsense mutation in an essential gene of the host. My lab worked with alanine racemate deficient strains of E. coli and cloning an alanine racemase would complement the defect (and you can use rich medium and not be limited to minimal). So there are many approaches.

But as Sebastian Schmitt

Thank you Sebastian Schmitt

As I need an antibiotic free system, I also thought of constructing a vector with GFP but the problems are the possible toxicity of the protein and the plasmid maintenance. So, I think I should go with the complementation system as Michael suggested.

GFP actually wouldn't work all that well for you since that is only a screen and not a selection.

However if you want to play around with stability and such you have a built in system already. If your plasmid keeps the intact lacZ alpha region and polylinker (without an insert) then it should confer blue color and weak Lac+ growth in the appropriate cloning strain DH5a or similar. You could try to directly select for growth on lactose minimal (it will be slow since the complimented LacZ expression is only partial) but should still work. Obviously this won't be a useful system if you intend to clone things into the polylinker, but might let you do some simple experiments regarding plasmid loss etc.