Reading these answers I wonder, why doesn't "gene therapy" use self-contained plasmids instead of trying to splice a length into a chromosome?
First, plasmids do not usually integrate into the chromosomes of human cells. In the laboratory, plasmids can be introduced into the cytosol of cells, and plasmid DNA will then be "read" by the host cell and used to make new proteins, for example. But plasmids usually remain separate from the cell's own DNA, and in that sense they are not effective for gene therapy.
Second, our cells are quite good at detecting and destroying plasmids, as well as other foreign DNA. This is a self-defense mechanism, evolved to get rid of viruses and other nasty parasites that might be trying to hijack the cell. In the lab, where we are working with isolated cells in a culture dish, with no immune system, plasmids are typically tolerated. But in a human being, immune cells that have specialized in detecting intruders are very competent at detecting such manipulations, and they will quickly sound the alarm and wreak havoc ... So if injected into the body, plasmids are quickly detected and destroyed.
Research in gene therapy aims to find more efficient ways of "editing" the DNA of human cells, ultimately in a way that can be safely applied in the body. Such a technique does not exist yet. The latest CRISPR-Cas9 gene editing system is a phenomenal tool, and is very efficient at getting into the DNA of human cells; but the CRISPR-Cas9 proteins must themselves be delivered into the cell, for which a virus is typically needed, and this would again cause immune reactions if attempted in a human body.
I believe we could use plasmid insertion to assist with gene therapy. EBV can maintain a plasmid in the cells it infects and the plasmids can effectively replicate. With a greater understanding of the latency and the implications of inserting genes using EBV, it may be possible for gene therapy to occur via genes on a plasmid. references: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2562867/ http://jvi.asm.org/content/65/1/483.full.pdf