Thousands of rare diseases are caused by a Mendelian genetic error. So far, more than 1800 genes associated with rare diseases have been identified1 and, in many cases, their expression patterns and functions have been unravelled. This information is a prerequisite for development of a therapeutic strategy. Depending on the disease’s severity, and by assessing feasibility and treatment alternatives, gene therapy can be viewed as an option in some instances (table2). Genetic disorders are phenotypically very heterogeneous. In considering the appropriate form of gene therapy, three basic parameters need to be established: whether a mutation leads to a loss or gain of function; whether or not a gene product’s function affects cell survival or development; and the disease gene’s tissue specificity. 3 There are four different gene therapy strategies. First, addition of a normal copy of the mutated gene. This approach is best suited to loss-of-function mutations and has been the focus of most gene therapy attempts so far. Second, modification of messenger RNA to avoid the consequences of mutation. This strategy can be viewed as a promising option when the mutated exon is not indispensable. 4 Third, inhibition of expression of a mutated gene. This approach is potentially useful to prevent the expression of a gain-of-function protein or to inhibit a cryptic splice site, thus preventing expression of an abnormally spliced product with deleterious consequences. The use of small interfering RNA seems to be preferred, 5 provided that unexpected toxic effects (as seen in the liver6) are not serious. Last, repair of the gene—an ultimate and elegant strategy aimed at reverting mutation. 7 This technology is based on the use of chimeric proteins composed of a DNA-sequence-specific binding domain and an endonuclease capable of inducing site-specific double-strand breaks in DNA. Simul taneously, a template encompassing the wild-type sequence that corresponds to the mutated stretch of DNA is introduced into the cell and acts as a substrate for repair by homologous recombination. Zinc-finger protein domains coupled to the FOK1 nuclease have been engineered and shown specifically to correct the IL2RG gene encoding the γc-chain of cytokine receptors (as noted in X-linked severe combined immunodeficiency [SCID-X1]) in 5–17% of cells. 7 However, the road to clinical application is still long, because of many technical concerns.

RNA viruses (eg, γ retroviruses, spumaviruses, and lentiviruses) are being used to mediate integrative gene transfer. 8–10 Either a viral promoter, such as the retroviral long-terminal repeat, or a potent viral or cellular internal promoter can be used. γ retroviral vectors are only effective in dividing cells, because the preintegration complex cannot cross the nuclear membrane, whereas lentiviral vectors can deliver genes to the genome of non-dividing cells. Substantial attention has been paid to the integration sites for provirus-derived vectors. Retroviral vectors integrate with a high frequency near CpG islands, 11–13 meaning that the long-terminal repeat can also target active genes albeit without a selective tropism to the transcription start site. Lentivirus-like vectors derived from HIV or simian immunodeficiency virus can target active genes. 14, 15 Intact retroviral