Q J Med 2003; 96: 873-874
© 2003 Association of Physicians
Biologic |
Misplaced
In a recent trial of gene therapy for an immunodeficiency state (X-linked severe combined immunodeficiency disorder), two of eleven children developed leukaemia. The haematopoetic stem cells of the patents, from the bone marrow, had been transduced with a vector based on an onco-retrovirus, the murine leukaemia virus (MLV). The vector expressed the common gamma chain of the interleukin receptor. In the two leukaemic children, the leukaemic clone contained vector DNA that had integrated in the immediate vicinity of the growth-promoting LMO2 gene. It is thought that this, in conjunction with the production of the common gamma chain, produced the uncontrolled growth of a lymphocyte precursor cell. This unhappy series of events has led to some general debate over the safety of retroviruses as vectors.
Retroviruses have been widely used as gene delivery vehicles. The reverse transcription of the viral RNA genome into a DNA copy that becomes integrated into the host cells genome is the critical step. These viruses are known to cause tumours when transcriptional elements in the retrovirus activate a growth-promoting gene (via the promoter or enhancer) and retroviral vectors often contain regulatory elements that are likely to activate genes. These transforming events, arising from what is called insertional mutagenesis, were believed to occur at random; if the insertion is random in position, the chances of causing trouble in therapy appear to be remote. This assumption must be questioned, however, since although current data suggest both that integration is favoured at transcriptionally active sites and that active regions are ‘disfavoured‘, the swing of information appears to be towards non-random insertions.
Recently, Schroder et al.1 reported that HIV integration favoured the sites of genes (they mapped 500 integrations of HIV-1 in the human genome). Wu et al.,2 in a comparative study looked at both HIV and MLV in HeLa cells. They found that 58% of the HIV integrations landed in genes, defined as insertion between the transcriptional start and transcriptional stop boundaries of one of the 18 214 RefSeq genes mapped to the human genome [www.ncbi.nlm.nih.gov/RefSeq/]. For MLV integrations, only 34% were so sited, but in MLV the integrations present were distributed evenly upstream and downstream of the transcriptional start site. They were very different from the HIV integrations, which did not involve upstream sites. So MLV has an apparent preference for the region surrounding the transcriptional start site, and HIV for the transcribed regions of genes. MLV-induced transcripts are therefore more likely to be full-length or nearly full length, but as the HIV vectors integrate further away from the transcribed region of the gene, they give rise to significantly truncated versions of the normal cellular transcript, and have a reduced chance of stimulating the cellular promoter. These data make clear that the interactions of the two types of virus with their host differ, and that the risks of using different viruses in gene integration therapies probably vary with the vector.
Viruses that infect terminally differentiated cells are less threatening in this context; such cells are hard to kick into uncontrolled proliferation. Most lentiviruses are more cytopathic than retroviruses, and may thus kill their hosts before they become transformed. But this is not a clear prescription for the use of non-retrovirus vectors, even if they do not integrate their genomes close to promoters (lentiviruses do not cause cancer—the malignancies in AIDS are related to loss of immune surveillance and superinfection). We do not have enough data.
Lois et al.3 point out that the generation of transgenic animals by lentivector-mediated transduction of embryonic stem cells shows that these vectors are not silenced by developmental processes, whereas retroviruses in the same experimental situation become silenced by methylation-dependent (and other) mechanisms. It would be useful to know if this difference was due to the avoidance of promoter regions by the lentiviral vectors.
In the trial referred to, more than 5 x 106 cells with MLV integrations were injected into each child. As Wu et al.2 have pointed out, if 20% of integrations are near transcriptional start sites, there will be 1 million integrations distributed among the 18 214 RefSeq genes, or an average of 55 integrations into the 5’ region of the LMO2 locus per treatment. So the chance of having an insertion that favoured proliferation is not remote. It should be remembered that the aim of therapy was to produce a proliferating clone of non-terminally differentiated cells; this may be a difficult process to fine-tune, for a self-replicating population of cells is not the same as a tumour. A hard look at different systems is justified.
References
1. Schroder AR, Shinn P, Chen H, Berry C, Ecker JR, Bushman F. Hiv-1 integration in the human genome favors active genes and local hotspots. Cell 2002; 110:521–9.[CrossRef][Web of Science][Medline]
2. Wu X, Li Y, Crise B, Burgess SM. Transcription start regions are favored targets for MLV integration. Science 2003; 300:1749–51.
3. Lois C, Hong EJ, Pease S, Brown EJ, Baltimore D. Germline transmission and tissue-specific expression of transgenes delivered by lentiviral vectors. Science 2003; 295:868–72.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||