Toward a Safer Integration Profile of MLV-based Retroviral Vectors

The stable retroviral integration of therapeutic transgene cassettes into patients' hematopoietic stem and progenitor cells enables gene therapy for various disorders of the blood and immune systems.^1,2,3,4 However, adverse events related to insertional mutagenesis were observed in the first clinical gene therapy trials.^5,6 These adverse events were caused mainly by the insertional activation of proto-oncogenes, such as LMO2 and MDS/EVI1, which occurred as a consequence of the retroviral integration preferences and the strength of the enhancer of the first-generation long terminal repeat (LTR)-driven gammaretroviral vectors. Two concepts have been proposed to achieve a “safer” integration profile. The first is targeted integration into genomic “safe harbors” by designer nucleases. Sadelain et al. defined safe harbors as chromosomal sites where transgenes can be stably and reliably expressed without adversely affecting endogenous gene structure or expression.⁷ The second concept is directing integration away from potentially harmful chromosomal sites, such as promoters of proto-oncogenes, by exploiting retroviral vectors with a more random integration pattern or, alternatively, by interfering with the integration and chromatin-tethering processes during retroviral integration (Figure 1).

Figure 1.

Strategies to prevent insertional mutagenesis. (a) Toward a safer vector architecture: from a long-terminal repeat (LTR)-driven vector to a self-inactivating (SIN) vector with a more physiological internal promoter. (b) Retroviral integration as a concerted action of retroviral integrase and host cell tethering factor. Integration-site preferences of murine leukemia virus (MLV)-based gammaretroviral (γRV) and HIV1-based lentiviral (LV) vectors. γRV integrations cluster with BET chromatin domains and are associated with promoter/enhancer sequences. LV vectors integrate preferentially into actively transcribed genes. BET-independent (−BET) γRV integrate less in proximity to transcriptional start sites and promoter/enhancer regions. cDNA, complementary DNA; SD/SA, splice sites; p(A), polyA site. Modified from artwork courtesy of Chris Baum.

In a paper in Molecular Therapy—Nucleic Acids, El Ashkar et al. from Rik Gijsbers's group in Leuven, Belgium,⁸ describe a promising new way to direct the integration profile of gammaretroviral murine leukemia virus (MLV)-based vectors away from their natural preference for transcriptional start sites (TSSs), CpG-rich islands, and promoter/enhancer regions.^9,10 Previous work by the authors,¹¹ as well as work from other groups,^12,13 identified the bromodomain and extraterminal (BET) family of proteins as the cellular tethering factors for MLV integrase (IN). El Ashkar and colleagues determined the specific amino acids and regions in IN that are responsible for binding to BET-proteins.⁸ Deletion of the 20 C-terminal amino acids of IN or introduction of a single-point mutant (IN_W390A) led to the creation of MLV-based vectors that no longer exhibited a typical MLV integration profile with a preference for promoter/enhancer regions and that associated less frequently with proto-oncogene TSSs. Their comparison of BET-independent MLV vector integration–site selection to a set of (epi) genetic modifications (distance to TSS, DNase I hypersensitivity sites, CpG islands, association to specific histone codes) showed that overall integration is distributed more randomly and is less frequently associated with markers of active chromatin. Accordingly, gene expression from these vectors was slightly decreased. The authors demonstrated that viral titers produced from the IN_W390A construct were similar to those produced from wild-type vectors, whereas the C-terminal IN truncation led to slightly reduced (one-sixth the production and twice the integration defect) viral titers. Importantly, titers are still in a range likely to be in accordance with the needs of gammaretroviral vector production for clinical gene therapy trials.

The current work by Gijsbers's group also extends previous studies aimed at deciphering the tethering mechanism and demonstrates some parallels to the HIV IN tethering factor lens epithelium–derived growth factor (LEDGF). LEDGF also binds HIV IN, protects it from proteasomal degradation, and tethers the lentiviral preintegration complex to the chromatin.^14,15,16 As in the BET-independent MLV IN setting, lentiviral preference for integration into active genes is decreased, arguing for a shift in integration preference.

Another strategy to shift integration preference of retroviruses entails substitution of the chromatin-interacting binding domain of the tethering factor with histone-binding domains,^17,18 as demonstrated for LEDGF and the histone-binding CBX1. Proof of principle was demonstrated as LV vectors integrated outside of genes, near regions occupied by CBX1. Ideally, however, this strategy requires abrogation of endogenous LEDGF and overexpression of the chimeric LEDGF-CBX1 fusion, which is very demanding from a practical standpoint. Therefore, direct IN changes as discussed here represent a more elegant and straightforward approach. In addition, complementing retroviral/lentiviral vector approaches with new developments to target stable integration sites should also be carefully considered (see Table 1). An example is the use of designer nucleases and homologous recombination of a donor sequence to reach “safe harbors” or to specifically repair defective genes as recently demonstrated in hematopoietic stem cells.¹⁹

Table 1. Strategies to influence integration-site selection.

Interestingly, Aiyer and co-workers²⁰ recently used solution nuclear magnetic resonance and protein interaction studies to demonstrate that the C-terminal tail peptide region of MLV IN is important for the interaction with BET proteins. It is reassuring that they also observed that disruption of MLV IN-BET interaction through truncation mutations affected the global targeting profile of MLV vectors.

What is required to translate these findings into the (pre)clinical arena? First, the new BET-independent MLV vectors should be tested in relevant preclinical safety and bone marrow/hematopoietic stem cell transplantation assays to validate their performance and to provide proof of concept for improved safety. Second, BET-independent MLV packaging constructs should be combined with next-generation gammaretroviral self-inactivating vectors, given that their vector architecture with weaker, more physiological internal promoters (representing weaker cis-activators of neighboring genes) will probably lead to a further increase in safety.

In summary, the findings reported by El Ashkar et al. will help achieve better control of integration-site preferences of retroviral vectors and, together with improvements in vector design, will increase the safety of gammaretroviral vectors for gene therapy.