The many platform and poster presentations at the last annual meeting of the American Society for Gene Therapy (ASGT) make it clear that nonviral, integrating DNAs have reached a tipping point in establishing themselves as potential vectors for gene therapy.1 These vectors currently include the Sleeping Beauty (SB) transposon system,2 the first such vector used in gene therapy experiments,3 the ϕC31 and ϕBT1 integrase systems from Streptomyces,4,5 and more recently the insect piggyBac6 and fish Tol27 transposon systems. Several other transposon systems are under development.
The common motivations to develop nonviral vectors are their lack of major immunogenic responses, the ease of purification and quality control, and the ability to insert precise sequences into chromosomes (Figure 1). The primary considerations used to evaluate these nonviral vectors have been their 1) activity to insert specific genetic sequences into chromosomes, 2) genetic cargo capacity, 3) sensitivity to different levels of the relevant recombinase, 4) integration-site preferences, and 5) their propensity to induce chromosomal rearrangements. The first three issues pertain to the particular nonviral systems with respect to the capability and ease of achieving a desirable therapeutic end, whereas the latter two concern safety issues, both theoretical and empiric.
Owing to its early debut, SB most often serves as a reference for gene-insertion activity, site specificity, and other parameters of recombination. In spite of its successful development in several animal models for gene therapy to treat human disease, the SB transposon has two perceived limitations: 1) the highest transposition rates require an optimal ratio of transposase to transposon and 2) transposition rates appear to be inversely proportional to genetic cargo size. High transposition levels require an optimized ratio of SB transposase molecules per transposon; fewer transposase molecules will not effectively hold together the two ends of the transposon to form a functional synaptic complex whereas a surplus of SB transposase molecules apparently will quench the reaction (called overexpression inhibition), potentially by keeping the appropriate couples of transposase molecules from interacting.
Figure 2 illustrates one model as an example of how overexpression inhibition may occur and interfere with transposition rates. Consequently, transposition occurs only during a narrow window of time because the expression level of the transposase (green line, Figure 2, right) is transient, resulting in limited periods (two blue lines) during which there is a functional ratio of the two components (light blue region in Figure 2, right). If overexpression inhibition did not occur, then transposition will continue as long as there is sufficient transposase. Moreover, if only two transposase molecules were required for transposition, the process could be active earlier and last longer than for SB transposition. This has been suggested for the Tol2 transposon system in zebrafish embryos.7 Thus we might expect higher activities over a longer time with a transposon system that is not subject to overexpression inhibition and requires only two transposase molecules per synaptic complex (red line in Figure 2, right).
With respect to the second perceived limitation, the SB transposon system has a limited cargo capacity in that expression cassettes greater than 5 kb are delivered to chromosomes inefficiently. In contrast to SB, Tol2 and/or piggyBac have exhibited neither overexpression inhibition nor limitations on size of the genetic cargo, which might be drawbacks for gene therapy with SB-based vectors (Figure 3, left). This is a bit surprising because once the genetic cargo length exceeds that of the plasmid carrier, the size of the transposon should not make much difference (Figure 3, right). That is, the ends of the transposon are separated both by its cargo and by the plasmid vector. One would expect that the shorter loop (Figure 3, right) would be the determining factor in complex formation. One way to reconcile the differences in effects of transposon size on transposition is to take nuclear transport into consideration. If a vector-integrase complex forms prior to entry into the nucleus with one system but not another, then overall delivery to chromosomes could be differentially affected and size would appear to be a factor in overall rates of transposition rather than the process as a whole. Regardless of mechanism, the ability to overcome both limitations on cargo size and defined levels of recombinase represent major new developments in nonviral vectorology.
Safety is a paramount issue in gene therapy. The nearly random integration of SB transposons has been considered risky because of the potential of integration into, or in the neighborhood of, a protooncogene or a tumor-suppressor gene. The advantage of the phage integrase system is the limited number of preferred (safe) sites into which the vectors integrate. Hence, projects detailing further modifications of both SB and ϕC31 to improve integration site specificity are of considerable interest, although the observation of some chromosomal rearrangements following delivery of ϕC31 constructs has led to concerns of genetic instability.8,9 Nevertheless, both SB and ϕC31 have been used for gene delivery in many sets of mice without any observed adverse events. Indeed, in order to find genetic effects of transposons on endogenous genes, several labs have produced multiple lines of mice that harbor in every cell SB transposons and transposase genes regulated by relatively strong, promiscuous promoters. Thus, except for mice with transposons especially designed to cause genetic havoc in cells, there has been no evidence of increased cancer or leukemia in multiple generations of mice.10 Clearly, more data are needed to understand fully the mechanisms of action of the various recombinase systems in the various cellular venues in which tests are run.
The good news is the enthusiasm of the growing number of labs to investigate gene delivery by nonviral, integrating agents. However, the short histories of the two most investigated integrating vector systems, SB and ϕC31, have shown that conclusions based on experiments in one cell type or for one set of vectors do not always hold in another setting. As we continue to develop the various vector systems, with spirited discussions that naturally focus on their respective shortcomings and uncertainties, we need to consider the consequences of overhyping early results with severe criticisms of alternative approaches. There may be adverse effects of overenthusiasm. As noted by Mark Kay at the 2006 annual meeting of the ASGT,11 the overall reputation of gene therapy is not as positive as most of us would like. That is not surprising: the principal sources of criticism of gene therapy are ASGT researchers. In the oft-quoted words of Walt Kelly, creator of Pogo: “We have met the enemy and he is us.” We need to keep in mind that we work with particular systems because of their advantages (which we want to increase) rather than because they have the fewest disadvantages.
The differences seen in the activities of the various nonviral integrating systems are modest compared to two major challenges: 1) delivery of vectors to desired tissues and 2) controlling integration into or near to transcriptional units. There are problems in directing the DNA to specific cells, transmission of the DNA through the plasma membrane, and subsequent incorporation into chromosomes. These problems should be the focus of the entire community, and solutions will likely come from many approaches. At the same time, our understanding of the consequences of insertional mutations is growing as the cancer community develops a deeper knowledge of the mechanisms involved with these diseases. Moreover, several of the new nonviral integrating systems may allow targeting of specific sites.
In sum, our experience over just the past few years with nonviral integrating vectors indicates exceptional promise for their future use in gene therapy.