The Promise of Polyclonality

I n this issue of Human Gene Therapy, Enssle and colleagues report that canine hematopoietic cells modified with lentiviral vectors establish polyclonal hematopoiesis, without progressing to clonal dominance in an extended observation period that lasted almost 6 years for two dogs (Enssle et al., 2010). Evidence of polyclonality was obtained by investigating the band pattern produced by linear amplification-mediated PCR (LAM-PCR) and by identification of the insertion sites. Although the number of animals was limited, this study is of great interest as it represents the longest follow-up of transplanted lentivirally modified hematopoietic cells available to date. With the advent of clinical trials in which lentivirus-based gene transfer reveals its potential to treat severe genetic diseases (Cartier et al., 2009), studies addressing the long-term integrity of the gene-modified cell population become increasingly important.

In the report by Enssle and colleagues, the duration of the gene-marked hematopoiesis in combination with the finding of the same integration site in both lymphoid and myeloid progeny cells strongly argue for the successful transduction of long-lived hematopoietic stem cells (HSCs). Although clones with insertions located in or close to proto-oncogenes were detected and shown to persist for up to 2188 days posttransplantation, there was no indication of uncontrolled outgrowth. The data suggest that the hematopoietic mosaic introduced by the semirandom distribution of lentiviral vector insertion sites in the cellular genome may stabilize in a peaceful coexistence of clones with either neutral or suspicious integration events (Enssle et al., 2010). These results are consistent with findings from a clinical trial in which lentivirally modified hematopoietic cells were used to treat patients suffering from adrenoleukodystrophy (ALD) (Cartier et al., 2009). Here, several hundred clones were detected and serial analyses of peripheral blood indicated that no single clone tended to dominate the numerous competitors, in contrast to findings made in some clinical trials performed with “first-generation” gammaretroviral vectors (Hacein-Bey-Abina et al., 2008; Howe et al., 2008; Stein et al., 2010). Although the follow-up of the lentiviral clinical trial (Cartier et al., 2009) is shorter and several methodological differences limit the ability to directly compare its results with the gene-marking study in dogs (Enssle et al., 2010), the synopsis of the two studies strongly supports the concept that lentiviral vectors, compared with previously used gammaretroviral vectors, not only have an improved potential to transduce HSCs but may also reduce the genotoxic risk of untargeted transgene insertion (Montini et al., 2009). This was not necessarily predictable, as studies in cultured cells have revealed the potential of lentiviral vector insertions to activate or truncate cellular genes (Hargrove et al., 2008; Arumugam et al., 2009; Bokhoven et al., 2009; Maruggi et al., 2009; Modlich et al., 2009), potentially leading to clonal dominance (Bokhoven et al., 2009; Modlich et al., 2009); and studies in mice have indicated that HSCs, as compared with more mature hematopoietic cell populations, are more susceptible to clonal dominance and malignant transformation in response to genetic lesions (Newrzela et al., 2008; Kustikova et al., 2009), although this may depend on the nature of the initiating lesion (Cozzio et al., 2003; Krivtsov et al., 2006). Polyclonality, as defined by the presence of numerous clones with different insertion sites, is thus not necessarily a promise of a favorable outcome. One may also expect that the chance of inducing a dangerous clone may directly correspond to the number of insertion sites and the number of engrafting HSCs.

But apparently, both the dog model and the ALD clinical trial show that clones with potentially growth-promoting insertions do not necessarily dominate hematopoiesis. This important observation may reflect cell-intrinsic and systemic balancing mechanisms.

The underlying cell-intrinsic mechanisms include the nature of the insertion sites and the control elements of the transgene. Of the numerous proto-oncogenes present in the mammalian genome that can potentially be activated by insertional mutagenesis, only a subset may indeed be potent enough to trigger long-term expansion of a gene-modified clone, thus increasing the chance to acquire the additional mutations that are necessary for malignant transformation. An especially worrisome subset of these proto-oncogenes may even be able to directly trigger genetic instability, as reported for EVI1 (Stein et al., 2010). With increasing insights into the mechanisms of retroviral integration site selection and high-throughput mapping of vector insertions (Lewinski et al., 2006; Cattoglio et al., 2007; Wang et al., 2008; Felice et al., 2009), it can be tested whether lentiviral vectors, or certain variants thereof, can be developed that largely avoid these genomic danger zones. Regarding the control elements of the transgene, various studies support the concept that the risk of cellular gene activation after transgene insertion is directly proportional to the strength of its enhancer elements (Hargrove et al., 2008; Arumugam et al., 2009; Maruggi et al., 2009; Modlich et al., 2009). If a vector lacking a strong enhancer hits a potentially dangerous site, this may not be functionally consequential, because of its inability to activate a proto-oncogene beyond the threshold required to execute its transforming program. Even relative differences in the integration pattern and the transgenic control elements may thus lead to major improvements of the long-term outcome. A final cell-intrinsic parameter would be related to potential signaling effects of the protein(s) encoded by the transgene. When correcting genetic diseases in which a defined protein is to be reconstituted, this variable may be the most difficult to utilize for increasing the safety of somatic gene transfer. Although in the dog model reported in this issue of Human Gene Therapy and in the ALD clinical trial, signaling effects of the transgenic protein are unlikely to be responsible for the failure to detect dominant clones, the impact of the expression cassettes used deserves further investigation.

The mechanisms operating on the systemic side are much less well investigated, although they may offer great potential to improve the prospects of hematopoietic gene therapy. The key question concerns the extent to which the proliferative stress operating on a given stem cell clone, both in the repopulation phase after myeloablative conditioning and in the long-term support of hematopoiesis, can be reduced by the coexistence of competing clones. Regardless of whether hematopoiesis occurs in a mode of clonal succession or by simultaneous activity of all HSCs, a larger pool of competing HSCs should reduce the number of clonal divisions, and thus lower the risk of secondary genetic lesions. If the transgene confers a selective advantage, it is not the total number of HSCs but specifically the number of gene-modified HSCs that counts in this equation. Therefore, the transduction protocols should be considered another important parameter that determines the likelihood of clonal dominance. Enssle and colleagues report findings made in dogs receiving HSCs that were cultured for a relatively short time (overnight or for a maximum of 2 days) (Enssle et al., 2010), thus limiting the loss of repopulation potential associated with extended culture in the presence of mitogenic cytokines, as typically used for gammaretroviral gene transfer (Mostoslavsky et al., 2005). If this concept is correct, every attempt should be made not only to improve the quality of the cell culture conditions, but also to reduce the potential unspecific toxicity of vector preparations.

In conclusion, two approaches can be envisaged to realize the therapeutic promise of polyclonality. The first follows the principle of “the more the better,” accepting the risk of rare insertions in dangerous loci and counting on the combination of improved transgene design and polyclonal competition to prevent clonal outgrowth. The second is based on an attempt to define the minimal clone number required to establish stable hematopoietic equilibrium: assuming about 100 clones to be sufficient to support life-long hematopoiesis and the risk of a dangerous insertion to be lower than 1:1000, this approach may provide the greatest degree of safety. Preclinical studies with long-term follow-up, as exemplified by Enssle and colleagues, may test these concepts.