Human gene therapy is a revolutionary approach to medicine that holds great promise for the cure of many devastating diseases that presently have no therapies. But gene therapy, like every other major innovative technology, needs time to develop and mature. Retroviral vectors are valuable tools for gene transfer technology and have shown great potential as vectors for gene therapy. Despite the progress made in the field of retroviral gene therapy in the past 20 years (Sinn et al., 2005) and despite the numerous positive results obtained in pre-clinical and clinical trials (Aiuti et al., 2002; Cavazzana-Calvo et al., 2000), further investigation will be required before retrovirus vectors can be used efficiently and safely in a clinic setting. Further advancements in vector design are required to improve the efficiency and specificity of gene delivery and expression for in vivo applications. To support these efforts, a better understanding of retrovirus-cell interactions is essential for the coherent design of engineered targeted vectors. In addition, a thorough evaluation of the vectors’ integration site preferences and the risks for insertional mutagenesis is needed. These studies should be coupled with the development of strategies to direct vector integration to specific sites in the genome for increased biosafety. Finally, the wide clinical application of these vectors for gene therapy will also depend on the availability of efficient large-scale manufacturing procedures. Hence, the optimization of upstream bioprocesses, the adaptation of packaging cells to serum-free media, the development of downstream processing strategies and the establishment of quality control tests for clinical-grade vector stocks are critical to advances in retroviral gene therapy.

Downstream processing of retroviral vectors is particularly challenging because of the low titers generated by most commonly used packaging systems and the instability of the viral particle. This thesis describes the development of two innovative and valuable methods for retroviral vector purification: heparin affinity chromatography and rate zonal ultracentrifugation. These methods were included in two complete strategies for the purification of the retroviral particles from crude supernatants to highly purified vector preparations which are described in detail in chapters II and III. A comparative analysis of these strategies is presented in this section followed by perspectives for future work.

Both strategies presented render high levels of purity as revealed by SDS-PAGE analysis. The electrophoretic profiles of the purified products showed the main viral proteins and a few other minor bands, many of which are likely to be host proteins incorporated into the viral particle or intermediate viral products as confirmed in Chapter III. Based on the visual inspection of the gels, the levels of purity reached using either strategy were estimated to be over 90%. Although the exact composition of the retroviral particle is still unknown, the impossibility to further purify the rate zonal purified virus preparation by size-exclusion chromatography or to digest proteins in this preparation with subtilisin treatment allowed us to safely estimate the levels of purity in the absence of a purified standard.

In terms of process time, both strategies are comparable. Microfiltration and ultra/diafiltration of a 2-L batch using the described methodology requires 6 hours of processing. These initial filtration steps were employed in both strategies resulting in the clarification, 20-fold concentration and partial purification and crude viral stocks with no significant losses of infectivity. To reach high levels of purity either rate zonal ultracentrifugation or 2 steps of chromatography were needed. While rate zonal centrifugation required 4 h of processing, chromatography purification was carried out in 5 h (1.5 h for heparin affinity chromatography and 3.5 h for size-exclusion chromatography). The total processing time for either strategy was around 10 h. Therefore, a coordinated effort would allow completion of either process within 1 working day.

Surprisingly, both strategies resulted in roughly the same overall recovery of infective retroviral particles which was approximately 40%. These overall recoveries are superior to those reported using other purification techniques (Kuiper et al., 2002; Williams et al., 2005; Ye et al., 2004). Few reports have shown a complete purification strategy for MoMLV-derived vectors where final overall recoveries are presented. An example was provided by Kuiper and collaborators who estimated that a final overall recovery of 5% would be obtained using their 3-step downstream processing protocol based on hydroxyapatite chromatography (Kuiper et al., 2002).

Both strategies presented here should be useful for retroviral particles regardless of the gene or Env-protein carried by the vector or packaging cell type used for production. None of these variables greatly affects the size and density of the viral particles which should allow the separation of various vectors by rate zonal ultracentrifugation. The general applicability of heparin affinity chromatography was demonstrated in Chapter IV. The usefulness of this methodology was not affected by the presence or absence of different Env-proteins. Moreover, vectors derived from 293 and HT1080 cells bind heparin columns with similar efficiency and affinity. The fact that these cell lines are currently the most attractive cell lines for retrovirus production (Merten, 2004) further supports our conclusions on the general utility of this method.

Each strategy also offers distinct advantages. Adsorptive chromatography can be easily scaled-up by increasing the column diameter. However, the possibility to scale-up the ultracentrifugation protocol depends on the availability of costly high capacity ultracentrifuge equipment suitable for retrovirus purification purposes that unfortunately, at the present time is not available in most laboratories. Additionally, chromatography is a very simple and reproducible method. Compared to chromatography, rate zonal centrifugation is labor-intensive and prone to inter-operator variability. These features make chromatography technologies ideal for use at industrial scale.

On the other hand, most chromatography methods are unlikely to remove significant amounts of defective vector forms and/or cell membrane vesicles (Vellekamp et al., 2001). The ability of rate zonal ultracentrifugation to separate these closely-related structures from retrovirus preparations was clearly shown in Chapter III. To date, no other technique used for MoMLV purification has shown the ability to separate these contaminants from vectors stocks. Moreover, these contaminants are extremely difficult to detect because they possess a composition similar to that of the retroviral particles, making their detection by SDS-PAGE analysis impossible in practice. This limitation can be circumvented by the use of electron microscopy which can successfully distinguish cell membrane vesicles from viral particles (active or inactive) in the hands of expert microscopists (Bess et al., 1997; Gluschankof et al., 1997).

The possible implications of having defective vector forms or cell membrane contaminants in retroviral clinical-grade preparations are rarely discussed in the literature. Moreover, to the best of my knowledge, there are no specific regulations that demand the quantitation or elimination of these contaminants from retroviral vector stocks for clinical studies. In contrast, the “Guidance for Human and Somatic Cell Therapy and Gene Therapy” first issued by the FDA’s Center for Biologics Evaluation and Research (CBER) in March 1998 (CBER, 2001), specifically states that the measurement of total vs. infectious particle ratios is required for adenoviral vectors and they recommend that it does not exceed 100:1 in the final product. These considerations for the use of adenoviral vectors are largely due to the fact that adenoviral particles are potentially toxic themselves. It should be also noted that a CsCl density gradient method that efficiently separates complete from empty adenoviral particles has been available for a long time (Niiyama et al., 1975), which allows researchers to meet this specifications.

Given that both strategies presented are comparable in terms of processing time, recovery of infective particles, ability to purify various types of retroviral vectors and reach high levels of purity that are acceptable for clinical applications, the choice of strategy may ultimately depend on the process scale and availability of specific laboratory equipment required. As a result, the rate zonal ultracentrifugation method will be particularly useful in academic laboratories equipped with ultracentrifuges while chromatography will be more attractive for large-scale vector production facilities.

In addition to the preparation of clinical-grade vectors for gene therapy, the methods developed in this work could be applied to fundamental research. The ability of rate zonal ultracentrifugation to separate defective virus forms may be particularly attractive for investigators interested in studying differences between defective and functional virus populations. Additionally, this method is extremely useful for obtaining highly purified virus preparations suitable for studying the composition of the poorly characterized retroviral membrane. These studies have been greatly complicated by the presence of contaminating cell membrane vesicles that co-purify with virions using standard ultracentrifugation techniques (Ott, 1997; Ott, 2002; Trubey et al., 2003). As described in Chapter IV of this thesis, retroviral particles are able to specifically interact with heparin ligands under physiological conditions regardless of the vectors’ cellular origin or Env-protein used for pseudotyping. These intriguing results may motivate further investigation on the nature of retrovirus-heparin interactions using heparin affinity chromatography.

By taking advantage of the expertise gained on virus purification, we are currently attempting to identify host proteins on the retroviral membrane that could bring light to the mechanism of virus attachment to target cells and contribute to the development of targeted retroviral vectors for in vivo gene therapy applications. For this purpose, a detailed characterization of the composition of the purified viral particles by proteomic analysis was performed in our laboratory. Three host-derived cell adhesion proteins were identified by 1-D-electrophoresis coupled with mass spectrometry analysis. The presence of these proteins on the virus membrane was confirmed by immunogold electron microscopy studies. Virus neutralization assays are being conducted in order to establish whether these proteins influence virus infectivity.

This thesis represents a framework from which to direct future research in retrovirus purification as there exist numerous opportunities for further optimization. First, taking into account that the vast majority of impurities contained in the vector supernatant come from the serum added to the culture media, progress towards development of serum-free adapted packaging cell lines would greatly facilitate downstream processes. To date, reports demonstrating successful production of retroviral vectors in serum-free media are scarce (McTaggart and Al-Rubeai, 2002). Alternatively, production of vectors in very low serum media helps reduce the levels of contamination and more importantly, often results in a higher specific vector productivity than with the 10% concentrations of serum typically used (McTaggart and Al-Rubeai, 2002; Merten, 2004). It would be interesting to test the performance of the strategies presented in this thesis for the downstream processing of supernatants produced in serum-free or low protein media. However, because serum is believed to play a role in the protection of viral particles, particularly when exposed to stressful conditions such as those encountered during concentration, it wouldn’t be surprising that the presence of low concentrations of serum at these early stages, as opposed to serum-free conditions, result in improved vector recoveries.

Ultrafiltration is the preferred method for large-scale processing of retroviral particles because it allows processing large volumes of supernatant in a relatively short period of time. Although stirred cell technologies are particularly attractive for ultrafiltration because they are very gentle and thus, result in high recovery of infective particles as shown in this work, the maximum capacity of a stirred cell unit is 2-L. Larger volumes of retroviral vectors (8 to 10 L) could be processed by tangential-flow filtration (Kotani et al., 1994; Makino et al., 1994; Paul et al., 1993) and thus, it would be appealing to test these technologies for large-scale production.

Due to the high amounts of heparin-binding contaminants contained in serum, purification of retroviral vectors by heparin affinity chromatography in a single step was not possible. In this work, efficient removal of heparin-binding proteins was achieved by size-exclusion chromatography. Alternatively, one could achieve removal of these contaminants by using the rate zonal centrifugation strategy described in this work. This approach is very attractive since it would combine the advantages of both purification strategies developed in this thesis, namely the scalability of chromatography and the high resolution of rate zonal ultracentrifugation. Although this would be an ideal situation in terms of purity and process scalability, it should also be noted that the theoretical overall recovery of infective particles, calculated as a factor of the recoveries obtained in each individual step, is expected to be lower (26.4%) than that obtained with the purification strategies described in this thesis.

Another way to optimize the heparin affinity chromatography method would be to use membrane chromatography instead of column chromatography. Membrane chromatography is attractive because it provides higher binding capacities for large bioproducts such as viruses and higher flow rates than the traditional column chromatography. Processes using membrane chromatography devices can be easily, cost-effectively, and linearly scaled up using multiple cartridges in parallel. Mustang Q anion-exchange membrane capsules have been successfully employed for the purification of lentiviral vectors (Slepushkin et al., 2003).

Engineering vectors by inserting tags or chemically modifying the viral Env-protein structure without reducing or eliminating the viruses’ ability to transduce cells has proved to be difficult (Katane et al., 2002; Palù et al., 2000; Pizzato et al., 2001; Tai et al., 2003). This has been partly attributed to the inability of the Env-protein to provide fusogenic functions for viral entry once its structure has been modified. Provided the composition of MoMLV particles was more deeply characterized, proteins other than the viral Env-protein could be chosen to insert tags for purification or targeting purposes. For instance, host proteins incorporated into the MoMLV could be genetically modified to contain such tags possibly without compromising virus infectivity. To date, this approach has not been tested likely due to our poor current understanding of the viral membrane composition and the lack of suitable purification methods for virus characterization studies.

Finally, developments in lentiviral vector manufacture and design have greatly benefited from the many years of experience and advances made with retroviral vectors. Additionally, the strategies presented in this work may also be exploited for the purification of these vectors. Since both types of vectors share common physical properties, it is likely that the rate zonal ultracentrifugation technique presented here for MoMLV-derived vectors will also be useful for the purification of lentiviral vectors. Moreover, the fact that soluble heparin also inhibits lentiviral vector transduction to target cells (Guibinga et al., 2002) and that HIV-1 particles were shown to specifically interact with heparan sulfate (Cladera et al., 2001; Mondor et al., 1998; Roderiquez et al., 1995; Vives et al., 2005) strongly suggests that these particles can also be purified by heparin affinity chromatography.

To conclude, the methods we have developed represent a significant contribution to the technological progress in the field of gene therapy and may also open the door to a finer characterization of retroviral particles.