4.1.Overview

The aim of this project is to achieve all the essential pre-clinical studies required for ex vivo gene therapy for recessive dystrophic epidermolysis bullosa (RDEB) using retroviral vectors. We have initially constructed a MSCV derived retroviral vector containing a functional copy of the COL7A1 cDNA. We have produced and purified high titers and quality viral suspensions containing this COL7A1 retroviral construct harbouring different envelope proteins. We have subsequently optimised the transduction conditions to obtain highly efficient gene delivery to human primary RDEB keratinocytes in culture. The corrected keratinocytes express and secrete type VII collagen molecules of expected molecular weight (290 kDa), and show no reduction in their growth capacity. However, grafting onto nude mice of genetically engineered RDEB skin derived from keratinocytes tranduced with this construct did not allow to assess long-term expression of the transgene in vivo. Moreover, the recent report of leukemia by insertional mutagenesis in two children treated by retroviral vectors in an ex vivo gene therapy trial for X-SCID lead us to amend our strategy . Indeed, this major complication leads us to consider the use of safer vectors (Self Inactivating (SIN) vectors) as a prerequisite to treat RDEB patients with retroviral or lentiviral COL7A1 containing vectors. We have now constructed 2 of these safer retroviral and lentiviral SIN vectors containing COL7A1 cDNA, which are described in detail in this report. However, a marked reduction in the production efficiency of these COL7A1-SIN vectors is expected, and optimisation of the production conditions remains to be achieved. The use of safer versions of retroviral and lentiviral COL7A1 constructs, together with alternative in vitro and in vivo skin equivalent models to assess long-term expression of the transgene, are the reasons for the extension of the initial project.

4.2. Design and construction of the COL7A1 gene therapy vectors

We have designed two different types of vectors to transduce primary RDEB keratinocytes. These include two retroviral vectors derived from the Moloney Murine Leukemia Virus (MoMuLV) and one lentiviral vector derived from the Human Immunodeficiency Virus I (HIV-1) (Figure 1). The rational of these constructs is developed below.

4.2.1 Retroviral vectors

In the first construct (Fig. 1A), the COL7A1 cDNA is under the control of a modified MoMuLV LTR (Long Terminal repeat) promoter in a "classical" retroviral vector. In the second construct (Fig. 1B), the transgene is in a SIN retroviral vector and is expressed under the control of the Keratin 14 promoter to permit specific expression in the basal layer of the epidermis.

The MoMuLV promoter based vector :

The SIN vector with a keratin 14 promoter :

Because the transcriptional activity of wild-type LTRs can affect the expression of host proto-oncogenes, we wanted to improve the biosafety of the COL7A1 retroviral vectors using the SFG vector, a SIN vector, which is derived from the original MFG vector but in which the enhancer of the 3' LTR is deleted. In this type of vectors, deletion mutations were introduced into essential regions of the 3' LTR to inactivate the 5' LTR of the integrated provirus (which is derived from the 3' LTR of the vector construct). This concept of transcriptional inactivation of the provirus is also termed self-inactivation, and these vectors are known as SIN (self inactivating) vectors . The SIN strategy is a major improvement in the biosafety of the vectors. The drawback of the inactivation of the transcriptional regulatory elements of the proviral LTRs is a substantial loss in viral titer, which is at least 10 to 100 fold lower than the parental retroviral vector. As a consequence of the LTR promoter inactivation, the transgene must be controlled in this type of vector by a heterologous promoter. Therefore, we have chosen the keratin 14 (K14) promoter to ensure specific COL7A1 cDNA expression in basal keratinocytes of the epidermis (Fig. 1B).

4.2.2 Lentiviral vector

4.3. Production of viral vectors

We have transiently produced these three vectors using a co-transfection procedure, to avoid the generation of replicative competent particles. The viral construct was co-transfected in 293T cells with the plasmid encoding the helper functions and a plasmid encoding the envelope protein (Ampho, VSV-G or GALV, see section 4.3. RDEB keratinocytes transduction).

4.3.1. Retroviral vectors

The recombinant genome containing the COL7A1 cDNA was co-transfected with a helper plasmid expressing the GAG and POL cDNA coding for the capsid proteins and the reverse transcriptase, respectively, and with a plasmid encoding the ENV protein

4.3.2. Lentiviral vector

The recombinant genome containing the COL7A1 cDNA was co-transfected with a helper plasmid expressing the GAG, PRO and POL cDNA coding for the capsid proteins, the protease and the reverse transcriptase, respectively, with a plasmid expressing the ENV cDNA coding for the envelope protein, and a plasmid coding for the REV protein, responsible for the nuclear export of the mature viral mRNAs via the RRE site.

Using these procedures, the recombinant genome expressing the COL7A1 cDNA was packaged into 293T cells. The cell supernatant containing the virions was collected 48 hours later and frozen. At present, only the cell supernatant containing the "classical" COL7A1 retroviral construct has been tested on RDEB primary keratinocytes. The SIN retroviral and the lentiviral COL7A1 constructs have not yet been tested on REDB primary keratinocytes.

Several parameters play a role in the transduction efficiency of primary cells. Each of these parameters has been tested in order to achieve the best transduction efficiency in primary keratinocytes.

4.4.1 The choice of the envelope protein

Using a retroviral MSCV-EGFP vector expressing E-GFP (Enhanced Green Fluroscent Protein) under the control of the MSCV LTR, we have infected RDEB primary keratinocytes and have determined the best transduction conditions by FACS (Fluorescent Activated Cell Sorter) analysis of the infected cell population. Three different envelope protein have been tested : the Amphotropic envelope protein (Ampho) of the MoMuLV, the VSV-G (G-protein of the Vesiculous Stomatitis Virus) envelope protein and the GALV (Gibbon Amphotropic Leukemia Virus) envelope protein. These three envelope proteins have the potential to infect a wide range of cell types from different species. However, while retroviral particles harbouring Ampho or GALV envelope proteins are fragile and cannot be prepared by ultracentrifugation of viral supernatants, VSV-G-pseudotyped vectors can be concentrated 100-300 fold by ultracentrifugation. The major disadvantage of the VSV-G protein is its cytotoxicity.

The ampho envelope protein gave excellent results, as good as the VSV-G envelope protein and better than the GALV envelope protein. Since the titer which we obtained was high enough and minding that the VSV-G protein, because of its cytotoxicity, could compromise a clinical trial, we have decided to use the "classical" retroviral COL7A1 construct harbouring the Ampho envelope protein.

4.4.2 Polybrene concentration, cell density and Mutliplicity Of Infection (MOI)

Infection with retroviruses is facilitated greatly by polybrene. This is a small, positively charged molecule that binds to cell surfaces and neutralizes surface charge. This apparently allows the viral glycoproteins to bind more efficiently to their receptors, because it reduces the repulsion between sialic acid-containing molecules. Cells vary in the amount of polybrene that they will tolerate, the concentration varying usually between 1 to 10 µg/ml. Increasing MOl from 1 to 50 were tested (a MOI of 1 corresponds to a virion/cell ratio of 1). The best results were obtained using a final concentration of 5 µg/ml of polybrene on low density primary keratinocytes, with a MOI of 20.

4.4.3 Re-expression of type VII coIlagen

Figure 2 shows immunostaining of type Vll collagen (in brown) in two colonies of RDEB keratinocytes, before (Fig. 2A) and after (Fig. 2B) transduction. The average percentage of corrected RDEB keratinocytes, expressing type Vll collagen is 85 %. Western blot analysis reveals that the type Vll collagen expressed by the transduced cell has the correct molecular weight (290 kDa) and is secreted in the cell culture medium (Fig. 3). Further biochemical analysis of the type Vll collagen expressed by the transgene, i.e. collagen helix stability and proteolysis resistance, will be performed in Dr Leena Brückner-Tuderman’s laboratory.

4.5. Grafting of genetically engineered RDEB skin onto nude mice