Setting up the system

The first step for the generation of retargeted lentiviral vector particles was their pseudotyping with the MV glycoproteins. As the unmodified MV H and F proteins did not pseudotype HIV-1 vectors (3.1), it was assumed that sequences in their cytoplasmic tails prevent pseudotyping. In both postulated mechanisms of pseudotype formation the glycoprotein cytoplasmic tail plays an important role. In the passive incorporation model, in which no direct interactions between virus proteins and glycoproteins are needed, but sufficient amounts of glycoproteins must be provided at the site of budding, the glycoprotein cytoplasmic tail has no active function but should not sterically hinder vector particle incorporation and morphology (Swanstrom and Wills, 1997). In the active model, it directly interacts with the viral core proteins or indirectly via cellular factors, which leads to pseudotyping.

At the beginning of this thesis there was only a single description of lentiviral pseudotype formation with glycoproteins of a paramyxovirus, namely Sendai virus, available. Thereby, truncation of the F protein cytoplasmic tail and addition of the cytoplasmic tail of the lentiviral Env to the cytoplasmic tail of the HN protein were necessary to allow pseudotyping (Kobayashi et al., 2003). For the generation of

HIV-1 vectors pseudotyped with the MV glycoproteins, in a first attempt, this strategy was adapted for this thesis. However, no MV-HIV vectors could be generated with this approach (data not shown). For this reason also the cytoplasmic tail of the H protein was truncated.

Fifteen H protein variants carrying stepwise truncations and amino acid exchanges in their cytoplasmic tails and two F protein variants (3.1.1 Figure 9) were screened in all combinations for their ability to pseudotype HIV-1 vector particles. The screen identified three combinations that led to efficient pseudotype formation, namely Hc∆18/Fc∆30, Hc∆19/Fc∆30 and Hc∆24+4A/Fc∆30 (3.1.1). The F cytoplasmic tail has to be truncated by 30 aa, leaving just 3 aa, two of which are positively charged.

In transmembrane proteins such membrane proximal, positively charged residues are often needed to stabilise their integration into the lipid-bilayer (Dalbey, 1990). The functionality of the Fc∆30 protein in respect of membrane fusion has been shown previously (Moll et al., 2002). Furthermore MV F proteins with a premature stop codon, resulting in a cytoplasmic tail truncated by 24 aa, are found in MV isolates from patients with SSPE (Cathomen et al., 1998).

Among the H protein variants, there was a clear peak of optimal truncation when 18 or 19 residues were deleted (variants Hc∆18 and Hc∆19). Further truncation reduced titers, although replacing some of the deleted residues by alanine could restore optimal titers in case of variant Hc∆24+4A. Obviously, the critical step was the identification of H protein cytoplasmic tail truncation mutants that allowed pseudotyping while retaining fusion helper function located in the membrane proximal region of the H protein cytoplasmic tail (Moll et al., 2002). All of the three H protein variants that allowed most efficient pseudotype formation are still active in fusion helper function (Moll et al., 2002). Apparently, the cytoplasmic tails of the unmodified MV H and F proteins contain sequences that prevent pseudotyping of lentiviral vectors. Western blot analysis of Hc∆18/Fc∆30 pseudotyped HIV-1 vector particles in comparison to HIV-1 particles produced in the presence of the unmodified H and F proteins demonstrated that these perturbing sequences in the cytoplasmic tails of H and especially F prevent incorporation of the unmodified MV glycoproteins into viral particles (3.1.3).

Although the overall expression levels of the truncated proteins in the packaging cells appeared increased compared to the unmodified proteins, cell surface expression levels, which are regarded more relevant for pseudotyping did not differ (3.1.3). Thus,

the interference of cytoplasmic tail sequences with particle incorporation is most likely the reason for absence of pseudotyping with the unmodified H and F proteins.

Interestingly, these same variants as well as other tested MV glycoprotein variants could not efficiently pseudotype MLV particles (3.1.2). Obviously, the restriction factors for the formation of lentiviral and γ-retroviral MV pseudotypes, respectively, are different. There is indeed evidence in literature for a different behaviour of MLV and HIV-1 vector particles in respect of pseudotype formation. For example, the unmodified glycoproteins of GALV and RD114 virus readily form MLV pseudotypes, because the MLV protease can cleave their R peptide, located in their cytoplasmic tails, but are unable to pseudotype HIV-1 particles. Only after truncation or modification of the R peptide sequence, these envelope proteins form HIV-1 pseudotypes (Christodoulopoulos and Cannon, 2001; Merten et al., 2005; Sandrin et al., 2004; Stitz et al., 2000).

A few month after our publication of the MV-HIV pseudotype formation (Funke et al., 2008), also the group around François-Loïc Cosset published data demonstrating pseudotyping of HIV-1 vectors with MV glycoproteins at titers comparable to ours (Frecha et al., 2008). While they also used Fc∆30, they identified Hc∆24 as the best candidate for pseudotyping. Pseudotype formation with this H protein truncation mutant was very inefficient in the screen described in this thesis (3.1.1). Furthermore, Hc∆24 is strongly impaired in the fusion helper function (Moll et al., 2002), for which reason it is even more surprising that it allowed efficient pseudotype formation.

Another discrepancy is the observation by Frecha et al. that Hc∆24/Fc∆30 also pseudotype MLV particles (Frecha et al., 2008). As mentioned above, in this thesis, for MV-MLV vector particle pseudotyping no gene transfer above background levels was observed, also including the Hc∆24/Fc∆30 combination (data not shown). Taken together, Frecha et al. confirmed the data in this thesis that MV-HIV pseudotypes are formed upon cytoplasmic tail truncations. In contrast to our data, a fusion helper function deficient H protein mutant was active and MV-MLV pseudotypes could be generated. Both issues are not in accordance with our results and can not be explained at the moment.

An interesting observation was the increase in MV-HIV titers, when the ratio of the plasmids encoding Fc∆30 and Hc∆19, respectively, was optimised for vector particle production. The optimal ratio was determined to be seven-fold more F than H plasmid, which resulted in a more than ten-fold increase in titer compared to particles

produced in presence of the same amounts of pCG-Fc∆30 and pCG-Hc∆19 (3.1.6).

Although it was known before that more F than H mRNA is present in MV infected cells (Cattaneo et al., 1987; Plumet et al., 2005), the extent of titer reduction observed when suboptimal H:F ratios were applied, was unexpected. Obviously, a fine tuned balance of the relative amounts of F and H has to be maintained in the packaging cells to allow most efficient formation of pseudotyped HIV-1 vector particles.

In conclusion, the successful pseudotyping of HIV-1 vector particles with modified MV glycoproteins was demonstrated. These particles mediated efficient and stable gene transfer into MV receptor-positive but not MV receptor-negative cell lines. This paved the way for the generation of retargeted HIV-1 vector particles. For proof of principle, the EGF receptor (EGFR) and the B cell surface marker CD20 (Cragg et al., 2005) were chosen as targets.