Detailed design and comparative analysis of protocols for optimized production of high-performance HIV-1-derived lentiviral particles

Abstract

Transgenic HIV-1-derived lentiviral particles are at the forefront of current gene therapy and tissue engineering initiatives, which will require optimal protocols for large-scale production of clinical-grade therapeutic lentiviruses. Production of latest-generation self-inactivating lentiviral particles requires cotransfection of mammalian production cell lines with two helper plasmids along with the lentivector, whose transgene-encoding expression cassette is the only genetic information stably transduced into target chromosomes. Capitalizing on a recently designed lentiviral expression vector family, we conducted rigorous analysis of production-relevant parameters including transfection, cell density, media composition, temperature, relative (helper) vector concentrations and genetic configuration. Comparative analysis of lentiviral particle performance (VP) was based on the viral titer (reflecting the number of transduction-competent lentiviral particles) relative to the number of lentiviral particles produced (correlating with p24 production levels) (VP=titer/viral particle number). Optimal lentiviral production parameters, resulting in up to 132-fold greater VP compared to standard protocols, required (i) CaPO₄-based transfection (ii) of helper plasmids and lentivector at a fixed concentration ratio (helper plasmid I:helper plasmid II:lentivector=1:1:2) (iii) into 1×10⁵ human embryonic kidney cells/cm² (HEK293-T) (iv) cultivated at 37 °C (v) in Advanced D-MEM medium supplemented with (vi) 2% fetal calf serum, (vii) and a culture additive containing 0.01 mM cholesterol, 0.01 mM egg's lecithin and 1× chemically defined lipid concentrate. (viii) Furthermore, constitutive transgene expression units placed in a forward polyadenylation site (pA)-free orientation relative to the lentivector backbone resulted in optimal transgene transduction/expression. Our studies suggest that detailed knowledge of lentivector design and the production of lentiviral particles will advance large-scale manufacturing of clinically relevant lentiviruses for future gene therapy applications.

Introduction

Owing to their efficient transduction of proliferating as well as mitotically inert cells in the absence of significant humoral immune responses, human immunodeficiency virus type 1 (HIV-1) lentiviral expression vectors have been designed for therapeutic interventions in (prototype) gene therapy scenarios (Crystal, 1995; Williams, 1995; Anderson, 2000; Somia and Verma, 2000; Kay et al., 2001; Nishikawa and Huang, 2001; Kang et al., 2002). Latest-generation HIV-1-based lentiviral transduction systems consist of multiply attenuated self-inactivating (SIN) viral genomes split among several (helper) plasmids harboring (i) gag (encoding major structural proteins), pol (coding for lentivirus-specific enzymes) and rev (a regulator of gag/pol expression and nuclear export of virus RNA), (ii) a vsv-g expression vector promoting pantropic transduction of pseudotyped lentiviral particles and (iii) the actual transgene(s)-encoding lentivector, which is the only genetic material transferred to the target cells (Dull et al., 1998). The lentivector harbors non-coding cis-acting elements, which manage encapsidation (extended packaging signal [${\psi }^{+}$], reverse transcription and integration (polypurine tracts [PPT, cPPT]; 5′ and 3′LTRs; Rev response element [RRE]). In order to prevent transcriptional interference with transgene expression and provirus-flanking chromosomal elements at the integration site, most lentivectors contain an enhancer-specific deletion in their 3′LTR, which renders the lentiviral particles SIN (Deglon et al., 2000). Standard lentiviral particle production requires cotransfection of two helper plasmids encoding a different pseudotyping envelope (Bartz and Vodicka, 1997; Mochizuki et al., 1998; Reiser et al., 2000; Beyer et al., 2002; Duisit et al., 2002; Kang et al., 2002) and complementing structural genes (De Palma and Naldini, 2002), along with the transgene-encoding lentivector, into human embryonic kidney cells (HEK293-T). Following transfection, lentiviral particles were produced and released into the culture supernatant, from which they were filtered for ready-to-use transduction batches (Naldini et al., 1996; Mitta et al., 2002; Koponen et al., 2003; Blesch, 2004). Stable helper cell lines, transgenic for constitutive complementation of structural genes and conditional expression of toxic vsv-g, have been considered as an alternative to triple transient transfections (Yu et al., 1996; Kaul et al., 1998; Kafri et al., 1999; Klages et al., 2000; Beyer et al., 2002). However, overall production performance did not always meet expectations (Rohll et al., 2002). Transient transfection thus remains the preferred technology for pilot production of lentiviral particles in a rapid, reliable and safe format (Naldini et al., 1996; Dull et al., 1998; Johnston et al., 1999; Morizono et al., 2001; D’Costa et al., 2003; Koponen et al., 2003; Mangeat et al., 2003; Mitta et al., 2004; Ventura et al., 2004). Although most established biopharmaceutical manufacturing scenarios use stable production cell lines (Zahn-Zabal et al., 2001), transient transfection is gathering momentum as a straightforward, consistent and cost-effective method for pilot production of biopharmaceuticals and recombinant viral particles (Zahn-Zabal et al., 2001; Derouazi et al., 2004; Maranga et al., 2004; Merten, 2004). There are a variety of protocols for the production of lentiviral particles which render a critical evaluation and comparative analysis from a bioengineering perspective impossible (Naldini et al., 1996; Zufferey et al., 1997; Mochizuki et al., 1998; Reiser et al., 2000; Zhao et al., 2002; Coleman et al., 2003; Condiotti et al., 2004). Furthermore, there is no generally accepted parameter describing overall virus performance. p24, the capsid protein encoded by gag (Oroszlan et al., 1979; Copeland et al., 1983; Coffin et al., 1997), is often quantified and provided as a measure correlating with virus titer. Yet, p24 levels fail to indicate the transduction quality of lentiviral particle preparations, since they only score the presence of this protein on the surface of lentiviral particles in a function-independent manner (2000 p24 proteins per lentiviral particle (Layne et al., 1992; Huang et al., 2001)). While virus titer generally correlates with transduction-competent lentiviral particles it fails to quantify the number of lentiviral particles generated during a production process.

We suggest that putting titer and lentiviral particles into relative perspective will enable a direct comparison of lentiviral particle production and transduction performance across platforms and protocols. Detailed multi-factor evaluation of different protocols for lentiviral particle production revealed critical parameters for optimal pilot production of high-quality recombinant lentiviral particles.