Lentiviral vectors in gene therapy: their current status and future potential

Summary

The concept of gene therapy originated in the mid 20^th century and was perceived as a revolutionary technology with the promise to cure almost any disease of which the molecular basis was understood. Since then, several gene vectors have been developed, and the feasibility of gene therapy shown in many animal models of human disease. However, clinical efficacy could not be demonstrated until the beginning of the new century in a small-scale clinical trial curing an otherwise fatal immunodeficiency disorder in children. This first success – achieved after retroviral therapy - was later on overshadowed by the occurrence of vector-related leukaemia in a significant number of the treated children, demonstrating that the future success of gene therapy depends on our understanding of vector biology. This has led to the development of later generation vectors with improved efficiency, specificity and safety. Amongst these are HIV-1 lentivirus-based vectors (lentivectors), which are being increasingly used in basic and applied research. Human gene therapy clinical trials are currently under way using lentivectors in a wide range of human diseases. The intention of this review is to describe the main scientific steps leading to the engineering of HIV-1 lentiviral vectors, and place them in the context of current human gene therapy.

1. A brief history of gene therapy

Gene therapy can be broadly defined as the treatment of a disease or medical disorder by the introduction of therapeutic genes into the appropriate cellular targets. These therapeutic genes can correct deleterious consequences of specific gene mutations, or re-programme cell functions to overcome a disease. For successful gene therapy, the exogenous therapeutic gene has to be specifically, efficiently and stably incorporated into the target cell.

The concept of gene therapy is not at all novel or recent, and it is accompanied by controversy since the moment it was proposed. Therefore, it is appropriate to approach this subject with a brief historical perspective. As it is the case with many scientific revolutions, everything started with a simple question. How are characteristics passed from generation to generation? It cannot be argued that the work carried out by Gregor Mendel during the 1850s to early 1860s, together with the later work of Ronald Fisher during the early 20^th century, represent landmarks in genetics (Weiling, 1991). Their work confirmed that organisms contained “encoded information” in the form of some kind of discrete biological material, termed “gene” in 1909 by the Danish botanist Wilhelm Johannsen (Falk, 1984). During the early 20^th century, it was clear that many human medical conditions, such as haemophilia, were transmitted from parents to offspring. There was also evidence that some diseases such as diabetes, retinoblastoma and colon cancer presented some underlying genetic contributions. Correction of these defects posed a medical challenge because it was not possible to directly manipulate genes (Keeler, 1947). Several breakthroughs in genetics, biochemistry and molecular biology from the 1940s to 1980s dramatically changed that view (Editorial, 1976). In a relatively short period of time, DNA was identified as the genetic material (Avery et al., 1979), its structure solved (Watson and Crick, 1953a; Watson and Crick, 1953b), the genetic code was broken (Ochoa, 1963) and gene cloning was becoming routine (Cline, 1985; Editorial, 1981; Hamer et al., 1979; Mantei et al., 1979; Mulligan et al., 1979; Nagata et al., 1980).

By the late 1970s, the molecular bases of many genetic disorders were well understood and gene therapy provided the “ideal and clean” solution (Friedmann, 1976). Restore the gene, cure the disease. Gene therapy was no longer considered an alternative treatment for genetic human conditions, but a potential solution (Friedmann, 1976; Friedmann and Roblin, 1972). In fact, the first human gene therapy trial was carried out in the early 70s. Three hyperargininaemic patients were intravenously inoculated with Shope papillomavirus, so that the virus-encoded arginase could correct the disease. Actually, intravenous administration in rabbits was asymptomatic and therapeutic. However, no reduction on blood arginine levels was observed in treated patients, and the authors attributed the failure to the virus instability (Friedmann and Roblin, 1972; Terheggen et al., 1975). In any case, the rapid scientific development suddenly placed human gene therapy from an unrealistic scenario to a near-future possibility, raising concerns within the scientific and non-scientific community. Some of these issues are still contemporaneous, such as human cloning and “designer babies” (Neville, 1976), having been re-raised by the more recent advances in somatic cloning, stem cell biology and regenerative medicine.

By the middle 1980s, gene transfer to mammalian cells was routinely performed. In fact, retrovirus-based gene transfer presented major advantages, due to stable integration of their genome in the host cell chromosomes. However, gene therapy was still controversial largely due to the poorly understood field of gene regulation, concerns about the possible influences of exogenous DNA in the host cell and other major ethical issues (Editorial, 1981; Williamson, 1982). Some of these ethical issues were materialized after the controversial, unauthorized human gene therapy trial carried out in 1980 in two β-thalassemia patients (Cline, 1985; Mercola and Cline, 1980).

The proof-of-principle of γ-retrovirus gene transfer into hematopoietic stem cells was demonstrated in the early 90s (Brenner et al., 1993; Deisseroth et al., 1994; Dunbar et al., 1995; Rill et al., 1994), and the first approved clinical trial to correct severe combined immunodeficiency (SCID) was carried out by Anderson and colleagues in 1991 (Anderson et al., 1990; Blaese et al., 1993; Levine and Friedmann, 1991). Peripheral blood CD34⁺ cells from patients were transduced with a γ-retrovirus driving the expression of adenosine deaminase. Although the direct benefits of this gene therapy trial are still under deliberation, this can be considered the first “successful” human gene therapy trial, at least regarding safety issues. A major breakthrough in gene therapy came in 2000, when X-SCID was corrected in 11 children by the introduction of the common interleukin receptor γ-chain in bone marrow using a retrovirus vector based on mouse leukaemia virus (MLV) (Cavazzana-Calvo et al., 2000). A similar approach was later published by Adrian Thrasher’s group in London (Gaspar et al., 2004).

Regrettably, in both clinical trials, several cases of leukaemia directly associated to the gene therapy itself were later reported, highlighting retrovirus induced insertional mutagenesis as a major complication. Interestingly, this problem had already been taken into consideration from a theoretical point of view in the early 1980s (Cline, 1985; Hacein-Bey-Abina et al., 2003; Howe et al., 2008). Following a similar approach to the X-SCID clinical trials, in 2006 the correction of X-linked chronic granulomatosis was reported, again using γ-retrovirus vectors encoding gp91 phox in bone marrow (Gottlieb, 2006; Moreno-Carranza et al., 2009). Clinical efficacy was observed, together with clonal amplification of corrected cells, probably as a result of insertional activation of MDS1-EVI1, PRDM16 and SETBP1 (Ryser et al., 2007).

In 2006, the first successful gene therapy trial for the treatment of melanoma was carried out. In this, a retrovirus encoding the α- and β-chains of a melanoma antigen (MART-1)-specific T cell receptor (TCR) was engineered. This TCR was cloned from a patient that showed regression after adoptive transfer therapy. Peripheral blood lymphocytes were transduced with this retrovirus and re-administered, leading to complete regression in two out of fifteen patients (Morgan et al., 2006). This was the first demonstration that immune cells could be genetically modified to fight cancer in a human gene therapy trial.

Nowadays, we are re-living exciting times. During the last decade immense scientific leaps have been made, from somatic cloning (Wilmut et al., 1997) and completion of the human genome project (Venter et al., 2001) to the discovery of microRNA-based gene regulatory systems (Lombardo et al., 2007). It is in this context that lentivectors have been developed with the promise of becoming the substitutes of retroviral vectors, the first gene carriers leading to long-term, full correction of human genetic diseases.

2. Retrovirus biology and basic engineering of lentivectors

Viruses are intracellular obligate parasites and they have evolved as efficient vehicles for the delivery of DNA or RNA to target cells. A large number of viruses with unique characteristics useful for gene therapy have been identified. This has led to the application of recombinant viruses such as adenoviruses, adeno-associated viruses, herpes viruses, poxviruses, retroviruses and more recently lentiviruses, both in the laboratory and clinic. Vectors based on retroviruses such as MLV were amongst the first to be developed (Mann et al., 1983) and to be “successfully” used in gene therapy trials (Anderson et al., 1990; Blaese et al., 1993; Levine and Friedmann, 1991). Currently, these vectors are used in about 21.2% of clinical trials (http://www.wiley.co.uk/genmed/clinical/[July 2009]).

In recent years, research has focused on the use of lentivectors, which - like their simple retrovirus counterparts - are devoid of viral proteins, free from replication competent virus, and additionally able to transduce non-dividing cells (Bukrinsky et al., 1993). This characteristic is advantageous in many gene therapeutic applications targeting post-mitotic, highly differentiated cells. Currently, these lentivectors are applied in about 1.4% of clinical trials (http://www.wiley.co.uk/genmed/clinical/[July 2009]).

2.1. The retroviral particle

Retroviridae is a family of single-stranded (ss) RNA spherical viruses of around 80 to 120 nm diameter (Vogt and Simon, 1999). The retroviral particle contains two copies of positive strand RNA, which are - together with the enzymes reverse transcriptase, integrase and protease - complexed with nucleocapsid protein. A second protein shell formed by capsid protein, encloses the nucleocapsid and delimits the viral core (Jones and Morikawa, 1998). Matrix proteins form a layer outside the core and interact with a cellular-derived lipid envelope, which incorporates viral envelope glycoproteins (env), responsible for the interaction with specific cellular receptors. Two units form these glycoproteins: TM (transmembrane), that anchors the protein into the lipid bilayer and SU (surface), which binds to the cellular receptor (Figure 1).

Figure 1.

Basic engineering of lentivectors. The self-inactivating (SIN) lentiviral transfer vector (a) and the resulting lentivector (b). The SIN vector contains a modified U3 region within the 3′ LTR, in which the enhancer/promoter sequences have been deleted as shown in the transfer vector. Abbreviations: CA, capsid; CTS, central termination site; IN, integrase; LTR, long terminal repeat; MA, matrix; NC, nucleocapsid; PPT, polypurine tract; PR, protease; RT, reverse transcriptase; SU, surface; TM, transmembrane; TRIP: triple helix; WPRE: woodchuck hepatitis B posttranscriptional element.

2.2. The retroviral genome

Based on their genome organization, the Retroviridae are divided in simple and complex retroviruses. Examples are oncoretroviruses, such as MLV and lentiviruses, such as HIV-1, respectively. Lentiviruses include primate retroviruses and non-primate retroviruses. Examples hereof are HIV and SIV (simian immunodeficiency virus) for the former and FIV (feline immunodeficiency virus), BIV (bovine immunodeficiency virus), CAEV (caprine arthritis-encephalitis virus), EIAV (equine infectious anemia virus) and visnavirus for the latter. Their genome is organized in the gag, pol and env gene. Gag encodes the structural proteins, whereas the pol gene encodes the enzymes that accompany the ssRNA. Of these, the reverse transcriptase carries out reverse transcription of the viral RNA to DNA, integrase catalyses the integration of the proviral DNA into the host genome and protease is involved in gag-pol cleavage and virion maturation (Katz and Skalka, 1994). Env encodes the viral envelope. In addition, complex retroviruses have accessory genes, whose concerted action regulates viral gene expression, assembly and replication (Coil and Miller, 2004). Moreover, the retroviral genome contains cis-acting sequences such as two long terminal repeats (LTR), with elements required for gene expression, reverse transcription and integration into the host chromosomes. Other important components are the packaging signal (psi or ψ), required for the specific RNA packaging into newly formed virions (Watanabe and Temin, 1982) and the polypurine tract (PPT), which is the site of the initiation of the positive strand DNA synthesis during reverse transcription (Charneau et al., 1992; Rattray and Champoux, 1989).

2.3. The retroviral life cycle

The retroviral life cycle can be broken down in several steps, starting with the binding of the viral envelope to its receptor (i), and fusion of the viral envelope with the cell membrane (ii). Subsequently, the particle is uncoated and the viral core released into the cytoplasm (iii). The ssRNA is reverse transcribed into dsDNA within the core (iv), which is transported to the nucleus (v) upon cell division for oncoretroviruses (Lewis and Emerman, 1994; Ryser et al., 2007) or through active transport in the case of lentiviruses (Bukrinsky et al., 1993). Herein lies the major advantage of lentiviruses over oncoretroviruses, their ability to transduce non-dividing cells. Once in the nucleus, the viral DNA is integrated into the host DNA (provirus) (vi), resulting in long-term expression of viral genes, which are transcribed (vii) and spliced (viii) during the life of the infected cell. The full-length viral RNA, as well as the RNA encoding all viral proteins are transported to the cytoplasm (ix), where they are translated (x). The unspliced full-length viral RNA is packaged and a viral particle assembled (xi). Virion maturation occurs together with budding of the particle from the cell (xii), as such resulting in infectious virions (Palu et al., 2000).

2.4. Basic engineering of retrovirus-based vectors

The use of gene delivery vectors based on retroviruses was introduced in the early 1980s (Mann et al., 1983). The most commonly used retroviral vectors are based on Moloney MLV and have a simple genome. From this genome, the polyproteins, gag, pol and env are required in trans for viral replication and packaging. Required in cis are the 5′ and 3′ LTR, the sequences for packaging of the viral RNA, as well as the tRNA-binding site and sequences involved in reverse transcription and integration of the provirus. The gag, pol and env genes are replaced with an expression cassette, containing the gene of interest, which is under the control of a chosen promoter (see sections 3.2.2.1. to 3.2.2.2.).

The major advantages of oncoretrovirus-based vectors are: (i) their lack of viral proteins, which renders them replication deficient and less immunogenic, in the sense that they do not elicit anti-vector immune responses (see section 5.1.) and (ii) their ability to integrate into the host genome, leading to persistent gene expression. However, there are some important limitations, such as: (i) instability of the viral particle (Andreadis et al., 1997), (ii) low viral titers (Le Doux et al., 1999), (iii) inability to transduce non-dividing cells (Lewis and Emerman, 1994; Ryser et al., 2007) and (iv) insertional mutagenesis. To overcome these shortcomings, vectors based on lentiviruses were developed (Naldini et al., 1996b). Lentivectors can also induce insertional mutagenesis, however at lower frequencies than their γ-retroviral counterparts, possibly due to different integration patterns (Hematti et al., 2004; Modlich et al., 2009).

Lentivectors are capable of transducing quiescent cells (Lewis and Emerman, 1994). For HIV, this process is facilitated by (i) the integrase protein (Naldini et al., 1996b), (ii) the matrix protein (Naldini et al., 1996b), (iii) vpr (Heinzinger et al., 1994) and (iv) the PPT (VandenDriessche et al., 2002). As a result of their capacity to transduce quiescent cells - although for some cell types, progression through the cell cycle is necessary (Breckpot et al., 2004; Korin and Zack, 1998) - lentivector development has received much interest. As with oncoretrovirus vectors, the design of lentivectors is based on the separation of cis- and trans-acting sequences.

3. Development of lentivectors (Figure 2)

Figure 2.

Characteristics of lentiviruses and development of improved recombinant lentivectors.

In general, lentivector particles are generated by co-transfection of 3 plasmids in human embryonic kidney (HEK) 293T cells (Naldini et al., 1996a): (i) a packaging plasmid, (ii) a transfer plasmid (Figure 1) and (iii) an envelope-encoding plasmid. In the latest generation vectors, the rev and tat genes are not longer included in the packaging plasmid, but provided in a fourth plasmid. Development of lentivector systems is reviewed elsewhere (Breckpot et al., 2007a; Breckpot et al., 2008).

3.1. The different generations of lentivectors

The lentiviral particles are divided into “generations” according to the packaging plasmid used for production. The first generation packaging plasmid provided all gag and pol sequences, the viral regulatory genes tat and rev and the accessory genes vif, vpr, vpu and nef. Identification of the HIV genes disposable for transfer of the genetic cargo allowed the engineering of the multiple-attenuated second generation packaging systems (Gruber et al., 2000; Zufferey et al., 1997). In these, the four accessory genes, vif, vpr, vpu and nef, were removed without negative effects on vector yield or infection efficiency, and as such improving lentivector safety, since any replication-competent lentivirus would be devoid of all virulence factors. Safety was further improved in the third generation packaging systems, consisting of a split-genome packaging system in which the rev gene is expressed from a separate plasmid and the 5′LTR from the transfer vector replaced by a strong tat-independent constitutive promoter (Dull et al., 1998).

3.2. Improving vector performance and safety

3.2.1. Modification of the packaging plasmid: integrase deficient lentivectors

Long-term, stable transgene expression, as a result of lentivector integration is very useful for diseases in which permanent cell correction is sought after. However, as already mentioned before, insertion of the γ-retrovirus led to oncogene transactivation and leukaemia in some of the X-SCID children treated by gene therapy. Therefore, the development of non-integrating lentivectors (NILV) has attracted much attention. Accordingly, selected mutations within the integrase-coding region of the packaging plasmid have been exploited to generate integrase-deficient lentivectors. These mutations eliminate integrase activity without affecting reverse transcription and transport of the pre-integration complex to the nucleus. Then, the lentivector DNA remains in the nucleus as an episome, leading to sustained expression in post-mitotic cells and tissues such as retina, brain and muscle (Apolonia et al., 2007; Philippe et al., 2006; Yanez-Munoz et al., 2006). NILV with multiple mutations either in the integrase itself or in the integrase attachment sites have been shown to be as effective as standard lentivectors, in a lymphoma model when dendritic cell (DC) constitutive activators were co-expressed with a surrogate tumour antigen (Karwacz et al., 2009). NILV could be applied in the case that post-mitotic tissue is the target, or in applications that do not require persistent antigen expression, such as vaccination. However, background integration by recombination can still occur (Apolonia et al., 2007). Recently, NILV have been used as donor DNA sequences for the activity of zinc-finger nucleases (Brown et al., 2007a; Cornu and Cathomen, 2007). This has allowed the targeted correction of gene mutations even in human stem cells, circumventing the drawback of random integration and insertional mutagenesis (Brown et al., 2007a).

3.2.2. Modification of the transfer plasmid (Figure 1)

The transfer plasmid consists of an expression cassette and the HIV cis-acting factors necessary for packaging, reverse transcription and integration, and can be modified to augment lentivector performance and safety. Nuclear import of the transfer construct was improved by including the PPT and its central termination sequence, together forming a triple helix (TRIP), resulting in higher vector titers and enhanced transgene expression (Sirven et al., 2000). Addition of other elements such as the woodchuck hepatitis B posttranscriptional regulatory element (WPRE), improves the lentiviral vector design by increasing gene expression by modification of polyadenylation, RNA export or translation (Zufferey et al., 1999). Engineering of self-inactivating (SIN) transfer vectors by deleting the enhancer/promoter unit in the U3 region of the 3′LTR (Miyoshi et al., 1998; Zufferey et al., 1998) minimized the risk of replication competent lentiviruses and decreased promoter interference (Ginn et al., 2003), which brings us to the topic of promoter engineering, another strategy to improve both performance and safety.

3.2.2.1. Cell- and tissue-specific promoters

Cell-specific promoters are advantageous since they are less sensitive to promoter inactivation and less likely to activate the host cell defence machinery (Liu et al., 2004). As such, improved stability and longevity of gene expression can be expected. Therefore, it is not surprising that much effort has been put in research on this topic. To date, specific gene expression has been described for several cell types, including erythroid (Moreau-Gaudry et al., 2001), endothelial (De Palma et al., 2003), central nervous system (Gascon et al., 2008; Greenberg et al., 2007; Kuroda et al., 2008; Lai and Brady, 2002; Liu et al., 2008), retinal (Miyoshi et al., 1997; Semple-Rowland et al., 2007), liver (Oertel et al., 2003; VandenDriessche et al., 2002) and cancer cells (Morgan et al., 2006; Seo et al., 2009; Uch et al., 2003; Yu et al., 2001). In addition, gene-specific expression has been achieved for immune cells, in particular DC that can contribute to induction (anti-tumour gene therapy) or suppression (auto-immune and graft-versus-host disease) of immune responses (Cui et al., 2002; Lopes et al., 2008; Zhang et al., 2009).

As an example, the cellular diversity within the central nervous system underscores the importance of restricting transgene expression to target cells (Costantini et al., 2000). Several promoters have been tested to drive gene expression in neurons, cells of the hypocampus and glial cells. Amongst these are the enolase promoter (Lai and Brady, 2002), SYN promoter (Gascon et al., 2008; Liu et al., 2008), synapsin 1 promoter (Kuroda et al., 2008), CD44 promoter, glial fibrillary acidic protein promoter and vimentin promoter (Greenberg et al., 2007; Liu et al., 2008). Some of these have been shown to drive high transgene expression (Kuroda et al., 2008; Lai and Brady, 2002), e.g. the CD44 promoter, which allows high and sustained GFP expression in Muller glia cells for over 6 months. In contrast, the glial fibrillary acidic protein promoter and vimentin promoter were less efficient (Greenberg et al., 2007), demonstrating the need for the testing of candidate promoters and strategies to enhance their activity, such as transcriptional activation in which activity is enhanced by the use of two promoter copies, or by artificial transcriptional activators. In this regard, Liu et al demonstrated an increase in transgene expression driven by a modified SYN promoter or glial fibrillary acidic protein promoter in neurons and glia cells, respectively (Liu et al., 2008).

Several promoters have been tested in view of curing retinal diseases. Amongst these is the rhodopsin promoter, which results in high levels of photoreceptor-specific expression (Miyoshi et al., 1997). In another study, two photoreceptor promoters, the interphotoreceptor retinoid binding protein and guanylate cyclase activating protein promoters were evaluated alongside the rhodopsin promoter to express two proteins in a retinal cell-specific manner (Semple-Rowland et al., 2007). The ability to engineer one lentivector that targets expression of multiple genes to single (cone cells) or multiple cells (cone cells and rod cells) in vivo was demonstrated. This type of application should prove useful in the development and delivery of complex, combination therapies.

Another proof of the benefits of cell-specific gene expression was shown in liver using the albumin promoter, leading to long-term transgene expression in rat liver, in contrast to CMV that was rapidly silenced (Oertel et al., 2003; VandenDriessche et al., 2002).

For anti-cancer gene therapy, envisaging the delivery of cell death inducing genes, it is important to restrict transgene expression to cancer cells. Human hepatocarcinoma cells have been transduced with lentivectors containing a suicide gene under the control of the α-fetoprotein promoter resulting in specific destruction of these malignant cells (Uch et al., 2003). Furthermore, a patient-derived prostate-specific antigen (PSA) promoter inserted into a lentivector has driven efficient activity in prostate cells with satisfactory efficacy and specificity (Yu et al., 2001). More recently, a PSA promoter-based lentivector has been used to deliver the diphtheria toxin A gene into prostate cancer cells, resulting in specific eradication of cancer cells in cell culture and in a mouse tumor model (Morgan et al., 2006). Another example is the metalloproteinase (MMP2) promoter which has been applied to introduce the pro-apoptotic genes Bax and tBID, resulting in cell death in MMP2 expressing cancer cell lines, but not healthy MMP2⁺ cells (Seo et al., 2009).

For “gene immunotherapy”, a specialized form of gene therapy, DC constitute an interesting target. The development and function of DC, as well as DC-targeting by lentivectors is reviewed in Breckpot et al (Breckpot et al., 2007a; Breckpot et al., 2008). Briefly, DC are specialized antigen-presenting cells involved in modulating immune responses, and can be exploited to fight cancer and infectious diseases or to re-establish tolerance in view of auto-immunity and transplantation. To achieve antigen-presenting cell-specific gene expression, Cui et al exploited the selective and high expression of MHC class II on antigen-presenting cells (Cui et al., 2002). Using the non-obese diabetic (NOD)/SCID mouse engraftment model, they showed specific expression in MHC class II⁺ human cells upon transduction with HLA-DRα promoter harbouring lentivectors. Other promoters, the dectin-2 and CD11c promoter have been used, in view of anti-tumour immunotherapy (Lopes et al., 2008) and for the generation of DC-specific transgenic mice (Zhang et al., 2009), respectively.

Taken together, there is a growing list of cell-specific promoters that are being applied successfully and that improve both safety and performance of lentivectors as gene transfer vehicles.

3.2.2.2. Regulatable promoters

The potential to regulate transgene expression is appealing for many gene therapy applications, as in the case of genetic diseases in which the gene expression levels, or timing of expression would be desirable, for example, diabetes. Therefore, several groups have focused their research on the use of inducible promoters. Many variations of each system have been developed, and the basic systems are briefly explained.

Probably, the tetracycline-based induction system is among the most widely used (Blomer et al., 1997; Gascon et al., 2008; Reiser et al., 2000; Seo et al., 2009; Vigna et al., 2002). For gene-therapy, an inducible vector dependent on the delivery of the antibiotic (tet-on) is preferred over one based on gene silencing (tet-off), which necessitates constant administration of antibiotics unless transgene expression is required. With the exception of some reports (Blomer et al., 1997; Gascon et al., 2008; Johansen et al., 2002), all the tetracycline-inducible lentivectors listed in the literature were constructed using the tet-on system (Farson et al., 2001; Georgievska et al., 2004; Reiser et al., 2000; Seo et al., 2009; Vigna et al., 2002). The authors showed this system in lentivectors allows dose-dependent rapid inducible expression. Other inducible systems have also been adapted to lentivectors, such as the drosophila ecdysone receptor (Galimi et al., 2005), that is based on: (i) a chimeric protein made of the herpes simplex virus protein VP16 activation domain and an ecdysone receptor (VgEcR), (ii) the retinoid X receptor (RXR) and (iii), the inducible promoter. In this system, ecdysone (or synthetic analogs) binds to the VgEcR-RXR heterodimer, which then binds to the inducible promoter driving gene transcription (Saez et al., 2000). As a drawback, multiple lentivector components have to be simultaneous administrated (Galimi et al., 2005). This multicomponent problem can be overcome by appropriate vector design, as with the tet-on/tet-off system in a single lentivector backbone using a fusion protein between the tetracycline repressor with the Kruppel-associated Box (KRAB) domain repressor as the regulator (Szulc et al., 2006).

This system has been used for tightly-regulated conditional transgene expression in the brain, gene silencing in hematopoietic cells and for the generation of drug-inducible transgenic mice (Szulc et al., 2006). Other systems adapted to lentivectors include glucocorticoid inducible promoters (Parker et al., 2009) and mifepristone-inducible systems (Sirin and Park, 2003).

3.2.3. Alteration of the viral envelope: targeting of lentivectors to specific cell types

Lentivector gene delivery requires entry into the cell of interest. The tropism of lentivectors is determined by their viral envelope glycoprotein, which upon interaction with its receptor trigger fusion of the viral envelope with the cell membrane. Since wild type HIV glycoproteins have restricted tropism, and do not allow production of high titer lentivector preparations, heterologous glycoproteins are used for lentivector production. This proces is termed “pseudotyping”. Lentivectors are often pseudotyped with the envelope of vesicular stomatitis virus (VSV G), a glycoprotein which interacts with a ubiquitous receptor, or a phospholipid component of the cell membrane (Coil and Miller, 2004). VSV G endows a broad host-cell range and confers high vector particle stability (Burns et al., 1993), two attractive properties in terms of ex vivo gene modification. However, restricting infection to specific cells, known as “transductional targeting”, is critical when it comes to efficient and safe in vivo gene delivery, and is key to enhance therapeutic effects, reduce side effects and possibly lower the amounts of vectors required. To achieve this goal, 2 methods can be used: (i) to take advantage of the natural properties of existing viral proteins and (ii) to use genetic engineering to retain, abolish or extend the original tropism of vectors.

There is an ever-growing list of glycoproteins that have been successfully used for pseudotyping of lentivectors, each with its (dis)advantages. Examples are glycoproteins from retroviridae, rhabdoviridae, arenaviridae, flaviviridae, paramyxoviridae, baculoviridae, and filoviridae (reviewed by Bouard et al (Bouard et al., 2009)). Although each of these glycoproteins preferentially interact with specific cell types, and can thus be used to restrict transgene expression, finding a natural envelope for cell-specific targeting may be challenging.

The modification of the viral surface by genetic engineering is an alternative to pseudotyping with existing envelopes. The aim is to alter the receptor attachment function in the glycoprotein, without hampering membrane fusion. For some viral envelope proteins such as that of Sindbis virus, which use more than one receptor for cell entry, this task was relatively straight-forward. Mutation of the most important contact residues prevented binding to its ubiquitously expressed receptor (heparane sulphate) resulting in an envelope protein variant that only bound to DC-SIGN. This is a DC-specific surface molecule, thus allowing DC-specific gene transfer (Morgan et al., 2006). However, engineering of re-targeted envelope proteins by fusion to natural ligands has proven to be difficult, and these strategies were atfirst applied with limited success (Waehler et al., 2007). Although binding of the vector to target cells was achieved, the inclusion of the ligand inhibited viral entry, except in one study. In that case the pH-dependent glycoprotein from Influenza virus was fused with epidermal growth factor (EGF), demonstrating specific infection of EGF receptor⁺ target cells (Hatziioannou et al., 1999). Alternative strategies have been developed based on specific requirements, such as the expression of specific proteases on the surface of target cells that can release the vector from the bound receptor (Szecsi et al., 2006). Such protease-activatable vectors were applied for the targeted infection of tumour cells that express MMP (Duerner et al., 2008; Springfeld et al., 2006).

Recently described targeted lentivectors exploit envelope proteins such as those of Sindbis virus and measles virus, in which the binding and fusion functions are provided by separate proteins. The Sindbis virus envelope consists of the E1 and E2 protein, which harbour the fusion peptide and receptor binding sites, respectively. Together, both proteins mediate pH-dependent cell entry. The targeting strategy involves mutation of the natural receptor binding sites of E2 and provision of an alternative binding method. Thus far, several studies have been published targeting tumour (Morizono et al., 2005; Pariente et al., 2007), endothelial (Pariente et al., 2008) and B cells (Morgan et al., 2006). Targeting is achieved by either covalent conjugation of a cell-specific antibody (Morizono et al., 2005; Pariente et al., 2008; Pariente et al., 2007), or its incorporation in the viral envelope alongside the engineered Sindbis virus glycoprotein (Morgan et al., 2006). However, this strategy requires endocytosis for the pH-dependent membrane fusion. In contrast, the measles virus proteins F and H mediate pH-independent membrane fusion, and thus allow direct entry at the cell membrane level. Similar to the approach for the Sindbis virus envelope, binding residues present in H were mutated and target ligands, such as EGF and a single chain antibodies against CD20 were fused to the mutated H protein. These lentivectors efficiently and specifically transduced EGF receptor⁺ and CD20⁺ cells, respectively (Funke et al., 2008), demonstratingthe feasibility of targeting to specific cell types. Moreover, transductional targeting has been combined with cell-specific promoters (transcriptional targeting) (Pariente et al., 2008; Pariente et al., 2007), as such further impacting on lentivector safety and performance.

4. Application of lentivectors in gene therapy

Insertional mutagenesis has been a major drawback for the use of γ-retroviral vectors in human gene therapy. Lentivectors also exhibit insertional mutagenesis in animal models and cellular-based systems prone to oncogenesis (Bokhoven et al., 2009), although overall they seem to be less mutagenic than their oncoretrovirus counterparts (Hematti et al., 2004; Montini et al., 2009). Actually, in the first approved human clinical trial using lentivectors for the treatment of HIV, no abnormal cell expansion or enrichment of integration sites near proto-oncogenes have been detected so far (Manilla et al., 2005; Themis et al., 2005).

Lentivectors have been tested for some time in many gene therapy animal model for metabolic diseases, such as β-talassemia (Zhao et al., 2009), SCID (Mortellaro et al., 2006; Throm et al., 2009), Wiskott-Aldrich syndrome (Mantovani et al., 2009; Marangoni et al., 2009), haemophilia (Brown et al., 2007a), metachromatic leukodystrophy (Biffi and Naldini, 2007), Fanconi anaemia (Jacome et al., 2009) and liver diseases in non-human primates (Menzel et al., 2009). The list of gene therapy models in which lentivectors are being used is increasing, and hopefully, the last lentivector generation will exhibit the safety and therapeutic efficacy that we have been waiting for.

5. Gene therapy of the immune system

The proof-of-principle of the therapeutic efficacy of gene therapy targeting the immune system has been established by the “successful” clinical trials for the treatment of melanoma and X-SCID (Cavazzana-Calvo et al., 2000; Gottlieb, 2006; Moreno-Carranza et al., 2009). At the present time, lentivectors are being tested for gene therapy of the immune system, and the main strategies are briefly summarized below.

5.1. Cancer immunotherapy

A very promising approach for “gene immunotherapy” is specific lentivector targeting to DC, the professional antigen-presenting cells of the immune system that regulate both immunity and tolerance. For cancer immunotherapy, the major goal is to break tolerance against tumour-associated antigens, since they are mostly either self- or quasi-self antigens. To drive an efficient anti-tumour immune response, antigens have to be presented to specific CD4 and CD8 T cells by mature immunogenic DC, which are characterized by high expression levels of co-stimulatory molecules, and pro-inflammatory cytokines. To meet these requirements, DC-specific transgene expression (section 3.2.2.1. and 3.3.) and induction of efficient DC maturation have to be achieved (Breckpot K, 2009). Antigen-presentation by mature DC will eventually result in expansion of tumour-specific T cells and inactivation of tolerogenic mechanisms.

Lentiviral vectors have been extensively used to transduce ex vivo generated monocyte-derived DC, demonstrating that these DC can be further activated and can present tumour-derived peptides to both CD4 and CD8 T cells (Breckpot et al., 2003). DC-based vaccines are patient-specific. Their generation requires specialized expertise, facilities and is time consuming. Therefore, researchers have evaluated the direct use of lentivectors as an off-the-shelve anti-cancer vaccines (Dullaers et al., 2006). It has been demonstrated that lentivectors do not provoke immunological tolerance, but instead elicit powerful cytotoxic T lymphocyte responses against transgene-encoded proteins (reviewed in (Breckpot et al., 2007a; Breckpot et al., 2008)). This suggests that lentivectors or components present in virus preparations activate innate viral-sensing pathways leading to strong adaptive immune responses. Many studies in human and mice have shown that this is in part mediated by plasmacytoid DC, a DC subset specialized in sensing viral infection. Although, it has been shown that human conventional DC are activated by lentivectors (Breckpot et al., 2007b; Tan et al., 2005), the contribution of conventional DC in vivo, the subset believed to orchestrate the anti-tumour immune response, remains elusive.

Nevertheless, many strategies have been developed to further enhance DC activation. These include: (i) the delivery of activation signals, such as constitutive active toll like receptor 4 (Xu et al., 2007) and CD40 ligand (Koya et al., 2003), (ii) overexpression of adaptor molecules, involved in DC activating innate intracellular signalling, such as TRIF and MyD88 (Akazawa et al., 2007a; Akazawa et al., 2007b), (iii) overexpression of viral molecules that interact with cellular signalling, such as vFLIP (Rowe et al., 2009), (iv) introduction of constitutive active forms of DC activators (Escors et al., 2008), as well as (v) removal of inhibitory mechanisms by RNA interference (Song et al., 2008). These strategies are extensively reviewed in (Breckpot K, 2009).

5.2. Immune silencing by gene therapy

Without any doubt, a major drawback of viral-based gene therapy is anti-vector and anti-transgene immune responses, leading to clearance of corrected cells by the immune system. In addition, lentivectors are relatively immunogenic, specially when the transgene is expressed in antigen presenting cells (Brown et al., 2006). Although this is an advantage in the case of cancer immunotherapy or vaccination, is a major drawback in gene therapy for metabolic diseases. Several approaches have been undertaken to prevent this problem. In a very elegant work, Brown and co-workers (Brown et al., 2007b) incorporated a target for the endogenous microRNA mir 142-3p present in cells of hematopoietic origin within the mRNA encoding the transgene. Transgene expression was strongly silenced by the endogenous microRNA only in antigen presenting cells, leading to long-term expression in immunocompetent mice. A second approach to silence transgene-specific immune responses involved lentivector expression of ERK and IRF3 constitutive activators together with the transgene (Escors et al., 2008). Constitutive ERK activation resulted in immature DC with down-regulated CD40 and expression of TGF-β, while IRF3 activation led to high-level secretion of IL-10. Activation of these pathways suppressed antigen-specific immune responses and expanded Foxp3⁺ regulatory T cells. These approaches can also be applied for the treatment of diseases with an auto-immune etiology such as diabetes, multiple sclerosis and rheumatoid arthritis.

5.3. Conclusions and future perspectives

For the first time, several human gene therapy clinical trials have been successful, at least from a therapeutic point of view. These have been mainly carried out using oncoretroviruses and regrettably insertional mutagenesis has proven to be a major complication. Meanwhile, lentivectors have arisen with the promise to become the substitutes of oncoretroviruses due to their improved performance and, possibly, enhanced biosafety. Lentivectors have been successfully applied in gene therapy models, generation of transgenic animals, and gene silencing by combination with siRNA and microRNA-based technologies. Overall, lentivectors offer greater advantages than their γ-retrovirus counterparts, and the first human clinical trials using them have already started. For the first time gene therapy has become a reality, and lentivectors may be the gene carriers we have been waiting for.

Abbreviations

CMV

cytomegalovirus

DC

dendritic cell

ds

double stranded

env

envelope encoding gene

gag

gene encoding structural proteins

HIV

human immunodeficiency virus

KRAB

Kruppel-associated Box

LTR

long terminal repeat

MMP

metalloproteïnase

MLV

murine leukaemia virus

NILV

non-integrating lentivector

NOD

non-obese diabetic

pol

gene encoding viral enzymes

PPT

polypurine tract

PSA

prostate-specific antigen

SCID

severe combined immunodeficiency

SIN

self-inactivating

ss

single stranded

siRNA

small interfering RNA

SU

surface

TCR

T cell receptor

TM

transmembrane

TRIP

triple helix

VSV G

vesicular stomatitis virus glycoprotein

WPRE

woodchuck hepatitis B posttranscriptional regulatory element