Lentiviral Vectors

Douglas Jolly

BioMedica, Inc.

San Diego, California, U.S.A.

I. INTRODUCTION

Lentiviruses are a genus of the family Retroviridae (retroviruses) (1) that have been modified to be used as gene transfer agents. The major attraction of this class of vector is its perceived potency in transducing and permanently modifying nonreplicating cells, not only in tissue culture but also in animals and, it is hoped, eventually in humans. This chapter summarizes the opportunities and issues surrounding the use of lentiviruses as clinical vectors, and so tries to cover a lot of ground. The references provide examples and leads for further information, and are not meant to be comprehensive.

II. BRIEF HISTORY

Retroviruses are RNA viruses that reverse transcribe their genome into DNA and then integrate it in the host cell genome. The name was coined after the characterization of murine leukemia viruses (MLVs) and their life cycle in the late 1970s and early 1980s. The molecular understanding of the viral life cycle led to, among other things, the development of retroviral vectors based on these types of viruses (mainly C-type retroviruses such as MLV) (2) and their extensive use in the clinic (refer to http://www.wiley.co.uk/ genetherapy/clinical/ and http://www4.od.nih.gov/oba/rac/ PROTOCOL.pdf).

Human immunodeficiency virus 1 (HIV-1) was the first well-characterized lentivirus (3). The identification and characterization of HIV-1 overlapped with the development of retroviral vectors described above; however, extensive characterization of the HIV system occurred somewhat later than the development of the MLV-based systems. Because retroviral vectors were perceived to have safety issues of unknown proportions (4,5), it was not attractive at the time to add to the degree of difficulty by using a known human pathogen as the basis of such vectors. However, it was known (and rediscovered several times) that C-type retrovi-ruses such as MLV do not efficiently transduce nonreplicat-ing cells while lentiviruses can (6), and this property of MLV, although conferring potential specificity, made some desirable applications impossible or, at best, more difficult than anticipated. These desirable applications included ex vivo transduction of cells and in vivo applications, such as delivery to liver, brain, lung, bone marrow, and other tissues (the majority) where there was normally little or no cellular replication.

The first lentiviral vectors were constructed by Sodroski, Haseltine, and their collaborators (7,8) and were based on HIV-1, using natural envelope that conferred tropism for CD4+ T cells. In fact, most of these early vectors were replication competent and were produced in a manner similar to the ''wild-type'' virus. Soon after this, Page et al. (9) described efficient pseudotyping with nonlentiviral envelopes such as the amphotropic MLV envelope, but this was seen as a molecular tool for laboratory use. In the meantime, the use of pseudotyping with the vesicular stomatitis virus G (VSV-G) protein for C-type vectors (10) and, crucially, the ability to easily concentrate these pseudotyped vectors by centrifugation and resuspension (11) were described. This information, and further data from extensive characterization of the HIV system in the mid-1990s, led to the demonstration that VSV-G pseudotyped HIV vectors could be made, concentrated, and used to transduce rat neural tissues in vivo (12). Such transduction led to long-term expression. The ability to use these vectors efficiently and easily in a research lab setting sparked a good deal of activity and initiated the current intense activity in the field of lentiviral vectors.

III. TYPES OF VECTOR—MOLECULAR STRUCTURE AND FUNCTION

A. Different Lentiviruses as the Bases for Vectors

The genome of HIV-1 (3) is shown in Fig. 1, as are those of other lentiviruses [feline immunodeficiency virus (FIV), equine infectious anemia virus (EIAV), simian immunodeficiency virus (SIV), bovine immunodeficiency virus (BIV), caprine arthritis encephalitis virus (CAEV)] that have been used to make vectors (13). The HIV viral genome encodes 9 proteins: the structural proteins, Gag, Gag-Pol, and Env, plus accessory proteins: Tat, Rev, Vif, Vpu, Vpr, and Nef (Table 1). The other viruses have some of the same accessory proteins and some others. All the viruses have a Rev/Rev responsive element (RRE) system that allows export of unspliced or partially spliced viral RNA from the nucleus. All the viruses appear to be able to infect nonreplicating cells. In general, the vectors and packaging systems have moved from the first generation (Helper genome with an envelope deletion) (12) through a second generation that eliminated some of the accessory proteins (14) to third-generation systems (e.g., 15-17), shown in Fig. 2 and realized for HIV and EIAV, where almost all the unnecessary accessory proteins have been removed leaving just Rev, or Rev and Tat only. Further refinements are described below. The most thorough understanding of structure and function is, of course, for HIV-1, and the most commonly used envelope is the VSV-G protein. Vectors from the different systems appear not to differ in potency a great deal, although it is hard to be definite about this because even those experiments that claim to do so may not compare the best of both systems (e.g., 18,19). In general, vectors generated by transient transfection on 293T cells with a VSV-G pseudotype give titers around 106 transducing units (TUs)/mL as measured by a p-gal assay in tissue culture on various cell types. If you can achieve this, then you can join the club!

B. Accessory Proteins

The functions of the accessory proteins have been intensively investigated for HIV-1, but remain incompletely understood (see Table 1 and Refs. 3,20,21). It is possible to delete or inactivate all of vif, vpr, vpu, and nef and still retain viral replication in tissue culture (21), and so it is believed that most of these play a role in human or animal infection and pathology. The information concerning accessory proteins in other lentiviruses is often derived by analogy from the HIV data. Nevertheless, as shown in Fig. 1, although SIV looks quite similar to HIV, the nonprimate lentiviruses (13,22) are not as similar and do not carry analogs to HIV vpr, vpu, and nef. The Rev, Tat, and Vif proteins and functions appear to be conserved, however, with the exception that EIAV has no Vif protein and FIV has no recognizable Tat, although some tatlike functions are performed by Orf A (23). In addition, the nonprimate lentiviruses carry other genes of unknown function (e.g., S2 in EIAV) and also encode a dUTPase in the Pol polypeptide. This level of complexity is fairly daunting, and the approach of vector makers has been somewhat cavalier, but also apparently successful so far. This approach has been to eliminate these functions from vectors and helper functions as far as possible and current vector systems usually incorporate only rev, and then only in the production stage. It is unknown whether expression of such accessory proteins could, under certain circumstances, influence the properties of the vectors in uptake, transduction, and expression in animals or clinical trial subjects. However, it is also true that some of these proteins are toxic, at least in tissue culture, and their functions are incompletely understood, so the expression of these functions adds an unknown level of risk for use in animals and clinical trial subjects. The bottom line is that the vectors appear to be quite potent without the accessory proteins.

C. Integration in Nondividing Cells

As noted in Section I, a key issue has been the ability of these viruses (and the derived vectors) to infect (or transduce) nondividing cells (3,6,12). This is achieved by the passage of the intact preintegration complex of nucleic acid and protein through the nuclear membrane by a mechanism that remains unclear. The preintegration complex appears to be less massive than for C-type viruses because most of the nuclear capsid proteins are removed in the cytoplasm, with the matrix (MA) peptide and reverse transcriptase (RT)/integrase (IN) peptides retained (24,25). Contribution to the nuclear entry function have been mapped to 4 different viral functions in HIV: the MA protein of the nuclear capsid (22,26,27); the IN molecule encoded as part of the reverse transcriptase polypeptide (22,28); the accessory protein, Vpr (22,29); and the central polypurine tract (22,30). It is clear, however, that not all these are needed for the integration to occur in nondividing cells (31-33) and, for example, in some vectors only the MA and IN contributions remain, but integration in nondividing cells proceeds efficiently. The central polypurine tract appears to make the vectors more efficient in some situations but not in all (33). The MA function by itself may not be sufficient to allow the nuclear transport to occur as attempts to transplant the function into MLV vectors allowed infection of replicating, but not of growth-arrested cells (32).

Nevertheless, the lentiviruses and lentiviral vectors do not transduce all cell types by any means, even when equipped with a pantropic envelope such as VSVg, and the efficiency can vary widely. For example, in the initial description of this property (12), it was clear that HIV vectors can transduce human macrophages more efficiently than MLV-based vectors. Nevertheless the efficiency with which this happens (as measured by the effective titer on those cells) is around 1% of the efficiency on replicating tissue culture cells such as HeLa cells (17). Another example for which the entire story is not yet clear is the transduction of liver cells in vivo following intravenous (IV) administration of the vectors to animal models (mainly mice). Some investigators have experienced difficulty transducing hepatocytes in vivo (34), but other pub

Figure 1 Genomes of lentiviruses that have been used as the bases for vector construction. The genomes are shown with the genes coding for the structural proteins in green, the accessory proteins in light blue, and the long terminal repeats (LTRs) in yellow. packaging signal; SD, splice donor; RRE, rev responsive element. The diagram is not to scale. See the color insert for a color version of this figure.

Figure 1 Genomes of lentiviruses that have been used as the bases for vector construction. The genomes are shown with the genes coding for the structural proteins in green, the accessory proteins in light blue, and the long terminal repeats (LTRs) in yellow. packaging signal; SD, splice donor; RRE, rev responsive element. The diagram is not to scale. See the color insert for a color version of this figure.

lications do not report such a problem (35). This issue has been difficult to resolve because it is not clear how much exposure the total complement of hepatocytes in mature liver has to vector in the circulation, and therefore what ''efficiency'' to expect. In any case, once again, compared with MLV, the efficiencies of lentiviral vector-mediated transduction are much improved in vivo. This issue is discussed in more detail in Section V. There may be a block to hepatocyte transduction in vivo under some circumstances, but this is likely due to something other than the ability of the vector capsid to penetrate the nuclear membrane. It was originally suggested that the vectors may be better at transducing cells arrested in the cell cycle (G2) rather than those that are in G0/ Gi (12) or that have no possibility of replicating, but this

Table 1 Functions of HIV Accessory and Regulatory Proteins

Protein

Function

Tat Activation of LTR transcription, other postentry functions, able to pass into neighboring cells intact, may set up neighboring cells for infection Rev Transport of unspliced or singly spliced message from nucleus to cytoplasm Vif Acts during virus assembly to make virus particle competent for subsequent infection—dependent on the cell type Vpu Facilitation of release of budding virus particles from surfaces of infected cells—cell-type dependent and not limited to HIV Vpr G2 cell-cycle arrest, nuclear transport, suppresses immune activation and apoptosis through regulation of nuclear factor kappa B, positive regulator of viral transcription and infectivity in primary human macrophages Nef Facilitates capsid disassembly upon infection, may down-regulate MHC

turned out not to be the case (36,37); in any case, the vectors are very efficient in transducing some postmitotic cells, such as neurons, in vivo (see below).

D. Packaging Signal

The packaging signal is the sequence that allows recognition and packaging of the genomic RNA into the viral capsid before the capsid proceeds to the internal surface of the coated pits where it picks up pieces of cellular membrane and some cellular surface proteins and buds from the cell (38,39). The packaging signal in general contains a set of ''lollipop'' RNA secondary structures that can be varied in sequence as long as the structure is maintained (22,40). This site is the area of the dimerization signal for inclusion of 2 viral genomes in the nascent particle, but this trans-interaction also appears to be structurally driven and not sequence specific. The packaging signal is recognized by nucleocapsid structural (NC) peptide from the Gag polyprotein to start formation of the viral capsid (41,42). It seems likely that this occurs through interaction with zinc fingers in the NC peptide.

The original definition of the packaging signal in HIV-1 was the 19 bp deletion between the LTR and the major splice donor by Lever et al. (8). Since then, the packaging signals for HIV-1 and -2, SIVmac, EIAV, FIV, and BIV have been quite well defined (43-49). The packaging signal for these, as for most retroviruses, lies somewhere in the region between the 5'R region and includes the start of the gag sequence. It is bipartite, with the area around the splice donor probably being expendable although the spacing between the 2 segments is likely important. It is important to define this area

Figure 2 Processes for making vector preparations with third-generation packaging systems. (A) Transient process where the various elements shown are cotransfected into the recipient cell line that is usually 293T cells. This is a typical system that retains the need for Rev to get good expression of the Gag and Gag-Pol proteins, and to boost vector genome production. P, promoter, usually the CMV IE1 promoter. (B) Packaging producer cell line. All the elements necessary to make vector are in the cell line permanently. Doxycycline is used to induce expression of the VSV-G protein through the interaction with the constitutively expressed rtTA transactivator that induces expression from the TRE promoter after interaction with doxycycline or tetracycline. This is the ''tet on'' system designed by Gossen and co workers (117), but other inducible systems can be used. The blue hexagon represents the cell and the brown oval, the nucleus. The yellow box is the gene encoded by the vector, and the vector structure is simplified here with only the LTR regions shown in green. A more detailed diagram of a typical vector is shown in Fig. 3. See the color insert for a color version of this figure.

as closely as possible for 2 major reasons: (1) to obtain as high titers as possible, and (2) to identify and minimize obligate areas of homology between the vectors and the packaging constructs because these have been shown to lead to homologous recombination between these sites.

Homologous recombination reunites elements that were separated during the design of the vector system and could be a first step on the road to generating a replication-competent entity. In general it is easier to define what is necessary (by deletion analysis) than what is sufficient. The latter requires a more complicated molecular manipulation and the results are usually not black and white. From the viruses' point of view, the inclusion of gag sequence in the packaging signal makes sense because it allows for selection of full-length un-spliced genomic RNA for encapsidation. From the vector makers' point of view, this is a nuisance because the gag sequence provides difficult-to-avoid sequence homology to the gag sequence in packaging plasmids provided in trans during vector production. Parenthetically, HIV-2 (and hence possibly SIVmac because these 2 viruses are closely related) is reported to be an exception and the gag sequence may be a relatively unimportant part of the packaging signal (44). The gag recombination issue for vector construction could be addressed by simply leaving out part of the packaging signal, but in general this reduces titers considerably. One effective solution that is now in common use is to build expression vectors for the viral packaging proteins, Gag and Gag-Pol, with optimized/different amino acid coding sequences (50,51). This has several advantages: elimination of homology to the packaging signal as discussed here, potentially higher levels of Gag-Pol expression, and elimination of the dependency of Gag-Pol expression on Rev expression in the packaging system.

E. Gene Expression and RNA Nuclear Export Enhancers

Lentiviral genomes carry a series of elements that, in the virus, ensure the orderly progression of the viral life cycle, but that complicate life for vector makers (22). These include the Tatmediated control of transcription that renders the HIV LTR a poor promoter in its absence, and the existence of RNA nuclear export inhibitory sequences in env and gag-pol primary transcripts, which are counteracted by the interaction of the Rev protein with the RRE. This allows export of the messages for the structural proteins and the full-length viral genome to the cytoplasm. These issues are addressed in the ''third-generation'' vector systems (15-17,46,47,52,53).

The Tat dependence of the LTR has been avoided by substituting other promoters (usually the CMVIE1 promoter that is very active in 293 cells) at the 5' end, while leaving in place those elements that are needed to allow the transcript to be reverse transcribed and integrated (tRNA primer site and R region, packaging signal) (see also Figs. 2 and 3). The Rev dependence has been manipulated by including the RRE in the vector genome, but positioning it so that it is excluded from the transcript of the desired transgene, as it was believed

Figure 3 Components of a minimal vector system. The figure shows the 3 components necessary for Rev-independent packaging systems, such as with those designed with codon optimized gag-pol genes (50,51). The vector is also expressed without the need for Rev, and is shown in its plasmid configuration. When the transcribed RNA vector genome is packaged into a vector particle, then reverse transcribed and integrated into the genome of a target cell, the 5'CMV promoter disappears and is replaced by the SIN LTR, which has no promoter function. The encoded gene is expressed from the inner CMV promoter. The system shown is used for transient transfection. For a packaging cell line using the VSV-G protein as envelope, the gene encoding VSV-G would need to be inducible. PCMV, CMV IE1 promoter. See the color insert for a color version of this figure.

Figure 3 Components of a minimal vector system. The figure shows the 3 components necessary for Rev-independent packaging systems, such as with those designed with codon optimized gag-pol genes (50,51). The vector is also expressed without the need for Rev, and is shown in its plasmid configuration. When the transcribed RNA vector genome is packaged into a vector particle, then reverse transcribed and integrated into the genome of a target cell, the 5'CMV promoter disappears and is replaced by the SIN LTR, which has no promoter function. The encoded gene is expressed from the inner CMV promoter. The system shown is used for transient transfection. For a packaging cell line using the VSV-G protein as envelope, the gene encoding VSV-G would need to be inducible. PCMV, CMV IE1 promoter. See the color insert for a color version of this figure.

to be desirable for high-level expression of the vector genomes. The transferred gene itself is expressed from an internal promoter. In most third-generation systems (Fig. 2), Rev is necessary anyway in the packaging cell to obtain good levels of expression of the Gag and Gag-Pol proteins. The packaging cell requirement for rev can be eliminated by the use of codon optimized gag and gag-pol genes, as noted above (50,51).

There has been benefit in some situations to add extra nucleic acid components to the vector genome to facilitate steps in the transduction cycle. The most commonly used elements include the cPPT element from the virus itself (see Section III.C) (30), which is part of the reverse transcriptase coding sequence and is a start site for viral genome replication (54,55); the Mason-Pfizer monkey virus constitutive transport element or analogs that facilitate export of unspliced RNA by interacting with a cellular activity (56); and the Woodchuck hepatitis virus posttranscriptional regulatory element element (57) that enhances the translation activity of the message in which it is incorporated by several possible mechanisms. All these elements can be shown to be quite potent in enhancing titer or potency of expression in some situations, but are also ineffective in others (e.g., 33). However, they have never been shown to be deleterious and so they or analogs are often included in the design of vectors without significant subsequent proof of their utility in the particular situation.

F. Envelope Proteins and Pseudotyping

The vast majority of lentiviral vectors have been made using the VSV-G pseudotype of the vector particles. This is for 2

main reasons: it works well in almost all vector systems to give high in vitro titer; and the vector produced can be easily concentrated up to about 1000-fold at the research scale, by centrifugation (11). This in turn yields preparations of vector that have titers of around 109 TU/mL, which turns out to be what is needed for a number of in vivo applications. However, numerous other pseudotypes (see Table 2 for some examples) have been made to try to provide tissue tropism, explore tar-getable envelopes, look for a nontoxic protein (as opposed to VSV-G) and hence simplify vector production, provide higher titers and also simply explore what novel properties such pseudotypes might possess. Although many such pseudotypes have been made, there is only preliminary information on the properties of such vector preparations in vivo, where it counts. An intriguing example has been the report of retrograde transport of the vector genome up the axons of neurons to the nucleus with rabies G pseudotyped vectors (58). This raises the possibility of introducing genes into the easily accessible muscle of a clinical subject and transducing the nuclei of neurons in the not-easily-accessible spinal cord, with prospects for treating motor neuron diseases.

Another aspect of the viral/vector life cycle that is beginning to be understood is the manner in which the viral capsid hijacks the endosomal transport pathways in order to travel to the inner side of the cell membrane. How envelope proteins gather there in the area of a coated pit to which the capsid arrives is less understood, as are the rules governing the subsets of the cellular surface proteins that gather there and bud off with the cellular membrane and the envelope protein (71). For example, HIV includes MHC class I and II molecules in the membrane of the particles (72). In MLV-based vectors, the amphotropic envelope probably allows inclusion of the CD55 and CD59 molecules that protect against complement attack in vivo, but VSV-G probably excludes them (73). The rules for such exclusion and inclusion of cellular markers are incompletely understood and have not been investigated in the context of the vectors or the different possible pseudotypes. One obvious consequence is that it is not known, for example, to what extent the vectors are resistant to degradation by human complement.

G. Other Cell and Tissue Tropism Issues

In vector systems, another classic way that does not involve envelope switches to achieve tissue-specific effects is to use a cell type-specific, or environment-specific, promoter and enhancer. This works quite well with lentiviral vectors (e.g., 74) (better, in fact, it seems than with MLV-type vectors), although there is not a large amount of data in vivo where, as noted before, it counts. The ease of accomplishment is related to the fact that ''SIN'' vectors (i.e., those with deletions in the 3'LTR that migrate to the 5'LTR after 1 round of reverse transcription and integration) (75) seem to work better with lentiviral vectors than with MLV-based vectors.

However, in addition to these intentional tissue-targeting issues there are innate tropisms in the vectors themselves that sometimes are unexpectedly observed. The best known of these is probably the failure of many HIV vectors to productively transduce some monkey cells (76). Further investigation shows that this phenomenon is quite common and can apply to nonprimate lentiviral vectors in primates (77). As primates are considered desirable models for human safety studies, and

Table 2

Examples of Various Pseudotypes of Lentiviral Vectors and Their Properties

Was this article helpful?

0 0

Post a comment