The second and third packaging systems, VSV-G envelope, selfinactivating features, cPPT and WPRE elements, have been widely used in lentiviral vectors. In this section we describe additional modifications that further reduce the risk of generation of RCLs during vector production. The mechanisms of genetic recombination have been extensively studied in γ -retroviruses.

Molecular analysis of the RCR (replication-competent retrovirus) has revealed that as little as 10 bp of nucleotide identity between a packaging construct and a vector constructs can mediate homologous recombination and RCR production. Homologous recombination can occur when two different RNAs are packaged into one virion followed by RT-mediated strand transfer (template switching).

Viral RT-mediated recombination is also observed in HIV-1 reverse transcription [58]. Given that HIV-1-based vectors have multiple overlapping cisacting sequences that are shared between vector and packaging constructs, recombination between these sequences is formally possible. Some of these sequences are difficult to impossible to remove from all constructs.

For example, the RRE sequence must be on both constructs. The gag gene must of course be present in the Gag expression plasmid, but a part of gag gene must also be included in the transfer vector for it to be efficiently packaged. In addition, the cPPT element in the pol gene can be on both constructs. Accordingly, packaging signal and gag recombinants have been frequently detected in HIV-1-based vector preparations, although no RCL has been observed in second- or thirdgeneration HIV-1 vectors.

However, it should be noted that an artificial replication-competent recombinant lentivirus with VSVG and accessory genes vif and vpr has been produced. To minimize the possible risk of generating RCLs, we can:

  1. further reduce the homologous sequences between vector and helper sequences; and
  2. further divide helper functions into multiple plasmids. On the basis of those concepts, additional modifications have been introduced into lentiviral vectors, which we summarize below, along with additional modifications leading to enhanced lentiviral vector performance in certain applications.

Codon optimization for Rev-independent Gag-Pol expression

The binding of HIV-1 Rev to RRE in HIV-1 transcripts results in the nuclear export of those intron-containing transcripts. Some studies have found AU-rich destabilizing sequences in the gag gene, which can be stabilized by the Rev–RRE interaction. Indeed, the HIV-1 genome, especially gag, pol and env genes, is highly AU-rich, and this imparts a codon bias that is highly different from the one used by human genes.

Intriguingly, codon optimization of the HIV-1 gag-pol gene not only increases Gag-Pol protein expression, but also makes these genes Rev-independent. Since many of the third bases are changed in codons to increase expression, this also fortuitously disrupts homology of the synthetic gag sequence with the native gag sequence needed for packaging of the transfer construct. This results in a significantly lower rate of recombination between the packaging and vector genome constructs.

Rev-independent transfer vectors

In contrast with the successful Rev-independent Gag-Pol expression, generating a Rev-independent transfer vector construct has been challenging. The CTE (constitutive transport element) from Mason–Pfizer monkey virus has been used to replace the RRE/Rev mRNA transport mechanism. Although CTE elements can rescue the Rev/RRE-independent Gag-Pol and/or vector genome expression, the resulting vector titres are generally low.

Trans-lentiviral vector system

The trans-lentiviral vector system has been developed through splitting the Gag/Gag-Pol packaging element into two separate plasmids, one that expresses Gag and PR, and another that expresses RT and IN. For efficient encapsidation of the RT and IN polyprotein, the RT–IN polyprotein is expressed as a fusion protein with the virion-associated accessory protein Vpr. These trans-lentiviral vectors are currently produced from four separate vector components:

  1. Gag-, Pro-, Vif-, Tat- and Rev-expressing packaging construct;
  2. Vpr–RT–IN-expressing packaging construct;
  3. VSV-G/Env-expressing construct; and
  4. Tat-dependent transfer vector.

Lentiviral vectors from other primate lentiviruses

To reduce safety concerns related to the generation of RCLs, White et al. developed a hybrid lentiviral vector system, where HIV-1 vector genome constructs are packaged by the viral proteins of a non-virulent SIV (simian immunodeficiency virus) strain, SIVmac 1A11. This design increases the predictable biosafety owing to the reduced homology between HIV-1 and SIVmac.

Lentiviral vectors derived from other primate lentiviruses (i.e. HIV-2 and SIV) have also been developed and demonstrated efficient gene transduction of dividing and nondividing cells. Although capsid-dependent post-entry restriction of SIV vectors has been observed in rodent cells, SIV vectors have demonstrated efficient gene transfer into haematopoietic stem cells in non-human primate models.

Lentiviral vectors derived from non-primate lentiviruses

Non-primate lentiviruses, including FIV (feline immunodeficiency virus) and EIAV (equine infectious anaemia virus) have highly restricted species tropisms and cannot replicate in human cells. Their narrow host ranges are primarily due to the lack of their functional receptors on human cells, whereas an additional intracellular block has been observed in human cells.

A series of non-primate lentiviral vectors have been developed with expanded tropisms through pseudotyping with the VSV-G glycoprotein and by the introduction of a strong internal promoter into the constructs. A variety of dividing or non-dividing cells can be transduced by nonprimate lentiviral vectors, including primary aortic smooth muscle cells, hepatocyes, dendritic cells and neuronal cells, although some studies have observed post-entry restrictions to FIV- or EIAV-based vectors in some human cells.

It has been proposed that non-primate lentiviral vectors may be safer alternatives to HIV-based vectors, because they should not be able to replicate in humans even if replication-competent viruses are produced. Conversely, it has also been proposed that these lentiviral vectors may be able to replicate in humans if the VSV-G gene is incorporated into the recombinant virus.

If so, it may well be safer to develop vectors from HIV-1, whose molecular biology and pathogenesis are better understood, and for which multiple drug therapies are available. In this light, it should be noted that in utero and neonatal gene transfer in mice with EIAV, but not HIV-based vectors, has led to a very high incidence of liver and lung tumours [85]. The causal mechanisms leading to this oncogenesis by the EIAV vectors remain elusive.

Chromatin-insulator elements and chromatin-opening elements

If a lentiviral vector integrates into a transcriptionally inactive region of the chromosome, their transgene expression may also be repressed by surrounding chromatin. In addition, epigenetic position effects can also affect the levels of transgene expression. Insulators are DNA sequences that can shield enhancers and promoters from the activation or silencing by adjacent chromatin (reviewed in).

Given this, chromatininsulator sequences have been incorporated into lentiviral vectors to increase the efficiency of transgene expression. The cHS4 (chicken hypersensitive site-4) insulator has been used to increase vector transgene expression, although some studies found minimal effects with this modification. Similar elements used to prevent vector silencing are the enhancerless UCOEs (ubiquitously acting chromatin-opening elements). Introduction of human HNRPA2B1-CBX3 UCOE (A2UCOE) has demonstrated highly reproducible and stable transgene expression in haematopoietic stem cells through blocking DNA methylation-mediated silencing of lentiviral vectors.

Packaging elements from different HIV isolates for altered vector tropism

Gag is essential for HIV genome packaging, virion assembly and budding, whereas Pol is necessary for virion maturation, reverse transcription and viral integration. Although there are many genetically divergent HIV-1 strains with diverse biological properties, most HIV-1 vectors have been derived from only a few well-characterized clones of laboratory-adapted T-cell-tropic HIV-1.

Therefore it is likely that more robust lentivectors or vector elements can be found in the wealth of HIV-1 genetic diversity beyond the common strains. Indeed, we and others have shown that naturally occurring substitutions in gag can actually significantly alter vector performance, vector titres and the ability to infect traditionally refractory cells.

On the basis of this, new lentiviral vector systems based on new strains are being developed. For example, new vectors based on a macrophagetropic HIV-1 YU-2 clone have demonstrated comparable gene transfer efficiency with widely available vectors.

Lentiviral vectors with a SIV accessory protein Vpx for improved transduction of monocytes and dendritic cells

Several primate lentiviruses have an accessory protein called Vpx. VSV-G-pseudotyped SIV vectors generated with only Gag, Pol, Tat and Rev helper functions cannot transduce human monocytes and dendritic cells. However, this defect can be rescued when Vpx protein is expressed in vector-producing cells. Similarly, Vpx-containing lentivectors, based on a pathogenic sooty mongabey monkey isolate (SIVsmm-PBj), can efficiently transduce freshly isolated human monocytes.

Recent studies identified SAMHD1 (sterile α-motif domain and HD domain 1) as the dendritic- and myeloid cell-specific HIV-1 restriction factor and that this factor is counteracted by Vpx. Codelivery of Vpx-containing viral-like particles not only enhances SIV vector transduction, but also potently enhances HIV-1- based vector infection of monocyte-derived dendritic cells


Targeted gene delivery and expression to organs and tissues of interest is the ultimate goal for many gene-delivery applications. There are two steps in lentivector infection, which can be modified for targeted vector transduction, i.e. vector entry and gene expression. In this section, we will review strategies to generate tissue-specific lentivectors.

Limited vector entry through pseudotyping with heterologous viral glycoproteins

Combining viral particles with a foreign envelope glycoprotein can alter the host tropism. Because of the successful incorporation into HIV-1-like particles and its high transduction efficiency in various cell types, lentivirus vectors are commonly pseudotyped with VSV-G. However, VSV-G-pseudotyped vectors can show cytotoxicity at high concentrations, whereas VSV-G-mediated non-specific gene delivery into undesired cell types poses a safety concern for their use in the clinic.

Many viruses use specific receptors for viral entry, which can be exploited for selective lentiviral vector delivery. γ – retroviruses, such as amphotropic (broad host range) MLVs, gibbon ape leukaemia virus, feline endogenous retrovirus RD114 and a xenotropic MLV-related virus XMRV (xenotropic MLVrelated virus), have been used to target their original receptors PIT2 [SLC20A2 (solute carrier family 20 member 2)], GLVR1 (SLC20A1), neutral amino acid transporter (SLC1A5 gene) and XPR1 (xenotropic and polytropic retrovirus receptor 1), which are broadly, but less ubiquitously, expressed in gene therapy target cells. Pseudotyping with the viral surface glycoproteins from other RNA viruses, including Ross River virus, Semliki Forest virus, lymphocytic choriomeningitis virus, rabies virus, Mokola virus and Ebola virus, also yields hightitre vectors.

Intriguingly, amphotropic MLV Env or RD114 Env-pseudotyped lentiviral vectors show efficient transduction of human CD34 + haematopoietic stem cells, whereas Ebola virus glycoprotein-pseudotyped vectors efficiently transduce airway epithelia from the apical surface. One caveat for the use of non-VSV-G glycoproteins for pseudotyping is that resulting vectors can be more fragile/unstable than VSV-G-pseudotypes. For instance, γ – retroviral Env-pseudotyped HIV-1 vectors are generally more sensitive to freeze and thaw. We have also observed large amounts of particle-free amphotropic MLV Env in vector preparations competing with MLV-pseudotyped vectors during vector transduction.

Targeting lentiviral vectors with bioengineered ligand/antibody-displaying Envs

Significant efforts have been devoted to generate targeting lentiviral vectors using a ligand protein or antibody fused to viral glycoproteins to retarget the lentiviral particles to specific cell-surface molecules. For instance, pseudotyping with an MLV Env glycoprotein engineered to display anti-CD3 single-chain antibody significantly improves transduction of primary lymphocytes.

Similarly, lentiviral vectors coated with a modified chimaeric Sindbis virus envelope conjugated with anti-P-glycoprotein antibody can selectively transduce Pglycoprotein on metastatic melanoma. One of the most promising re-targeting envelope platforms is based on the F and H glycoproteins of measles virus that allows targeted lentiviral entry into CD20-positive B-lymphocytes or quiescent T-cells.

Selective vector transgene expression through introduction of a tissue-specific promoter

Use of a cell-type-specific regulatory element/promoter as an internal promoter can target vector transgene expression to the cells of interest. Unlike the envelope-mediated entrytargeting strategies, vectors with a tissue-specific promoter can basically enter and integrate in any cell types. However, their transgene expression is limited to a certain cell type by the internal promoter. Various tissue-specific promoters have been incorporated into lentiviral vectors, including neuron-specific, dendritic cell-specific, tumour angiogenesis-specific, vascular differentiation-targeted or hepatocytespecific promoters.

Selective vector transduction through miRNA (microRNA)-mediated silencing in undesired cells

A total of 21–22 nucleotides of non-coding miRNAs mediate post-transcriptional gene regulation. Lentivector-carrying miRNA vector has been developed as a new approach to limit undesired vector transgene expression. For instance, to minimize off-target transgene expression in antigenpresenting cells, lentivectors have been engineered to carry a haematopoietic-specific miRNA, miR-142-3p, which leads to reduced transduction of antigen-presenting cells and antitransgene immune responses.