Selection of novel vesicular stomatitis virus glycoprotein variants from a peptide insertion library for enhanced purification of retroviral and lentiviral vectors

Abstract

The introduction of new features or functions that are not present in an original protein is a significant challenge in protein engineering. For example, modifications to vesicular stomatitis virus glycoprotein (VSV-G), which is commonly used to pseudotype retroviral and lentiviral vectors for gene delivery, have been hindered by a lack of structural knowledge of the protein. We have developed a transposon-based approach that randomly incorporates designed polypeptides throughout a protein to generate saturated insertion libraries and a subsequent high-throughput selection process in mammalian cells that enables the identification of optimal insertion sites for a novel designed functionality. This method was applied to VSV-G in order to construct a comprehensive library of mutants whose combined members have a His6 tag inserted at likely every site in the original protein sequence. Selecting the library via iterative retroviral infections of mammalian cells led to the identification of several VSV-G-His6 variants that were able to package high-titer viral vectors and could be purified by Ni-nitrilotriacetic acid affinity chromatography. Column purification of vectors reduced protein and DNA impurities more than 5,000-fold and 14,000-fold, respectively, from the viral supernatant. This substantially improved purity elicited a weaker immune response in the brain, without altering the infectivity or tropism from wild-type VSV-G-pseudotyped vectors. This work applies a powerful new tool for protein engineering to construct novel viral envelope variants that can greatly improve the safety and use of retroviral and lentiviral vectors for clinical gene therapy. Furthermore, this approach of library generation and selection can readily be extended to other challenges in protein engineering.

FIG. 1.

VSV-G-His₆ library design. (A) Structure of the pCLPIT VSV-G-His₆ vector, which expresses the VSV-G-His₆ library from a tetracycline regulatable promoter and puromycin resistance from the viral long-terminal repeat (LTR). IRES, internal ribosome entry site; tTA, tetrecycline-controlled transactivator; TRE, tetracycline response element. (B) Peptide and DNA sequences for the His₆ insert after insertion. X₁ and X₂ will depend on the five-host nucleotides (N) duplicated during insertion. The digested insert sequence is underlined and in bold. (C) The pCLPIT VSV-G-His₆ plasmid library and clones were cut once in the His₆ insertion and once in pCLPIT. Successful insertions into vsv-g yield fragments of 1.6 to 3.2 kb in size. Lanes: 1, pCLPIT VSV-G; 2, pCLPIT VSV-G-His₆ library; 3 through 12, randomly selected pCLPIT VSV-G-His₆ clones.

FIG. 2.

Expression of library proteins. (A and B) Immunostaining of cells transfected with pCLPIT VSV-G-His₆ or pCLPIT VSV-G to detect intracellular (A) and surface (B) expression of VSV-G (white, 63× objective). Cells are counterstained with TO-PRO-3 (gray). (C) Western blot detection of VSV-G-His₆ library proteins binding to Ni-NTA.

FIG. 3.

VSV-G-His₆ library selection. (A) Schematic of library selection by using retroviral infection of cells. (B) Viral titers for each round of selection for replication. Error bars represent the standard error of the linear regression used to determine titers.

FIG. 4.

Immunofluorescence detection of VSV-G-His₆ clones. Detection of (A) intracellular and (B) surface expression of VSV-G (white) from individual VSV-G-His₆ clones (63× objective). Cells are counterstained with TO-PRO-3 (gray).

FIG. 5.

Column purification of VSV-G-His₆-pseudotyped retroviral and lentiviral vectors. (A) Representative titers of retroviral and lentiviral vectors expressing eGFP pseudotyped with VSV-G-His₆ variants. Results for the G-25LH₆-pseudotyped retroviral vector reflect packaging at 30°C. All other vectors were produced at 37°C. Error bars represent the standard error of the linear regression used to determine titers. (B) Recovery of vectors pseudotyped with VSV-G-His₆ variants after Ni-NTA purification. Error bars represent the standard error of the linear regression used to determine titers. (C) Optimized purification profile of G-19LH₆- and G-24LH₆- pseudotyped lentiviral vectors. Error bars represent the standard error of the linear regression used to determine titers. (D) Silver staining of column fractions. Lanes: 1, marker; 2, IMDM with 10% fetal bovine serum (1:10 dilution); 3, vector supernatant (1:10 dilution); 4, ultracentrifuged virus; 5, column flowthrough; 6 to 8, successive washes; 9 to 12, successive eluates.

FIG. 6.

Behavior of VSV-G-His₆-pseudotyped lentiviral vectors in vivo. (A) Representative images of injections with VSV-G-His₆-pseudotyped vectors display equivalent eGFP expression (green) to vectors pseudotyped with WT VSV-G (10× objective). Cells were counterstained with TO-PRO-3 (blue). (B) Representative images show that VSV-G-His₆ and WT VSV-G-pseudotyped vector tropism are equivalent in the brain (63× objective). Cells were stained with antibodies against NeuN (blue) and GFAP (red) to identify mature neurons and astrocytes, respectively. (C) Vector spread through the brain for each preparation based on the number of eGFP⁺ sections. The P value was >0.3 by analysis of variance (ANOVA). Error bars represent the standard error of the mean of each preparation. AP spread, anterior-posterior axis spread. (D) Overall volume was assessed by each vector preparation based on eGFP expression in 22 to 26 sections per animal. The P value was >0.4 by ANOVA. Error bars represent the standard error of the mean of each preparation. AP spread, anterior-posterior axis spread. (E) Reduction in immune response by using column-purified vectors. Immunostaining of CD8⁺ T cells (red, OX8) and macrophages (red, ED1) from animals that were injected with column-purified or conventionally purified viral stocks (10× objective). Cells are counterstained with TO-PRO-3 (blue). Images are representative areas of high eGFP expression that were at least 200 μm away from the site of injection and the corpus callosum to avoid bias introduced by enhanced transport in these areas.