Adenovirus DNA Polymerase Loses Fidelity on a Stretch of Eleven Homocytidines during Pre-GMP Vaccine Preparation

Abstract

In this study, we invented and construct novel candidate HIV-1 vaccines. Through genetic and protein engineering, we unknowingly constructed an HIV-1-derived transgene with a homopolymeric run of 11 cytidines, which was inserted into an adenovirus vaccine vector. Here, we describe the virus rescue, three rounds of clonal purification and preparation of good manufacturing practise (GMP) starting material assessed for genetic stability in five additional virus passages. Throughout these steps, quality control assays indicated the presence of the transgene in the virus genome, expression of the correct transgene product and immunogenicity in mice. However, DNA sequencing of the transgene revealed additional cytidines inserted into the original 11-cytidine region, and the GMP manufacture had to be aborted. Subsequent analyses indicated that as little as 1/25th of the virus dose used for confirmation of protein expression (10⁶ cells at a multiplicity of infection of 10) and murine immunogenicity (10⁸ infectious units per animal) met the quality acceptance criteria. Similar frameshifts in the expressed proteins were reproduced in a one-reaction in vitro transcription/translation employing phage T7 polymerase and E. coli ribosomes. Thus, the most likely mechanism for addition of extra cytidines into the ChAdOx1.tHIVconsv6 genome is that the adenovirus DNA polymerase lost its fidelity on a stretch of 11 cytidines, which informs future adenovirus vaccine designs.

Keywords: HIVconsvX; T7 polymerase; adenovirus DNA polymerase; polymerase fidelity; protein engineering; vaccines.

Figure 1

Preparation of the virus seed stock. (Left) Schematic map of BAC p4056. The dsDNA ChAdOx1.tHIVconsv6 genome was excised from BAC p4056 using the PmeI restriction endonuclease. (Right) The VSS preparation started transfection of the genome into M9 cells in T25, and the whole cell culture harvest was amplified in T175. The resulting virus preparation was titred and tested by ID, flank-to-flank and purity PCRs, as well as for transgene product expression using Western blot of infected cell lysates assessed for Gag p24-specific mAb reactivity and size. The first round of SVI purification started by infection of 96wp cultures with an average of either 3 or 1 IU per well. The cell monolayers were inspected regularly over the following week, and only wells with a single visible plaque were collected and expanded first in 24wp wells and then in T75 to make enough virus for titration, PCR and expression testing. Three virus clones were taken into the next round of SVI. After SVI-3, the best growing/expressing clone was expanded in three T175 flasks of M9.S cells to generate the VSS, which served as the GMP starting material.

Figure 2

Control of critical steps and process intermediates. At critical stages of the clonal purification and expansion cycles, virus cultures were tested for tHIVconsv6 protein expression in infected HeLa cells using anti-HIV-1 Gag p24 mAbs (ab9071) specific for the C-terminal region 1 in a Western blot (a–e). The amount of loaded protein in each lane was visualized by direct protein staining (b–e). Transgene presence was assessed by flank-to-flank (b) and ID (c,d) PCRs (only one or the other is shown). (a) Passage-1 cells were transfected with BAC-excised linear ChAdOx1.tHIVconsv6 genomic DNA, the rescued virus was expanded by one passage and the transgene expression was confirmed. M_r markers in kDa (lane 1); MOI 500 (lane 2); MOI 1000 (lane 3); and RG vaccine ChAdOx1.tHIVconsv6 was used as a positive control (lane 4). (b) Passage 4. Following SVI-1, tHIVconsv6 expression (left) and transgene presence (bottom right) were confirmed for clones 1AC8 (lane 2), 1BE4 (lane 3), 3BA5 (lane 4), 3BG1 (lane 5) and RG (lane 6). (c) Passage 7. SVI-2 yielded clones 1BE4.3AG3 (lane 2), 3BG1.1AC11 (lane 3) and 3BG1.1BE5 (lane 4), which were compared to the RG vaccine (lane 5) and uninfected HeLa cell lysate (lane 6) for HIVconsv6 expression (left) and transgene presence by ID PCR (bottom right). (d) Passage 10. Following SVI-3, the transgene product expression (left) and transgene presence (bottom right) were analysed for clones 3BG1.1A11.1AE1 (lane 2), 1BE4.3AG3.1CC4 (lane 3), 3BG1.1BE5.3AA5 (lane 4) and 1BE4.3AG3.1AD12 (lane 5). SVI-2 clone 3BG1.1BE5 (lane 6), the RG vaccine (lane 7) and uninfected Hela cells (lane 8) were used as controls. (e) Passage 12. The VSS was generated from 3BG1.1AC11.1AE1, and tHIVconsv6 expression (lane 2) was compared to SVI-3 clone 3BG1.1AC11.1AE1 (lane 3), SVI-2 clone 3BG1.1AC11 (lane 4), uninfected Hela cells (lane 5) and the RG vaccine (lane 6) (left), with the total loaded protein visualized (right).

Figure 3

The VSS quality control. (a) The VSS established from clone 3BG1.1AC11.1AE1 was tested for genetic stability over five blind passages in the M9.S suspension cells (VSS+5), and the tHIVconsv6 protein expression was assessed in HeLa cells infected at vp MOIs of 2, 20 and 200 (lanes 2–4). Protein expression was compared to uninfected (lane 6) and RG vaccine-infected (lane 8) cells in a Western blot using HIV-1 Gag p24-specific mAb ab9071. (b) Using one cell passage from the VSS, material for the pre-clinical toxicity study in an animal model (ToxLot) was prepared and purified using AEX chromatography. A single virus peak was eluted at wavelengths of 260 nm (red) and 280 nm (blue). (c) Mouse immunogenicity. The VSS, ToxLot, VSS+5 and the RG vaccine were used to immunize BALB/cJ mice intramuscularly at a dose of 10⁸ IU per animal, and the elicited T cells were tested for recognition of eight well-defined epitopes in tHIVconsv6, of which the AMQMLKETI is downstream of the mutated polycytidine region. Reactive splenocytes were enumerated as SFU in an IFN-γ ELISPOT assay. Mean ± SD (n = 3) are shown. (d) DNA sequencing traces across the 11 cytidine regions of the starting BAC DNA, three SVI-3 clones and the VSS using one forward and two reverse primers indicated on the left. (e) tHIVconsv6 detection in Western blot by two HIV-1 Gag p24-specific antibodies given below in HeLa cells infected with the VSS (lanes 1 and 7), SVI-3 clone 3BG1.1AC11.1AE1 (lanes 3 and 8), SVI-2 clone 3BG1.1AC11 (lanes 4 and 9), uninfected (lane 5) and the RG vaccine (lane 10). (f) Schematic representation of the tHIVconsv6 protein indicating its regions, 6-5-4-3-2-1, originating from Pol (green) and Gag (navy blue); the regional junctions (yellow), including junction 2-1 generating 11 cytidines and two CD8⁺ T cells (red); and one mAb 91-5 (blue) epitopes used in the analyses. (g) Initially, the entire transgene in the VSS was sequenced directly from the purified genome (VSS). Then, the 11-C fragment was excised from the VSS genome and subcloned into pUC19. Next, 20 and, later, 100 bacterial colonies were sequenced. The table indicates the lengths of the cytidine run and their frequencies.

Figure 4

Workflow of the aborted pre-clinical development of the ChAdOx1.tHIVconsv6 vaccine. Numbers in the boxes indicate the clone numbers after SVI-1 (X), SVI-2 (X.X) and SVI-3 (X.X.X). BAC—bacterial artificial chromosome; SVI—single virion infection; Ex—expression of tHIVconsv6 protein with p24 confirmed in infected HeLa cells; Im—murine immunogenicity to 8 defined epitopes, of which AMQMLKETI is in Gag p24 downstream of the junctional 11-cytidine run; VSS—virus seed stock; VSS+1—VSS passaged one time in the M9.S cells; VSS+5—VSS plus five blind passages in the M9.S cells; DS—drug substance is the crude virus harvest at the end of the upstream process.

Figure 5

Slippage during a one-step in vitro transcription/translation. (Top) A schematic representation of DHFR ORF coupled to 11, 12 or 13 homocytidine runs and mAb tags designated Pk (aka SV5), Myc and HA (influenza virus A hemagglutinin) in three reading frames. The amino acid lengths of the expressed proteins, as well as their estimated M_r in kDa, are shown next to the schematics. The ORF nucleotide and predicted protein amino acid sequences are available in Figure S1. (Bottom) An in vitro T7 polymerase transcription/E. coli translation reaction was conducted with each of the plasmid ORFs indicated above as the template, followed by a Western blot analysis of the three separately run reactions loaded next to each other. Three identical Western blot membranes were incubated, with individual tag-specific mAbs indicated below, and a secondary enzyme-conjugated mAb, followed by ECL. The amount of loaded protein is given in the right-most panel. The predicted M_r of DHFR was 18 kDa.