A novel method for generating and screening peptides and libraries displayed on adenovirus fiber

Abstract

Capsid-displayed adenoviral peptide libraries have been a significant, yet unfeasible goal in biotechnology. Three barriers have made this difficult: the large size of the viral genome, the low efficiency of converting plasmid-based genomes into packaged adenovirus and the fact that library amplification is hampered by the ability of two (or more) virus to co-infect one cell. Here, we present a novel vector system, pFex, which is capable of overcoming all three barriers. With pFex, modified fiber genes are recombined into the natural genetic locus of adenovirus through unidirectional Cre-lox recombination. Modified-fiber genes can be directly shuttled into replicating viral genomes in mammalian cells. The 'acceptor' vector does not contain the fiber gene, and therefore does not propagate until it has received a 'donor' fiber gene. Therefore, This methodology overcomes the low efficiency of transfecting large viral genomes and bypasses the need for transition to functional virus. Thus, with a fiber-shuttle library, one can generate and evaluate large numbers of fiber-modified adenovirus simultaneously. Finally, successful fiber genes can be rescued from virus and recombined back into shuttle plasmids, avoiding the need to propagate mixed viral pools. For proof of principal, we use this new system to screen a capsid-displayed peptide library for retargeted viral infection.

Figure 1.

Methods for site specifically transferring modified fiber gene cassettes into adenoviral plasmid vectors or replicating viral genomes in mammalian cells through unidirectional Cre–lox-mediated recombination. (A) The large viral plasmid vector pFex has two regions for accepting transgene cassettes, the Amp-resistant-E1-region cassette and the SacB-fiber-region cassette. The fiber-region-cassette can be recombined before (bottom) or after E1-region-cassette recombination (top). E1 cassettes are recombined through homologous recombination of adenovirus left-hand and right-hand homology regions in RecA positive BJ5183 coli. Recombinant plasmids are selected according to the newly acquired resistance cassette. On the other hand, fiber cassettes are recombined through half-mutant lox site (see legend: Green, wild-type half site; Red, mutant half site; Black and Gray represent non-compatible spacers) recombination in Cre recombinase expressing coli. Non-compatible spacers (gray center versus black center) prevent intragenic recombination. The resulting recombinant plasmids are selected by growth on sucrose containing plates. Following recombination, the donor plasmid product contains fully mutant lox sites (red circles), where the resulting recombinant vector contains fully wild-type lox sites (green circles); therefore, preventing any further recombination between the shuttles and viral vectors. The final resulting vectors are linearized and transfected into mammalian packaging cells to create viral particles (B) Fiber gene cassettes can be directly shuttled into adenoviral genomes in mammalian cells. E1-cassette-containing pFex vectors, such as AdTrack-pFex, can be pseudotyped (packaged in cells that express wild-type fiber) and used to infect packaging cell lines that express Cre recombinase. Transfection of infected cells with a fiber-gene shuttle results in site specific incorporation of the fiber cassette into replicative adenovirus. Only following recombination into a viral genome can the modified fiber gene be expressed. The resulting recombinant virus will be packaged in the newly synthesized modified-fiber capsids and amplified in other cell lines, such as 911–S11.

Figure 2.

E1 and fiber cassette exchange in plasmid vectors. (A) PacI restriction digestion of parental vector, pFex, and resulting recombinant vectors pAdTrack-Fex and pAdTrack-Luc-Fex following E1 cassette exchange in BJ5183 E. coli. The corresponding E1 Cassettes contain an additional PacI site, which results in the diagnostic 4.6 Kb band (arrow). (B) PCR across a single lox site of 294cre clones demonstrates the presence of fiber containing pFex (5′-primer within fiber gene and 3′-primer within pFex). The shuttle control does not contain lox sites and therefore does not recombine with pFex. On the other hand, PCR within the pFex vector demonstrates presence of the ‘acceptor’ pFex in both RP-Fib and shuttle-control samples. (C) XhoI restriction digest of 12 mini-preps randomly screened for the presence of the desired fiber-containing clone following fiber cassette exchange into pFex. The fiber gene contains an additional XhoI site, which generates the diagnostic 3.6 Kb product (arrow). All of the screened clones contained the desired plasmid product.

Figure 3.

Evaluation of adenovirus generated by the pFex system. (A) Viral burst comparison of pFex derived and AdEasy derived virus, both containing wild-type fiber, demonstrates equal replication and spreading rate as indicated by GFP-positive viral burst size. (B) Equal PFU of purified adenovirus containing wild-type fiber (AdTrack-WTFib), CAR-ablated fiber (ΔTAYT) or CAR ablated and RGD4C retargeted fiber (ΔTAYT-RGD) were boiled and evaluated by western blotting (top gel). Similar fiber quantities and appropriate size are demonstrated. Viral DNA of equal PFU of purified adenovirus was also subjected to PCR to evaluate content, size and homogeneity of the fiber-HI loop region (lower two gels). The larger fiber-PCR band of AdTrack-RGD4C-2 (ΔTAYT-RGD) is caused by the peptide encoding region. AdTrack-FBR2 (ΔTAYT) and AdTrack-WTFib (wild-type) do not contain recombinant HI loop peptides and therefore have identical size HI loop products. Hexon PCR was included as a reference control. (C) PC-3 and PC3-CAR prostate cancer cells were infected with 100 MOI of wild-type, CAR-ablated (ΔTAYT) and CAR-ablated and integrin retargeted (ΔTAYT-RGD) GFP-expressing adenovirus. Infected cells were identified by GFP-positive fluorescent microscopy. As predicted, ΔTAYT mutation completely detargeted virus and inclusion of the integrin targeting peptide, RGD4C, restored viral infection in a CAR independent manner.

Figure 4.

Efficiency and sensitivity of fiber gene exchange. Serial dilution mixing experiments demonstrate fiber cassette exchange sensitivity. Fiber shuttle RPuc-WTFib was serially diluted from 1:10 to 1:1 000 000 and mixed into a constant amount of mutant Rpuc-Fib1R1 (N541S) plasmid. This plasmid mix was recombined into AdTrack-pFex genomes by transfection into 293cre57 cells infected with pseudotyped AdTrack-Fex at an MOI of 1. Cell lysates were used to infect 293 cells, which were then overlayed by agar. Recombinant AdTrack-WTFib viruses were identified by viral burst (block arrow). At least a single viral burst was detectable in dilutions of 1:10 000 to 1:100 000 wild-type fiber shuttle, indicating a sensitivity of 0.01–0.001%. Some non-spreading viral infections were present (small arrows), indicating infection by AdTrack-pFex or AdTrack-N541S, which was pseudotyped in cells co-inhabited by AdTrack-WTFib viral genomes. Fiber N541S alone control (0:1) was unable to create any viral bursts.

Figure 5.

Schematic for generating, selecting and rescuing adenoviral peptide libraries. Clockwise from top left: CAR-ablated fiber-peptide library cassettes are unidirectionally shuttled into fiber-less pFex viral genomes in 293cre57 cells. Some of the resulting CAR detargeted and peptide retargeted viruses infect the target cells. Resulting viral plaques can be isolated and amplified. The floxed fiber cassette can be amplified by PCR and recombined with RPuc-Rescue in 294cre E. coli. Growth in appropriate antibiotics and 5% sucrose selects new fiber shuttles that are directly applicable to sequencing, further modification or recombination to generate new virus in plasmid or viral form.

Figure 6.

CAR-independent infection by selected retargeted adenovirus. (A) The 293 cells were transfected with two separate anti-CAR siRNAs, a negative control siRNA or non-transfected (UNT). Cells were infected with each virus at an MOI of 1, 72 h after transfection. Infected GFP-positive cells were visible the following day by fluorescent microscopy (6×) at equal exposure times. The infection rate of adenoviruses with wild-type fiber was reduced by over 50% with both anti-CAR siRNAs. CAR knock-down had no effect on infected cell numbers in CAR-detargeted and peptide retargeted viruses (B) PC-3 and the CAR overexpressing subline, PC3-CAR, were infected with each virus at an MOI of 10. Infected GFP-positive cells were visible the following day by fluorescent microscopy (6×) at equal exposure times. The infection rate of adenoviruses with wild-type fiber increased several fold in the CAR overexpressing line. However, there was no visible increase in infected cell number in CAR-detargeted and peptide retargeted viruses.