Directed evolution of hyperactive integrases for site specific insertion of transgenes

Abstract

The ability to deliver large transgenes to a single genomic sequence with high efficiency would accelerate biomedical interventions. Current methods suffer from low insertion efficiency and most rely on undesired double-strand DNA breaks. Serine integrases catalyze the insertion of large DNA cargos at attachment (att) sites. By targeting att sites to the genome using technologies such as prime editing, integrases can target safe loci while avoiding double-strand breaks. We developed a method of phage-assisted continuous evolution we call IntePACE, that we used to rapidly perform hundreds of rounds of mutagenesis to systematically improve activity of PhiC31 and Bxb1 serine integrases. Novel hyperactive mutants were generated by combining synergistic mutations resulting in integration of a multi-gene cargo at rates as high as 80% of target chromosomes. Hyperactive integrases inserted a 15.7 kb therapeutic DNA cargo containing von Willebrand Factor. This technology could accelerate gene delivery therapeutics and our directed evolution strategy can easily be adapted to improve novel integrases from nature.

Graphical Abstract

Figure 1.

PE optimization and PhiC31 integrase targeted insertion into the human genome. (A) Schematic of PE directed integrase gene insertion. PE efficiently inserts sequences <100 bp. PE is first used to insert the 35 bp attB site at a desired genomic locus, such as ROSA26. The attB site acts as a landing pad for PhiC31 integrase which can efficiently insert multi-kb sequences. During a second step, the integrase recombines the attB site found in the genome with an attP site found on the 6.6 kb donor plasmid. Following recombination, the entire donor plasmid becomes inserted into the genome and the attB and attP sites are rearranged to form attL and attR sites flanking the insertion. Because the integrase cannot recombine the newly formed attL and attR sites, the donor plasmid containing the hygromycin antibiotic resistance gene and DsRed fluorescent protein gene is permanently inserted into the genome. (B) Amplicon sequencing was used to compare efficiencies of PE for insertion of PhiC31 integrase attB (35 bp) or attP (41 bp) sites at AAVS1, Xq22.1 and ROSA26 in HEK293 cells. For each target sequence, twin epegRNA were used (listed on the x-axis). Amplicon sequencing primers were designed to flank each target sequence. PE efficiencies reflect sequencing reads containing the intended insertion of the att site. Indel efficiencies reflect reads with unintended insertions or deletions. n = 2. (C) Integration efficiency of PhiC31 integrase mutants from Keravala et al. (39) at pre-inserted attP sites (AAVS1 and ROSA26) in clonal HEK293 cell lines. Cells previously modified to contain a PhiC31 attP site were co-transfected with a 6.6 kb donor plasmid containing an attB site, hygromycin resistance gene, and DsRed gene along with a helper plasmid containing one of four integrase variants. Integration of the donor plasmid was measured by ddPCR using a forward primer and probe designed to bind the genome and reverse primer designed to bind the inserted sequence. n = 3. (D) PE plasmids, including attP41 4a + 4b pegRNA from (B) designed to insert the PhiC31 attP site, were co-transfected with a donor plasmid containing an attB site and a helper plasmid encoding PhiC31 integrase variants in HEK293 cells. Integration efficiency at ROSA26 was measured by ddPCR. n = 3. Data are shown as mean + s.d. All ddPCR measurements were normalized using the product/reference ratio to account for differences in the number of target sequences in the genome (Supplementary Note S1 and Supplementary Table S1).

Figure 2.

IntePACE. Schematic of IntePACE. The integrase that is to be evolved is encoded in the phage genome. The phage gIII, which is essential for phage infectivity, is removed from the genome. PIII, encoded by gIII, is naturally separated into N-terminal and C-terminal domains by a flexible linker. The accessory plasmid (AP) contains the attP site and C-terminal portion of gIII. The complementary plasmid (CP) contains the attB site and a promoter driving the N-terminal portion of gIII called the leader sequence. Expression can be tuned by replacing a promoter or ribosome binding site (RBS) of desired strength. Both AP and CP are maintained in the host E. coli cells, and because gIII is split, no functional pIII is expressed. Key to IntePACE is the link between gIII expression and the desired activity of the integrase, specifically, the recombination of the attB site on the CP with the attP site on the AP. Upon crossover between the att sites the AP and CP join to form a single recombined plasmid (RP) and the promoter, leader sequence, and C-terminal portion of gIII join to permit the expression of pIII. During IntePACE, host cells harboring both the AP and CP are continuously added to a fixed volume vessel called the ‘lagoon’. The cells are then infected by the selection phage (SP) leading to expression of the integrase. Highly active integrase variants are expected to recombine more plasmids than less active variants. This results in higher levels of pIII and higher numbers of infectious progeny from the phage that encode the integrase variant capable of recombining the att sites. The lagoon is continuously drained at the same speed as the host cells are added. Integrase variants with poor efficiency of recombination do not produce enough infectious progeny and are diluted from the system. Stringency can be modulated by controlling the speed of the outflow of the host cells. Stringency can also be increased by reducing the strength of the promoter on the CP to limit the amount of gIII expression. During stringent conditions, only highly active integrases can recombine enough AP and CP at low pIII levels or recombine quickly enough to produce progeny before the lagoon drains the cells. New mutations are incorporated into progeny from phage encoding successful integrase variants due to the presence of the mutation plasmid (MP6) (40) in the host cells. Therefore, each cycle of phage replication generates a new mutation library from the fittest variants of the previous library.

Figure 3.

IntePACE evolved hyperactive PhiC31 integrase. (A) IntePACE lagoon phage titers. Six lagoons were inoculated with phage containing SP encoding PhiC31 P1, P2 and P3 integrase (two lagoons each). Host cells containing the MP, AP and CP with one of four gIII promotors (in order of strongest to weakest) ProB, ProA, Pro3 and Pro1 (53) were exchanged to increase stringency at 0, 76, 124 and 176 h time points (black dotted lines), respectively. The flow rate was increased from 2–3 lagoon volumes per hour (black solid line) at the 188 h time point. Phage were collected at least every 4 h. Titers counted by activity-independent plaque assay are shown on the y-axis in log scale. (B) Activity in E. coli of PhiC31 integrase evolved libraries from IntePACE time points assayed by overnight enrichment assay. Cells containing AP and CP were infected by phage encoding integrase variants isolated from IntePACE. Active integrase variants recombine the AP and CP resulting in pIII expression and increased phage progeny. Recombination activity was measured by counting plaques generated by the phage. P2 (lagoons L1 and L2), P1 (lagoons L3 and L4), and P3 (lagoons L5 and L6) (shown in bold font) represent starting point phage at the 0-hour time point (x-axis). Time point labels begin with lagoon number (L1–L6) followed by the hour the phage was collected (x-axis) n = 2. (C) Schematic for strategy for assaying IntePACE derived clones in human cells. During IntePACE, integrase variants are challenged by stepwise increases in stringency as phage concentration is measured at least every 4 h. Promising mutation libraries are chosen from time points with high phage titer (black circles). Next, the activity of the different libraries is compared in E. coli using the overnight enrichment assay. Integrase variants from top libraries are cloned into mammalian expression plasmids. PE and integrase plasmids are transfected into HEK293 cells. Following insertion of the att site by PE, integrase variants insert the donor plasmid by recombining the att site found on the donor plasmid with the att site now found in the genome. The ddPCR probe and forward primer are located in the genome. Both PE and insertion efficiency are measured using ddPCR using reverse primers found on the att site or donor plasmid, respectively. (D) Helper plasmids encoding evolved PhiC31 integrase mutants were co-transfected with a 6.6 kb donor plasmid into a clonal HEK293 cell line containing a pre-installed attP site at ROSA26. Two lagoons were used for each starting variant. Helper plasmid clone names containing starting variant P1, P2 or P3, and the first or second lagoon number L1 or L2 for each, and clone number are listed on the x-axis. All mutations are listed in Supplementary Table S7. n = 3. (E) Integration of 6.6 kb donor plasmid by select evolved PhiC31 integrase mutants with PE insertion of the attP site into unmodified HEK293 cells at ROSA26 (data for all mutants shown in Supplementary Figure S2D). n = 3. Data are shown as mean + s.d. (*) P value < 0.05 derived from a Student's two-tailed t-test if significantly higher than the pre-evolved integrase control (shown in bold on the left).

Figure 4.

IntePACE evolved hyperactive Bxb1 integrase. (A) Comparison of both PE efficiency of insertion of the att site and integration efficiency of a 6.6 kb donor plasmid of wildtype PhiC31, evolved PhiC31, Pa01 and Bxb1 integrases in HEK293T cells measured by ddPCR. n = 3. (B) Bxb1 IntePACE lagoon phage titer (colored lines) measured by activity-independent plaque assay shown on the y-axis in log scale. Similar to PhiC31 IntePACE (described in detail in Figure 3A), stringency was increased throughout the experiment. The gIII promoter strength was decreased by switching to host cells containing sequentially weaker promoters: ProB, ProA, Pro3 and Pro1. Host cells were changed every 2 days in six lagoons during the first experiment (black dotted lines) which lasted 5 days. In PACE 2, two lagoons were seeded with selected phage from PACE 1 (L7 was seeded with PACE 1 L4-120 h and L8 was seeded with PACE 1 L6-100 h). Flow rates were increased incrementally each day (black solid line). (C) Bxb1 integrase mutants isolated from IntePACE were cloned into mammalian expression plasmids and used to measure insertion efficiency of a 6.6 kb donor plasmid (process described in Figure 3C). Integration efficiency of evolved Bxb1 integrase mutants in a clonal HEK293 cell line with a preinstalled ROSA26 attB site was measured by ddPCR. n = 3. (D) PE efficiency of insertion of the att site and integration efficiency of a donor plasmid by evolved Bxb1 integrases in unmodified HEK293 cells was measured by ddPCR. n ≥ 4. Data are shown as mean + s.d. (*) P value < 0.05 derived from a Student's two-tailed t-test if significantly higher than the pre-evolved integrase control (shown in bold on the left).

Figure 5.

Bxb1 integrase combination mutants. (A) Five beneficial mutations were selected from IntePACE for further characterization. Insertion efficiency measured by ddPCR of integrases with all combinations of one to five mutations was compared in a clonal HEK293 cell line containing a Bxb1 attB site at ROSA26. n = 4. (B) PE efficiency of Bxb1 attB site insertion and integration efficiency of a donor plasmid was measured by ddPCR to compare combination mutants in unmodified HEK293 cells. n = 4. Data are shown as mean + s.d. (*) P value < 0.05 derived from a Student's two-tailed t-test if significantly higher than the pre-evolved integrase control. (C) Alphafold model of Bxb1 integrase. Bxb1 integrase domains defined by Ghosh et al. (52). Pymol (69) was used to align with LI prophage integrase (pdb 6dnw) co-crystalized with attP half-site DNA (70). (D) Alphafold model of Bxb1 integrase dimer aligned with LI prophage integrase to visualize DNA. The five mutations selected for combination variants are shown in red and purple.

Figure 6.

Insertion of large cargos by evolved Bxb1 integrase with PE. (A–E) PE efficiency of insertion of the att site and integrase insertion efficiency from combination mutants (Figure 5A) of a 6.6 kb donor plasmid during optimized transfections of (A) HEK293T, (B) HeLa, (C) K562, (D) U2OS and (E) HDFa cells. Cells were sorted three days post transfection for DsRed expression to exclude non-transfected cells before analysis by ddPCR. n = 3. (F) Schematic depicting delivery of therapeutic cargo to HEK293T cells. Cells were transfected with PE components, integrase helper plasmid, and a 15.6 kb vWF donor plasmid. Cells were sorted for DsRed expression 3 days post transfection to exclude non-transfected cells. Half the cells were pelleted for ddPCR detection of vWF donor integration at ROSA26 (G) and the remaining cells were cultured for an additional 11 days to minimize transient expression of the therapeutic transgene. n = 3. (H) Detection of secreted vWF in culture media by ELISA. Data are shown as mean + s.d. (*) P value < 0.05 derived from a Student's two-tailed t-test if significantly higher than the pre-evolved integrase control (shown in bold on the left).