CRISPR/Cas9-generated p47phox-deficient cell line for Chronic Granulomatous Disease gene therapy vector development

Development of gene therapy vectors requires cellular models reflecting the genetic background of a disease thus allowing for robust preclinical vector testing. For human p47phox-deficient chronic granulomatous disease (CGD) vector testing we generated a cellular model using clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 to introduce a GT-dinucleotide deletion (ΔGT) mutation in p47phox encoding NCF1 gene in the human acute myeloid leukemia PLB-985 cell line. CGD is a group of hereditary immunodeficiencies characterized by impaired respiratory burst activity in phagocytes due to a defective phagocytic nicotinamide adenine dinucleotide phosphate (NADPH) oxidase. In Western countries autosomal-recessive p47phox-subunit deficiency represents the second largest CGD patient cohort with unique genetics, as the vast majority of p47phox CGD patients carries ΔGT deletion in exon two of the NCF1 gene. The established PLB-985 NCF1 ΔGT cell line reflects the most frequent form of p47phox-deficient CGD genetically and functionally. It can be differentiated to granulocytes efficiently, what creates an attractive alternative to currently used iPSC models for rapid testing of novel gene therapy approaches.

The BsrG1 analysis of one nucleofected PLB-985 clone displayed the same band pattern as of a Δ GT p47 phox -deficient CGD patient ( Fig. 1D and Supplementary Fig. S1), suggesting a Cas9-mediated disruption of the BsrG1 site in both NCF1 alleles (homozygous mutation efficiency 4.5%). The BsrG1 digestion analysis of the remaining clones suggested heterozygosity in these clones ( Supplementary Fig. S1).
To confirm the presence of the Δ GT in mutated NCF1, genomic DNA of WT and pPX458-NCF1-treated PLB-985 cells were used for PCR co-amplification of the NCF1, NCF1B, and NCF1C (Fwd1, Rev2 primers, Fig. 1B). The barcoded PCR products were analyzed by single molecule real-time sequencing (SMRT-seq) for the presence of GT-dinucleotide, Δ GT, as well as for the presence of one copy of 20-nucleotide (20-nt) intronic repeat (1 × 20nt) derived from NCF1 or two repeats (2 × 20nt) derived from NCF1B or NCF1C (Fig. 1B). SMRT-seq analysis showed that PLB-985 WT and PLB-985 NCF1 Δ GT cell lines displayed almost identical percentage (61%) of reads with one or two copies of the 20-nucleotide repeat (Fig. 1E), indicating that all three NCF1 loci were co-amplified with comparable efficiencies. Moreover, all SMRT-seq reads from PLB-985 NCF1 Δ GT displayed the Δ GT, while in PLB-985 WT 37.4% harbored the GT-dinucleotide sequence, as expected from the two NCF1 alleles.
Potential off-target sites of utilized sgRNA were predicted (Supplementary Table S2) and DNA sequences of 14 sites with highest scores were analyzed for presence of insertions or deletions (indels) by Surveyor assay. In none of the potential off-target sites indels were observed (Supplemetary Fig. S2).
The iPSC model of Δ GT p47 phox -deficient CGD 11 can be differentiated to monocytes, macrophages 11 and granulocytes 15,16 . Nevertheless, the use of iPSCs is laborious, and time-consuming, as phagocytic differentiation requires long culture periods of 35-43 days 15,17 requiring large amounts of cytokines and continuous surveillance of the culture. In contrast full granulocytic differentiation of PLB-985 NCF1 Δ GT cell line takes 7 days, requires only fetal calf serum (FCS) restriction (5%) and supplementation with 0.5% N,N-dimethylformamide (DMF). Furthermore, the differentiation of neutrophils can be easily assessed by flow cytometry (Supplementary Fig. S3).
The function of the NADPH oxidase complex within PLB-985 NCF1 Δ GT cells was reconstituted by transduction with two γ -retroviral vectors encoding p47 phox protein (Fig. 2D). Transgene expression was driven either by the ubiquitously active spleen focus-forming virus (SFFV) promoter or by the myelospecific microRNA 223 (mir223) promoter 18 . The expression cassette was composed of truncated Low-affinity nerve growth factor receptor (Δ LNGFR) and p47 phox linked by the 2 A self-cleaving peptide 19 . Transduction efficiency reached 8.9% and 11.6% for mir223, and SFFV-driven γ -retroviral vectors (Fig. 2E). Functional reconstruction of the NADPH oxidase complex activity in transduced and differentiated PLB-985 NCF1 Δ GT cells was assessed by NBT assay (Fig. 2F) and chemiluminescence assay (Fig. 2G). Percentage of NBT-positive cells reached 13.4 ± 2.1% with the mir223, and 11.8 ± 1.8% with the SFFV-driven Δ LNGFR-2A-p47 phox expression constructs ( Fig. 2F and Supplementary Fig. S5). Chemiluminescence assay of transduced and differentiated PLB-985 NCF1 Δ GT cells confirmed reconstitution of ROS production. These results clearly indicate that the defect of the NADPH oxidase complex caused by the Δ GT mutation within NCF1 in the PLB-985 NCF1 Δ GT cell line can be corrected.
In summary, the CRISPR/Cas9-generated PLB-985 NCF1 Δ GT cell line fully recapitulates the genetic background and functional NADPH oxidase defect found in majority of p47 phox -deficient CGD patients, can be corrected genetically and differentiated into functional neutrophils. The PLB-985 NCF1 Δ GT cell line therefore represents a promising cost-effective tool for rapid gene therapy vector testing.

Materials and Methods
Reagents and antibodies. Single stranded DNA was purchased from Microsynth (Balgach, Switzerland).
Cell culture and differentiation. We utilized PLB-985 cell line 12 , a subclone of HL-60 21,22 , as it is capable of granulocytic differentiation. Cells were cultured in RPMI 1640 medium supplemented with 10% (vol/vol) FCS, and 1% (vol/vol) penicillin/streptomycin in a humidified incubator at 37 °C and 5% CO 2 . Granulocytic differentiation of logarithmically growing PLB-985 cells at density of 0.8·10 6 cells/mL was induced by reduction of the FCS content to 5%, and supplementation with 0.5% (vol/vol) DMF. After 3 days, an equivalent of initial volume of differentiating medium was added and the differentiation continued until day 7.  (Qiagen, Hombrechtikon, Switzerland). Parallel PCR co-amplification of the NCF1, NCF1B, and NCF1C was performed using published primers 23 (Fwd1 and Rev1, Fig. 1B). For PCR reaction using Fwd1/ Rev2 primers Fwd1 primer was barcoded for each template producing products of 417 bp from NCF1 and of 435 bp from NCF1B and NCF1C. PCR products were pooled for SMRT sequencing and analyzed according to barcode identities.
Single molecule real-time (SMRT) sequencing. The NCF1 gene and NCF1B and NCF1C pseudogenes were PCR amplified in parallel utilizing primers Fwd1 and Rev2 (Fig. 1B) 23 . The primer Fwd1 was barcoded for each FACS sorted clone and obtained PCR products were of 417 bp for NCF1 and 435 bp for NCF1B and NCF1C templates. PCR reaction consisted of initial denaturation (95 °C, 3 minutes), 40 cycles of denaturation (95 °C, 30 seconds), annealing (60 °C, 30 seconds), elongation (72 °C, 30 seconds), and a final elongation step (72 °C, 3 minutes). PCR products were gel purified using QIAquick Gel Purification Kit (Qiagen). 10-20 ng of gel-purified PCR products were pooled and analyzed by SMRT sequencing by Functional Genomics Center Scientific RepoRts | 7:44187 | DOI: 10.1038/srep44187 Zurich, ETH/University of Zurich, Zurich, Switzerland. Briefly, DNA amplicon librabry was produced using DNA Template Prep Kit 1.0 (Pacific Biosciences, Menlo Park, California, United States). The input DNA concentration and quality was measuered using Qubit Fluorometer dsDNA Broad Range assay (Life Technologies, Zug, Switzerland) and Bioanalyzer 2100 12 K DNA Chip assay (Agilent Technologies AG, Basel, Switzerland). The SMRT bell template was prepared by end-repair of the DNA amplicons, followed by blunt-end ligation of overhang adapters and exonuclease treatment. The SMRT bell template was complexed with polymerase using P6 DNA/Polymerase Binding Kit 2.0 (Pacific Biosciences) according to the manufacturer's instructions. The samples were sequenced using Pacific Biosciences RS2 platform. From the raw reads, high quality circular consensus reads (CCS) were obtained through the Reads Of Insert protocol available in the SMRT Analysis suite (Pacific Biosystems). CCS reads were then de-multiplexed by exact matching of the sample barcodes, starting at the base immediately preceeding the Fwd1 primer sequence. Reads with a length between 400 and 450 nucleotides were retained and matched against the NCF1 reference sequence using blast 24 . Surveyor Assay. Off-target sites were predicted utilizing the Optimized CRISPR Design (http://crispr.mit. edu, F. Zhang laboratory, MIT 2015). Loci of predicted sites (see Supplementary Table S2) were PCR amplified from gDNA of PLB-985 WT and PLB-985 NCF1 Δ GT cells. Corresponding PCR amplification products were mixed in a ratio of 1:1, while PCR amplification product of PLB-985 WT was used as a control. The samples were denatured at 95 °C for 10 minutes, slowly renatured, and digested using Surveyor ® Mutation Detection Kit For Standard Gel Electrophoresis (Integrated DNA Technologies, Leuven, Belgium) according to the manufacturer's instructions.
Data Availability. All SMRT sequencing data can be obtained upon request.
Nitroblue-tetrazolium (NBT) test. Differentiated PLB-985 WT, PLB-985 X-CGD, and PLB-985 NCF1 Δ GT cells, as well as transduced PLB-985 NCF1 Δ GT cells were incubated in growth medium supplemented with 100 μ g/mL NBT in the presence of 200 ng/mL PMA at 37 °C and 5% CO 2 for 30 minutes. Subsequently, the cells were fixed in 1% (vol/vol) formaldehyde. 250 cells per cytospin slide were analyzed manually for NBT activity using a Leica DM IL Fluo light microscope equipped with a DFC420 digital camera and LEICA application suite acquisition software (Leica Microsystems AG, Glattbrugg, Switzerland).