KEY TAKEAWAYS
- Cattle cells with exposure to the CRISPR-Cas9 system did not show any significant differences in DNA variants and protein types when compared with those without exposure to the CRISPR-Cas9 system.
- The biosafety concern of the CRISPR-Cas9 system is low because the created CRISPR system was designed to make a cleavage on target bacterial genomes but not on cattle genomes, and to be delivered into bacterial cells but not into cattle cells.
- The nucleotide sequence and length of 20 bp for the gRNA currently used to target stx genes provided sufficient specificity.
BACKGROUND
The CRISPR-Cas9 system has emerged as a programmable and versatile tool for precise gene editing purposes. In addition to gene editing, the CRISPR-Cas9 system can be developed as “antimicrobials” to kill target bacteria. In previous studies, a CRISPR-Cas9 targeted killing system was developed with a guide RNA designed to specifically recognize the Shiga-toxic genes (stx1 and stx2). Delivery of this system into bacterial cells could effectively kill Shiga-toxin producing E. coli (STEC). This current study was conducted to estimate potential biosafety risk associated with a CRISPR-Cas9 system when applied to kill STEC cells in a bovine cell line model system using next generation sequencing (NGS) and proteomic analyses.
Methodology
A bovine cell line was cultured to reach 90% confluence. Then the bovine cells were subjected to one of four treatments: 1) bovine cell control: without any CRISPR treatment; 2) CRISPR/gRNA: treated with phages that carry the CRISPR system with the guide RNA targeting stx genes; 3) CRISPR+O157: treated with phages that carry the CRISPR system but without the guide RNA targeting stx genes and E. coli O157:H7 strain Sakai cells; and 4) CRISPR/gRNA+O157: treated with phages that carry the CRISPR system with the guide RNA targeting stx genes and E. coli O157:H7 strain Sakai cells. Each treatment was conducted in four replicates for a total of 16 samples. After application of treatments, bovine cells from each sample were collected and divided into two portions: half for whole genome sequencing and half for protein analysis. Whole genome DNA from each sample was extracted, purified, and sent to Novogene Bioinformatics Technology for library construction and NGS. Raw reads were subjected to quality control (QC) procedures to remove unusable reads. Clean reads after QC were aligned to the bovine reference genome using Burrows-Wheeler Aligner (BWA) with default parameters. Based on the mapping results, SAMtools was used to detect individual SNP/InDel variants, and ANNOVAR was used for functional annotation of the detected variants. Protein in each sample was extracted and analyzed by liquid chromatography coupled with tandem mass spectrometry. The resulting data were searched against the UniProt bovine protein sequence database to identify proteins. Relative quantitation between the control and treated cell cultures was performed using a label-free approach and determined using spectral counting (SpC). A one-way ANOVA or a student’s t-test with Benjamini-Hochberg correction was 55 applied to determine protein species that were significantly different in abundance between the control and CRISPR treatments.
Findings
A total of 1,078 Gb of output with 3,593.12 million paired end reads (150 bp) were obtained for all 16 samples after NGS. After QC, a total of 1,073 Gb of clean data output were obtained for all 16 samples. The average output for the four replicates within the control, CRISPR/gRNA, CRISPR+O157, and CRISPR/gRNA+O157 treatments was 59.1, 72.3, 72.3, and 65.9 Gb, respectively. The mapping rate of each sample ranged from 99.48% to 99.75%, and the 1X coverage ranged from 97.98% to 98.67%. A total of 94,796,157 SNPs and 11,949,421 InDels were identified in all 16 samples when compared with the reference genome. Of the total number of functional SNPs, 62.43% were identified to have functions in intergenic regions (regions between genes), 36.77% in intronic regions (non-coding sequences of genes), and 0.80% in exonic regions (coding sequences of genes). Of the total number of functional InDels, 62.40% had functions in intergenic regions, 37.45% in intronic regions, and 0.14% in exonic regions. The number of SNPs and InDels within each functional class (e.g., stop loss, stop gain, frameshift insertion/deletion, non-frameshift insertion/deletion, at slicing sites, and upstream or downstream from transcription termination sites) showed no significant differences (P > 0.05) among the control and the three CRISPR treatments.
For protein abundance analysis, none of the total 1,803 proteins identified from the 16 samples in the experiment were significantly different among the control and CRISPR treatments when Benjamini-Hochberg correction was considered in our proteomic analysis. When B-H correction is not considered, 107 proteins had expression levels different from the control (P < 0.05) depending on the CRISPR treatments being compared. A total of 61 proteins had higher abundance levels than the control. A total of 46 proteins in all three CRISPR treatments had lower abundance levels than the control. The fact that these 107 proteins were significantly different or not when Benjamini-Hochberg correction was considered or not, respectively, indicates that even if there might be changes in protein expression levels, all the changes were within a very modest range that can be ignored when Benjamini-Hochberg correction was introduced.
Implications
Results from this study will provide insights on how to further improve approaches or develop criteria on biosafety evaluation of the CRISPR-Cas9 system. In addition to coding regions, continued analysis of the variants in non-coding regions will occur. Results will provide the beef industry with biosafety information regarding application of CRISPR as an alternative to antibiotics in cattle production.