Bypassing GMO regulations with CRISPR gene editing

To the Editor:

The US Department of Agriculture (USDA) recently announced that regulation of a CRISPR-Cas9 gene-edited mushroom, in which a polyphenol oxidase (PPO) gene had been mutated to avoid (or delay) browning, fell outside of genetically modified organism (GMO) legislation¹. However, we remain concerned that the approved mushroom may still contain tiny fragments of foreign DNA in its genome. If foreign DNA is present after CRISPR-Cas9 editing, regulatory approval would be required under current GMO legislation.

In a letter (available in ref. 1 and at https://www.aphis.usda.gov/aphis/home/) submitted to the USDA, Yinong Yang, a researcher at Pennsylvania State University, outlined how the gene-edited mushroom was produced. Mushroom protoplasts (fungal cells with no cell wall) were transfected with plasmids encoding Cas9 and a guide RNA (gRNA) specific for a PPO gene. Transfected protoplasts were regenerated to produce mushrooms that contained small deletions (1–14 base pairs (bp)) in the PPO gene. The researchers analyzed the genome of edited fungi using PCR and Southern blot analysis and found no evidence of foreign DNA in the genome. In response, the USDA decided that the gene-edited mushroom fell outside of GMO regulations and that the US government had no authority to regulate this product.

We wish to point out that gene disruption by transient transfection of cells with plasmids encoding Cas9 and gRNAs can result in cells or organisms that contain small portions of foreign DNA (up to hundreds of base pairs) that are derived from the introduced plasmids. We showed that human cells transfected with Cas9 and gRNA plasmids often contain small insertions (58–280 bp) at off-target sites². One hypothesis is that introduced plasmids are fragmented in cells, and the resulting small DNA fragments are recombined into nuclease cleavage sites. Alternatively, plasmid DNA might be used as a template by a cellular DNA polymerase that functions in DNA double-strand break repair. Unlike small insertions at on-target sites, any insertions at off-target sites can only reliably be detected using whole genome sequencing. This is because small insertions of up to several hundred base pairs cannot be detected by PCR or Southern blot analyses.

To evaluate whether plasmid-derived DNA sequences are captured at Cas9 target sites in plant cells, we transfected Arabidopsis protoplasts with Cas9 and gRNA plasmids and analyzed edited protoplast DNA using targeted amplicon sequencing (Supplementary Methods and Fig. 1). Plasmid-derived sequences were found at the target site, regardless of the amount of plasmids used. The frequencies of these insertions were low (0.06–0.14% of total mutations) but are likely underestimated, because insertions of >50-bp sequences were excluded from our deep-sequencing analysis. Insertions at genome-wide off-target sites, which are difficult to identify unless one performs Digenome-seq³ or other unbiased off-target profiling methods, cannot be detected by targeted amplicon sequencing. As expected, no plasmid DNA insertions were detected at the target site, when we used the Cas9 ribonucleoproteins (RNPs)² that comprise recombinant Cas9 protein and an in-vitro-transcribed gRNA.

Figure 1: Insertions of Cas9 and gRNA plasmid-derived DNA sequences at the genomic target site in transiently transfected Arabidopsis protoplasts.

(a) T7 endonuclease I (T7EI) assay showing targeted mutagenesis in PHYB in protoplasts. Arabidopsis protoplasts were transfected by variable amounts of Cas9 and gRNA plasmids. Arrows indicate the expected positions of DNA bands cleaved by T7EI. Mutation frequencies (indels (%)) were measured by targeted amplicon sequencing. (b) Mutant DNA sequences containing plasmid-derived DNA sequences. The numbers of inserted nucleotides are indicated. Inserted sequences are underlined. Origins of inserted sequences are indicated in parentheses. The protospacer-adjacent motif is shown in bold. WT, wild type.

We do not believe that the presence of small insertions derived from Cas9 (or Cas9 variants and Cpf1) and guide RNA plasmids in a plant genome poses any risk to human health or the environment. However, if foreign DNA sequences were to be found in the genome of a gene-edited crop that becomes commercially available in the future, it would set back both the regulatory authority and the industry.

We propose that whole genome sequencing should be carried out to make sure that gene-edited crops do not contain plasmid-derived DNA sequences. An alternative to whole genome sequencing is to breed crop lines in order to remove any unwanted insertions, although this might be time-consuming and labor-intensive. Genome editing in crops could be carried out using preassembled Cas9 (refs. 2,4) or Cpf1 (refs. 5,6) RNPs, rather than plasmids encoding these components. This would reduce, or eliminate, any potential for unwanted foreign DNA in a crop genome.