Patricia F. Fitzpatrick Dimond Ph.D. Technical Editor of Clinical OMICs President of BioInsight Communications
Researchers Seek Ways to Minimize Off-Target Effects
Elegant in its simplicity and easily accessible to most laboratories, CRISPR/Cas9 technology has been widely adopted as a DNA-editing tool in research labs, and a group of investigators in China has begun to use it in a clinical trial to disable the PD-1 gene in lung cancer patients to slow or stop the cancer’s progression. But the technology’s off-target effects have raised serious concerns about its ultimate applicability to human therapies.
CRISPR proponents say it’s less clunky to use that other recently emerged genome-editing technologies, including zinc-finger nucleases (ZFNs) and transcription activator–like effector nucleases (TALENs). ZFNs and TALENs work by using a strategy of tethering endonuclease catalytic domains to modular DNA-binding proteins for inducing targeted DNA double-stranded breaks (DSBs) at specific genomic loci.
In contrast, Cas9 is a nuclease guided by small RNAs through Watson-Crick base pairing with target DNA, described by F.A. Ran et al. as a system that is “markedly easier to design, highly specific, efficient, and well-suited for high throughput and multiplexed gene editing for a variety of cell types and organisms.”
CRISPR/Cas9 nucleases consist of the Cas9 bacterial enzyme and a short, 20-nucleotide RNA molecule that matches the desired target DNA sequence. In addition to the RNA/DNA match, the Cas9 enzyme must recognize a specific nucleotide sequence called a protospacer adjacent motif (PAM) adjacent to the target DNA.
The most commonly used form of Cas9, derived from the bacteria Streptococcus pyogenes and known as SpCas9, recognizes PAM sequences in which any nucleotide is followed by two guanine DNA bases. This limits the DNA sequences that can be targeted using SpCas9 only to those that include two sequential guanines.
Off Target Effects
But almost as soon as the technology was introduced, scientists raised concerns about off target effects. Said Xiao-Hui Zhang et al. of the College of Veterinary Medicine, South China Agricultural University and coauthors at MIT in a 2015 Molecular Therapy—Nucleic Acids article, “The high frequency of off-target activity (≥50%)—RGEN (RNA-guided endonuclease)-induced mutations at sites other than the intended on-target site is one major concern, especially for CRISPR technology therapeutic and clinical applications.”
The growth of any new technology, the authors note, including CRISPR/Cas9, demands progressive enhancement. And while research in CRISPR/Cas9 has made “huge strides in the evolution of gene editing,” it and other RGENs, which include ZFNs and TALENs, have more severe off-target effects than other nucleases due to their inherent structure and mechanism.
At an American Society of Hematology workshop on genome editing, CRISPR pioneer J. Keith Joung, M.D., Ph.D., of Massachusetts General Hospital said, “In the early days of this field, algorithms were generated to predict off-target effects and [made] available on the web miss a fair number of off-target effects. He added, “These tools are used in a lot of papers, but they really aren’t very good at predicting where there will be off-target effects,” according to STAT.
Observations in the recent literature have raised more alarms among CRISPR cognoscenti, all of whom would agree than technical improvements are needed. In one of the most blogged-about papers on CRISPR, “Unexpected mutations after CRISPR–Cas9 editing in vivo,” Kellie A. Schaefer and colleagues at Stanford concluded that “More work may be needed to increase the fidelity of CRISPR/Cas9 with regard to off-target mutation generation before the CRISPR platform can be used without risk, especially in the clinical setting.”
The authors had, they reported in a 2016 study by W.H. Wu et al. in Molecular Therapy, used CRISPR/Cas9 for sight restoration in blind rd1 mice by correcting a mutation in the Pde6b gene. Mice homozygous for the rd mutation have hereditary retinal degeneration and have been considered a model for human retinitis pigmentosa.
Citing persistent concerns about secondary mutations in regions not targeted by a single guide RNA (sgRNA)—concerns also expressed by a number of other scientists—Kellie A. Schaefer, Ph.D., at Stanford University and colleagues at Howard Hughes and Massachusetts General Hospital performed whole genome sequencing (WGS) on DNA isolated from two CRISPR-repaired mice (F03 and F05) and one uncorrected control.
CRISPR/Cas9-treated mice were sequenced at an average depth of 50×, and the control to 30× to identify all off target mutations. The sequencing the authors said identified an “unexpectedly high” number of single nucleotide variations (SNVs), contrary to the widely accepted assumption that CRISPR causes mutations mostly at regions homologous to the sgRNA.
CRISPR’s penchant for promiscuous behavior has spawned an entirely new research field focused on fixing it. Patents have already been filed on the fixes and the developers believe that these advances will incrementally enable more reliable CRISPR performance. Most efforts have concentrated on modifying CRISPR nuclease Cas9 using structure-design based changes in the enzyme, chemical modifications, and amino acid substitutions at critical sights to better predict and control its function.
Writing in Nature in 2015, Benjamin P. Kleinstiver, Ph.D., of the Molecular Pathology Unit and Center for Cancer Research at the Massachusetts General Hospital and colleagues noted that although CRISPR/Cas9 nucleases are widely used for genome editing, the range of sequences that Cas9 can recognize is constrained by the need for a specific PAM.
The investigators reported that they could modify Cas9 to recognize alternative PAM sequences using structural information, bacterial selection-based directed evolution, and combinatorial design. The altered PAM specificity variants, they said, could edit endogenous gene sites in zebrafish and human cells that are not targetable by wild-type SpCas9. Further, they said, the variants’ genome-wide specificities are comparable to wild-type SpCas9 as judged by GUIDE-seq analysis and establish the feasibility of engineering a wide range of Cas9s with altered and improved PAM specificities.
Kleinstiver and other investigators working in Joung’s also developed the unique endonuclease SpCas9-HF1, which they describe as a high-fidelity enzyme variant with alterations designed to reduce non-specific DNA contacts. The scientists hypothesized that reducing iCas9 and the target DNA interactions might help eliminate off-target effects while still retaining the desired on-target interaction.
Since certain portions of the Cas9 nuclease can itself interact with the backbone of the target DNA molecule, the team altered four of these Cas9-mediated contacts by replacing the long amino acid side-chains that bind to the DNA backbone with shorter ones that could not bind.
SpCas9-HF1 retained on-target activities comparable to wild-type SpCas9 with >85% of single-guide RNAs (sgRNAs) tested in human cells. Notably, with sgRNAs targeted to standard nonrepetitive sequences, SpCas9-HF1 rendered all or nearly all off-target events undetectable by genome-wide break capture and targeted sequencing methods.
Jiang and Doudna pointed out in a 2017 piece in Annual Review of Biophysics how Cas9 locates specific 20-base-pair (bp) target sequences within the genomes that are millions to billions of base pairs long and, subsequently, how it induces sequence-specific double-stranded DNA (dsDNA) cleavage remain critical questions, not just in CRISPR biology, but in the efforts to develop more precise and efficient Cas9-based tools.
Molecular insights from biochemical and structural studies such as those describe above will provide a framework for rational engineering aimed at altering catalytic function, guide RNA specificity and PAM requirements, and reducing off-target activity for the development of Cas9-based therapies against genetic diseases.
A recent study in BioRxiv questioned the methodology, data interpretation, and conclusions of a prominent study in Nature Methods by Schaefer et al. The BioRxiv paper by Wilson et al. noted that the Schaefer et al. study had too small a sample size (n=2 for treated mice and n=1 for untreated mice) for the researchers to make any valid scientific observations. Additionally, Wilson and team argued that the observed genomic differences between the control and experimental animals could have existed prior to edits made with CRISPR/Cas9. Shaefer's team also pointed out that the specific gRNA used in the experiment is prone to off-target specificity: “While perhaps acceptable for research purposes, a gRNA with a predicted high off-target profile would be immediately excluded as a therapeutic candidate.” In summary, Wilson's team concluded that the mutations highlighted in the Schaefer paper could not be attributed to CRISPR.
Intesrestingly, a coauthor of the Wilson paper in BioRxiv is George Church, who has a vested interest in the success of CRISPR technology. As a comment attached to the open-source paper revealed, Church is a founding scientific advisor at Editas Medicine, which has various CRISPR projects in the pipeline. As of June 23, 2017, the paper did not include a conflict of interest statement.