In the process of gene editing, scientists try to cure disease by using molecular tools to insert, delete, or change DNA in a precise way. The rise of breakthrough gene editing technologies has revolutionized our ability to study and treat diseases.
All gene-editing agents must accomplish two goals: 1) find a DNA sequence, and 2) change that target DNA’s sequence.
Almost all gene-editing agents accomplish the first task by using one of the three major DNA-targeting technologies: zinc finger proteins, TALE proteins, and CRISPR proteins. All three technologies are proteins that bind DNA. Unlike zinc finger and TALE proteins, which use protein sequences to find DNA sequences, CRISPR uses an RNA sequence called the guide RNA. Using RNA, CRISPR can target DNA much more easily than can zinc fingers or TALE proteins. As a result, researchers have adopted CRISPR overwhelmingly to target DNA, as shown below.
In addition to finding a target DNA sequence, editing agents must change the structure of the target DNA molecule. Gene-editing agents can change human DNA cells in at least three ways: DNA cutting, base editing, and prime editing.
Three major classes of nucleases have been used to cut DNA: ZFNs (zinc finger nucleases), TALENs (TALE nucleases), and CRISPR-Cas9 nucleases. All of them cut the DNA double helix into two pieces. Having evolved in nature, the CRISPR-Cas9 nuclease uses viruses to cut DNA, disrupting viral genes and ending the viral life cycle. Similarly, the other nucleases cut a DNA sequence and disrupt the target gene. Nucleases are like scissors that can cut DNA but cannot change one DNA sequence into another specific sequence.
Base editors can use zinc finger proteins, TALE proteins, or CRISPR to find a target DNA sequence. Instead of cutting the target DNA, they change one DNA letter, such as an A, to a specific DNA letter, such as a G. Because base editors change DNA sequences precisely and predictably, they have been compared to pencils that change DNA misspellings one letter at a time. Importantly, unlike nucleases, base editors do not cut the DNA double helix.
Prime editing uses CRISPR to find a specified DNA sequence. Instead of cutting the target DNA, or changing one letter of DNA, prime editors replace an entire segment of DNA with a segment of the researcher’s choosing. Since prime editors perform search-and-replace on DNA, they have been compared to DNA word processors. Like base editors, prime editors do not normally cut the DNA double helix.
Many readers may be familiar with the fundamental differences among nucleases (~scissors), base editors (~pencils), and prime editors (~word processors) but may not understand how human DNA sequences change after the various treatments. Because the outcomes can be so different, each is suited to different applications.
Fortunately, the raw DNA sequencing files are transparent in many gene-editing research publications, and public tools like CRISPResso2 can decode them, helping researchers understand how nucleases, base editors, and prime editors change DNA sequences. To explain the outcomes, we conducted CRISPResso2 analyses on DNA sequencing files from representative experiments and will discuss the outcomes below.
CRISPR Cas-9 Nuclease Editing and Its Unintended Consequences
CRISPR Cas-9 nuclease results in double-strand breaks in DNA, cutting a chromosome into two DNA molecules. These double-strand breaks stimulate the cell to heal the break by rejoining the ends of the broken chromosome. If the rejoining occurs perfectly, then the nuclease can recut the target sequence. The rejoining process can be imperfect, so rejoining and recutting recurs until the process has made enough errors in the location of the DNA cut that the nuclease no longer recognizes the altered DNA sequence as a target. The result is that nucleases such as CRISPR Cas9 result in “indels”––mixtures of insertions and deletions––at the site of DNA repair. Since indels typically disrupt the function of a gene, a nuclease can disrupt genes effectively. Unfortunately, indels––like the insertion of two extra letters or deletion of five DNA letters––caused by double-strand breaks are a function of biology and cannot yet be controlled by gene-editing agents.
In some cases, gene disruption is the goal and can lead to therapeutic outcomes. Researchers at Intellia Therapeutics (NTLA) and Regeneron Pharmaceuticals (REGN), for example, made history at the Peripheral Nerve Society Conference last year with the first data on CRISPR Cas9-based in-vivo gene editing therapy in patients with Hereditary Transthyretin Amyloidosis (hATTR) polyneuropathy (PN). hATTR is a disease in which amyloid proteins build up and cause multiple organ failures, typically the heart and nerves. Notably, researchers demonstrated that disrupting the TTR gene with CRISPR-Cas9 delivered a one-time treatment that reduced sustained serum TTR, obviating the need for chronic therapy. Although the sample size was small, sustained serum TTR reductions suggest the possibility of a one-time therapy to cure hATTRv-PN patients.
Furthermore, after administering a CRISPR Cas-9 nuclease to 75 Sickle Cell and Beta-Thalassemia patients, researchers at Vertex Pharmaceuticals (VRTX) and CRISPR Therapeutics (CRSP) demonstrated that disrupting a gene that silences a backup set of hemoglobin genes effectively treated these blood diseases.
Researchers have shown that in addition to making indels that disrupt target genes, double-strand breaks can cause undesired cellular consequences, including chromosomal abnormalities, such as translocations and large deletions. In fact, the US Food and Drug Administration (FDA) has placed holds on some clinical trials in response to such adverse events.
The CRISPResso2 tool tabulates DNA sequence results in a way that allows us to measure the editing outcomes from human cells treated with gene-editing agents, as shown below. The starting DNA sequence of the cells before treatment is at the top of the chart. The commonly observed outcome following treatment of a population of cells is on the first row, and less frequent outcomes on subsequent rows. The frequency of each outcome is shown on the right. Substitutions are in bold, insertions are in red boxes, and deletions are dots. Shown below is the CRISPResso2 analysis of the DNA sequences treated with a CRISPR-Cas9 nuclease.
The analysis highlights how effectively nucleases disrupt genes. Treating human cells with DNA-cutting nucleases typically results in a mixture of disrupted DNA sequences, each with small insertions (boxed in red) or deletions (grey boxes with dashes) near the cut site. In this example, only 12% of the treated cells in the target sequence did not change. In other words, after treatment, 88% of the cells contained one of dozens of indels in the targeted gene. No indel dominated the outcomes, which is the reason nucleases can disrupt genes but cannot make precise corrections to disease-causing DNA errors. In fact, each disruption can have different biological consequences, heightening the uncertainty and risks.
Emerging Gene-Editing Methods: Base Editing and Prime Editing
When the goal is not to disrupt a gene but to induce a specific change within the DNA, two innovative methods are proving useful––base editing and prime editing. Because these technologies change the target DNA sequence directly into a different sequence without cutting the double helix, they avoid the large uncontrolled mixtures of DNA indel outcomes that result from treating cells with nucleases. Because they do not cut DNA, base editors and prime editors also can sidestep undesired consequences like chromosomal abnormalities caused by CRISPR-Cas9 and other nucleases.
The table below compares the three gene editing methods along four dimensions: Size, PAM Dependence, breaks, and addressable clinical variants.
Base editing is used to correct disease-causing mutations caused by single letter misspellings. Single letter misspellings are the most common cause of genetic disease, accounting for roughly half of the known pathogenic DNA mutations.
A base editor converts one DNA letter, or base, to another by using a deaminase, an enzyme that performs chemistry on a DNA base. Base editors find the target DNA sequence using a programable DNA-binding, such as a zinc finger protein, a TALE protein, or a disabled CRISPR-Cas9 protein that can no longer cut the DNA double helix. Base editors nick only one strand instead of both strands of DNA, to which cells respond very differently. For perspective, human cells experience thousands of DNA nicks naturally every day, but rarely experience double-strand DNA cuts.
While CRISPR-Cas9 nuclease editing disrupts only a gene, base editing can disable or enable a specific genetic function by correcting a single “letter” in the genome. Base editing, therefore, can correct mutated DNA letters that cause thousands of genetic diseases. Beam Therapeutics (BEAM), for example, is working on treating Glycogen Storage Disease la (GSDIa), a disease in which glycogen does not metabolize into glucose. In pre-clinical studies, its researchers have shown that correcting the R83C mutation could cure GSDIa.
To understand the outcomes of treating a population of human cells with base editors, we used CRISPResso2 to analyze the DNA sequences that resulted from treating cells with a base editor programed to change two As to two Gs in PCSK9, a gene that controls the levels of “bad” cholesterol. In a clinical trial recently initiated through a partnership between Verve Therapeutics (VERV) and Beam Therapeutics (BEAM), a similar base editor treated patients with high cholesterol in an attempt to reduce the risk of heart disease, as shown below.
According to the CRISPResso-2 allele plot, the desired base edit was successful ~56% of the time, while ~36% of the time the original sequence was unchanged. In contrast to those treated with a nuclease, cells treated with the base editor showed a limited range of outcomes: rarely (3%) was a nearby A changed to a G, an outcome called “bystander editing,” or was only one of the two targeted As changed to a G (1.5%). Overall, after treatment with the base editors, 92% of the outcomes at the target site were either the desired edit (56%) or an unchanged sequence (36%).
Prime editing can accomplish all twelve possible base-to-base swaps, as well as small insertions and deletions. A prime editor uses an engineered reverse transcriptase enzyme, which writes DNA directly from an RNA template in a prime-editing guide RNA (pegRNA). Like CRISPR-targeted base editors, prime editors use a disabled CRISPR Cas-9 protein that cannot make double-stranded DNA breaks.
Because of its versatility and ability to create many different edits precisely, prime editing has the potential to treat rare and complex diseases by correcting the majority of known pathogenic DNA mutations to normal DNA sequences. Scientists at the Hubrecht Institute, UMC Utrecht, and the Oncode Institute, for example, published a paper on the use of prime editing to correct the cystic fibrosis mutation in cultured human stem cells. Meanwhile, scientists at Yonsei University in Korea published a paper on the success of prime editing in ameliorating liver and genetic eye diseases in adult mice.
In the prime editing example below, scientists used a state-of-the-art prime editor in human cells to edit a G to a T at the PRNP locus in HeLa cells. This edit protects against prion diseases like Mad Cow Disease.
The intended prime edit occurred 32% of the time, but 64% of the time left the DNA sequence unchanged and 0.93% of the time resulted in an undesired byproduct. While its efficiency needs to improve, prime editing delivered the cleanest outcomes relative to the nuclease and base editing experiments that we analyzed.
No gene editing technique is perfect in all situations. The optimal gene-editing modality depends on the specific application, biological context, and potential trade-offs. That said, CRISPR-Cas9 nuclease treatment can result in dozens of uncontrolled indel outcomes in the effort to disrupt genes and does not result in the precise conversion of one DNA sequence to a specific sequence; its double-strand breaks also can cause unintended genetic consequences. Base editors make A-to-G, G-to-A, C-to-T, and T-to-C edits precisely and result in fewer indel or other editing outcomes but can cause bystander edits that may or may not have biological consequences. Prime editing seems to be the most versatile of the gene-editing technologies, resulting in few undesired outcomes at the expense of efficiency.
Understanding the risks and benefits associated with each gene-editing method is important. A careful examination of all the DNA sequence outcomes reveals how different gene-editing technologies can change target DNA, informing researchers of the best use cases for each. These breakthrough technologies aim to correct DNA mutations associated with thousands of genetic diseases––permanently. They are not treating or controlling symptoms of disease. Medicine has entered a new era.