A few years ago, someone asked me to give a layman explanation of CRISPR/Cas9 gene editing. They had heard about it in the evening news for several months by this point, but they never felt like they fully grasped what it did at the DNA level.
A word processing cut-and-paste metaphor was my immediate response, explaining that the complex effectively finds the defective form of a gene sequence and replaces it with a preferred form.
As CRISPR/Cas9 continued to evolve, so too did my metaphorical thinking; it was less cut-and-paste than find-and-replace.
Even though the gene-editing platform continues to be a hot topic, in both the public and scientific eyes, the technology itself continues to evolve as we see a transition from find-and-replace to something more akin to spellchecking.
From gene to base
For genetic disorders, the progression from small-molecule and biologics interventions to cell and gene therapies has been an effort to find deeper and longer-lasting treatments that—in their ultimate form—could be curative.
Some disorders have been more amenable to genetic exploration—e.g., stable monogenic disorders—whereas others have been more intractable—e.g., molecularly heterogeneous tumors.
The University of Victoria’s Francis Choy and Chloe Christensen work in the former category, focusing on mucopolysaccharidoses (MPSs), lysosomal storage disorders where the absence of a single metabolic enzyme leads to a buildup of toxic byproducts. Effectively, the cellular garbage disposal is broken, and the molecular trash overwhelms the cell.
“Enzyme replacement therapy [ERT] has really been the interest over the years for lysosomal diseases and MPS in particular,” Christensen explains, but this approach offers a variety of limitations, not the least of which is the inability of ERT to cross the blood-brain barrier (BBB).
Thus, for conditions like MPS IIIB (aka Sanfilippo syndrome), ERT can ameliorate the bodily impacts of the disorder, but not the neurological impacts.
“There is also the issue of ERT being a chronic therapy, something that needs to be used over time,” she continues. “Patients have to go back for repeat treatments, and for that reason, as well as because the MPSs are quite rare, ERT can be very expensive.”
To find a longer-lasting solution, Choy’s lab and others have looked at gene therapy as a potential solution, using vehicles like adeno-associated viruses (AAVs) to deliver the missing genes to cells where they would produce the deficient enzyme.
Because this genetic material doesn’t integrate into the patient’s DNA, however, even this is not a permanent solution. Over time, as infected cells divide, the AAV-delivered genes dilute out, requiring additional treatments.
“Even more recently, more interest has been in gene replacement therapies, where the gene can actually be knocked-in to the genome of the patient,” Christensen notes, pointing to the recent CHAMPIONS and EMPOWERS clinical trials led by Sangamo Therapeutics.
“They’ve taken the IDS and the IDUA genes—which are implicated in MPS I and MPS II, respectively—and delivered them using AAVs to patient hepatocytes,” she explains. “The idea is that if this is delivered alongside a zinc-finger nuclease (ZFN), which is like the first version of genome editors, that this gene could be knocked-in to a safe harbor locus and could produce the enzyme long term.”
What makes this approach particularly attractive for lysosomal storage diseases is a phenomenon called cross-correction, says Choy. When lysosomal enzymes are produced in the cell, they are secreted into circulation and ultimately interact with, and are taken up by, other cells via a ligand-receptor interaction. In MPSs, the ligand on the enzyme is typically mannose-6-phosphate. Thus, even if a cell is not gene-modified, it can benefit from its neighbors.
“The beauty of the concept of cross-correction is related to what we call the disease threshold,” offers Choy. “For example, in MPS I (Hurler’s syndrome), researchers have documented the threshold amount of normal enzyme to correct the major disease effects is quite low.”
Most MPS I patients have near-zero enzyme activity, he explains, but research has shown that as little as 1 to 5 percent of normal enzyme levels can treat most of the outcomes.
“Because of this low disease threshold for the enzyme correction level, cross-correction becomes the method of choice,” he states.
“This is particularly interesting to us because MPS IIIB disease has a neurological component,” adds Christensen, suggesting that they are applying similar genome editing ex vivo to patient-derived induced stem cells.
“Once edited for a patient-specific mutation, you can take those cells, differentiate them into neuronal and glial precursors, and deliver them back into the patient using intracerebral transplantation,” she remarks. “Then, hopefully, those precursors would turn into neurons and glia, be able to produce functional enzyme, and secrete the enzyme to be taken up by existing neurons in the patient brain.”
This is not to say that genome-editing platforms don’t have their challenges.
ZFNs, for one, are owned by Sangamo, and therefore can be expensive to acquire. As well, they can be difficult to bioengineer because of the protein-DNA interaction that guides their genomic site-specificity.
“The success recently, particularly with the CHAMPIONS and EMPOWERS studies, has resulted from the focus on this single locus, the safe-harbor locus in the albumin gene,” Christensen notes. “So, they were essentially able to just design one version of the zinc finger and use a different gene knock-in construct to insert into that particular location.”
She contrasts that with CRISPR/Cas9 genome editing, where targeting relies on RNA-DNA interactions, making them much easier to design and much less expensive to use.
The challenge with CRISPR/Cas9, she points out, comes with its induction of a double-strand DNA break (DSB) and its reliance on host cell DNA repair machinery to induce the genome modification.
If DSB repair occurs via non-homologous end-joining (NHEJ), the result is likely a gene knock-out, whereas if it occurs via homology-directed repair (HDR), the sequence of an endogenous homologue or introduced DNA fragment is used to specify how the DSB is fixed.
“This is particularly important in the context of off-target effects as well as when considering large numbers of simultaneous edits,” says Jamie Freeman, corporate innovation partner at Horizon Discovery. “In traditional approaches, there will be a level of chromosomal translocations between break points.”
“With a low number of target genes, these [off-target effects] can be mapped and any risks assessed,” he continues. “However, when there is also off-target guide RNA binding, these additional off-target break points must also be considered, together with the risk of unknown off-target effects.
“Equally, as the number of target genes increase in order to make more influential cell phenotypes, predicting this risk becomes exponentially more difficult.”
Christensen adds that CRISPR/Cas9 DSBs have also been shown to induce a p53 response in cells, which can lead to higher rates of cellular apoptosis than in untreated cells. Likewise, the reliance of genome-editing on the host DNA repair machinery can limit precisely what cells can be modified.
“Traditional technologies are capable of generating these edits in rapidly dividing cells, albeit at a relatively low efficiency due to the dependence on HDR,” says Freeman, “but this has limited functionality in non-dividing cells such as many primary cells, often the target of cell and gene therapies, where this pathway is often inactive.”
To circumvent many of these issues, Choy’s lab and many others have begun to explore the use of base editors, systems designed to precisely modify an individual base within a target sequence.
An evolution of the CRISPR/Cas9 system, base editors also use a sequence-targeting guide RNA and Cas9 enzyme, but rather than induce DSBs, the Cas9 endonuclease has been modified to only nick or not change the target DNA.
In addition to the Cas9 enzyme, however, the base editor also includes an adenine or cytosine deaminase moiety that catalyzes the base changes A > G or C > T, respectively.
Thus, explains Christensen, the base editor complex has all the machinery it requires to operate, reducing its reliance on HDR and therefore extending its use to non-dividing cells. And the absence of a DSB, abrogates p53 induction.
In a recent letter, Yi Zhang and colleagues at Zhengzhou University used an adenine base editor (ABE) to alter the effects of PD-1 on the inhibition of CAR T-cell therapy vs solid tumors. Specifically, they targeted the codon of an N-linked glycosylation site on PD-1 that stabilizes the checkpoint and thereby compromises anti-tumor immunity.
Not only were the researchers able to achieve all three transitions of the AAC codon—that is, GAC, AGC and GGC—but also, each of the transitions resulted in diminished surface and total PD-1 levels. The base-editing did not, however, impair CAR T cell proliferation or activation.
More importantly, when activated by tumor cells, the base-edited CAR T cells demonstrated “enhanced cytolytic capacities and increased secretions of IL-2 and IFN-γ.
“Compared with CRISPR/Cas9, ABE has a narrower editing window and much less frequent off-target events, representing a safer and more precise approach for gene editing,” the authors concluded. “ABE-mediated point mutation can downregulate the inhibitory PD-1, therefore providing an alternative approach to augment T cell immunotherapy.”
Earlier this year, University of Ulsan’s Yongsub Kim and colleagues described their use of a cytidine base editor (BE3) to perform a functional analysis of BRCA1 variants. Using 745 guide RNAs targeting all exons in a high-throughput in-vitro screen, the researchers identified loss-of-function (LOF) variants, as well as variants with heretofore unknown functions.
“The introduction of BRCA1 variants with LOF mutations into a cell results in cell death with increasing passage numbers, and this can be detected through analysis of mutation frequencies,” the authors reported.
Transfecting the guide RNAs into BE3-expressing HAP1 cells, the researchers tracked BRCA1 mutations over 21 days by targeted deep sequencing. As expected, the relative mutation frequencies of pathogenic variants diminished over time, while those of benign variants remained similar.
They then examined the relative frequencies of three variants of unknown significance and noted they each decreased in over time, suggesting they were pathogenic. The same was true for a variant induced in the 5’-untranslated region (-UTR) of BRCA1.
“The 5′-UTR region of BRCA1 might regulate the transcription level, and several mutations in the 5′-UTR region are known as pathogenic variants,” the researchers stated. “To further validate the c.−97 C > T variant, we performed a luciferase reporter assay in HEK293T/17 cells, which showed that the c.−97 C > T mutation in the 5′-UTR caused a two-fold down-regulation of gene expression.”
The identification of a novel potentially pathogenic mutation in the 5′-UTR region of BRCA1, they pressed, highlighted the importance of the UTR region for clinical genetic testing.
Base editors continue to undergo development, whether to increase efficiency, enhance specificity or to potentially broaden activity.
As the Zhang studies suggest, base editors are not free of challenges either, and whereas those experiments could accommodate multiple and varied A > G transitions within a single codon, potential use of base editing in gene therapy might not.
“I would say that the biggest challenge is currently around what are called bystander edits,” explains Freeman, which occur when there are multiple target residues within the editing window, as above.
“This is of particular concern when the aim is to precisely repair a point mutation, for example, as a gene therapy for a monogenic disorder, which would be repaired by transition mutation,” he says, “but is less of a concern when the application is to knockout a gene, or cause exon skipping.”
Another challenge, he suggests, is delivery, which is a sentiment echoed by Christensen.
“For the CRISPR/Cas9 system, the best method so far seems to be the delivery of the actual Cas9 nuclease in conjunction with an already-present guide RNA, instead of using a plasmid,” Christensen says. “Unfortunately, for base editors, because they’re so new they are only available from Addgene as plasmids, and it’s a very large plasmid.”
“So, even though transfection methods have really come a long way in recent decades, it is still very challenging to get a 10-kb plasmid into hard-to-transfect cells,” she continues. “And then beyond that, to ensure that this plasmid is actually transcribed and translated into the appropriate base-editor protein.”
Thus, Christensen sees one of the big next steps for the field coming with the production and commercialization of the base-editor protein.
Delivery is also why she predicts most of the patient therapy focus turning toward ex-vivo cell modification—as the Choy lab is pursuing—rather than direct delivery to patients. This would allow researchers to prescreen cell therapies for any off-target modifications.
In a recent review, Choy, Christensen and colleague Rhea Ashmead noted the challenge with off-target modification and described efforts in another lab to modify BE3 to reduce off-target effects without reducing base-editing efficiency, variants given the appellative Selective Curbing of Unwanted RNA Editing, or SECURE.
Existing base editors also suffer from limitations in the types of mutations they can correct.
According to Choy, “The base editor relies on the deaminase enzyme to change the mutated base to the correct one, but right now it is limited to changes of a purine to purine or pyrimidine to pyrimidine. But if the original sequence is a purine that mutated to a pyrimidine, then the deaminase will not be able to correct it.”
But even that challenge is being addressed with the introduction of Prime editing a couple months ago.
Here, in place of a deaminase, Cas9 has been modified with a reverse transcriptase, and the guide RNA has been extended to include a segment with homology to the corrected gene sequence, facilitating the repair via reverse transcription.
“Prime editors are like a hybrid of the base editor and the CRISPR/Cas9 HDR,” says Christensen. “It can use this correction template to fix any type of mutation you have at that site.”
That said, Freeman adds, there are concerns over the editing efficiency and rate of DSB formation relative to base editing, so base editing may be a more appropriate technology to use in engineering cells destined for therapeutic use.
“I see Prime editing as being an exceptionally complementary technology to base editing,” he presses. “Where suitable—i.e., appropriate transition mutations and multi-gene knockouts—I believe base editing to be the preferred technology. Where base editing is not applicable—i.e., transversion mutations, small insertions/deletions—then Prime becomes the best available technology.”
Where genome- and base-editing may provide significant inroads in dealing with monogenic disorders, however, life becomes more complicated when one begins to explore multifactorial conditions, whether involving multiple mutations or metabolic pathways.
To address these situations, some researchers have turned to engineering.
From gene to circuit
“The way we have traditionally thought about biology is very static,” offers Senti Bio CEO Tim Lu. “Here’s a gene, the gene has a function. You knock out the gene or you over-express the gene.
“Biological systems are a lot more dynamic than that. It should be obvious, but it is something that the field has really come to grips with.”
As much as biology is about what genes are expressed, it is even more about how they are expressed, when they are expressed and in what combinations of expression with other genes. That dynamic interplay, he continues, is at the core of everything, from physiological development to disease pathology.
This is not to say Lu dismisses the recent advances in cell and gene therapy. He sees those as truly transformative for certain diseases, like monogenic disorders or certain cancers where a simple, single-target approach is sufficient.
“But if we want to go toward more complex diseases, solid tumors, for example, that likely involve multiple targets we have to hit, or diseases that may involve more than one target or may have a narrow therapeutic window,” Lu clarifies, “those are all things that are going to require therapies that are more sophisticated and more programmable. So that’s really what the field of synthetic biology is about.”
And that is why he moved from MIT to co-found Senti Bio.
“The goal for what we’re trying to do is provide the technologies to build those dynamic therapies,” he explains. “So, can we build gene therapies that can be turned on and off? Can we build gene therapies that are highly specific to a certain cell type? Can we build gene therapies that are conditionally on during a period of time?”
Senti Bio is tackling this issue with gene circuits.
Earlier this year, Lu and two colleagues from MIT, Ming-Ru Wu and Barbara Jusiak, reviewed the use of gene circuits vs. cancer, and they quickly jumped into language more of reminiscent of computer science than cell biology.
“Synthetic biology applies engineering principles to modify living cells, enabling them to perform sophisticated decision-making processes in order to produce a user-defined outcome,” the authors wrote. “This approach often involves programming artificial multi-gene circuits that consist of three components: a sensor that detects user-defined input(s), a processor that decides on the response to the inputs, and an actuator that produces the desired response.”
And like computer circuits, these gene circuits rely on logic gates that define their response to their environment.
With an OR gate, for example, if factor A or factor B is present, the output is triggered, and the cell is killed. This approach, Lu says, helps deal with parameters like heterogeneity, where not all cells in a tumor may have the same molecular profile.
A NOT gate, meanwhile, keeps you from destroying healthy tissues that may have similar profiles to the tumor cells, but distinctions as well; e.g., factor A NOT factor B = kill.
Likewise, an AND gate offers you greater specificity as it demands action only when both factors are present; e.g., factor A AND factor B = kill.
“Synthetic biology circuits are often based on transcriptional regulation but may also involve post-transcriptional regulators, such as microRNAs (miRNAs) or protein-based signalling cascades,” the authors described. “The inputs to a circuit can be exogenous, such as user-provided small molecules or specific wavelengths of light, or endogenous, such as transcripts or proteins that can distinguish between cell types of interest.”
Human cells are not computer chips, however, so designing highly functional gene circuits continues to be challenging.
“Often, when we try to gauge any project, it is really about number one, having the tools available, ones that you’ve built before and know work reasonably well,” Lu explains.
“Number two is being able to rapidly design, build and test,” he continues. “We take a very engineering approach here, where we’re not typically just looking at a single construct. We’re trying to develop methods where we can look at hundreds, thousands or tens of thousands of different variants, and then rapidly determine which ones work well, which ones don’t.”
It is that design/build/test/learn cycle that he feels will ultimately bring the company success. The more times you can go around that cycle, the faster you can converge on the final system.
An example of this is Lu’s efforts with The Hebrew University of Jerusalem’s Yuval Tabach and others to identify synthetic promoters with enhanced cell-state specificity (SPECS), which they described in 2019.
The authors reported that earlier efforts to design SPECS consisted of tandem repeats of transcription factor binding sites (TF-BSs) for a handful of TFs known to be active only in the cell state of interest. Thus, the SPECS were largely built from prior knowledge.
Instead, the researchers wanted to design a high-throughput experimental and computational screen that did not require prior information.
“For this purpose, we designed a library of synthetic promoters that corresponds to 6,107 eukaryotic TF-BSs reported in two databases,” they explained. “Each construct in the library comprises tandem repeats of a single TF-BS encoded upstream of an adenovirus minimal promoter to control the expression of mKate2 fluorescent protein.”
The team then used FACS, next-gen sequencing and machine-learning to identify optimal SPECS.
The researchers were able to identify promoters with up to 1,000-fold activity differential between the cell states of interest and their counterparts, highlighting “(i) distinct spatiotemporal activity in an organoid differentiation model; (ii) specificity for either a breast cancer or a normal breast cell line; and (iii) discrimination of stem-like glioblastoma cells from their differentiated counterparts.”
The high-throughput identification of SPECS also addresses another key factor in gene circuit design: modularity.
“I think modularity is one of the central things you need in engineering those complex systems,” Lu says. “It’s true for computers. In biology, it would be great if things were perfectly modular, but in reality, it is never that clean.”
Again, he presses, design is key.
“How do you design feedback loops or certain architectures to maximize modularity?” he asks. “And then on top of that, having that downstream optimization effort to make sure that indeed these components when you put them together do work for the application of interest is important.”
Lu says the company takes a hybrid approach, reusing components that they know work well across a variety of contexts, but also optimizing components to specifically fit particular applications.
In early proof-of-concept work published in 2017, Lu and colleagues designed an RNA-based immunomodulatory gene circuit versus ovarian cancer. Using an AND gate, the binding of two promoters led to the synthesis of surface T-cell engagers as well as secretion of chemokine CCL2, cytokine IL-12 and an anti-PD1 checkpoint inhibitor.
When delivered in vivo to mice, the gene circuit not only triggered robust anti-tumor responses, but also increased mouse survival. They also noted that only 15 percent of tumor cells needed to express the circuit to elicit the anti-tumor response, suggesting that delivery efficiencies need not be perfect to provide therapeutic effect.
“In addition, we demonstrated that our circuit can be readily modified to target breast cancer cells by replacing the synthetic input promoters,” the authors highlighted.
As with everything, it seems, delivery continues to be a challenge as Senti Bio seeks to translate their circuits toward the clinic.
“The way we think about this is that the gene circuit is the software that dictates the function, but then you have to find the hardware for delivery,” Lu says. “And depending on the application, sometimes the cell makes more sense and sometimes a virus makes more sense.”
“The way we’re approaching this currently is trying to take advantage of existing delivery modalities: lentivirus, AAV, T cells, for example,” he continues. “Ones where the manufacturing know-how is there, that we can really rapidly translate toward the clinic.”
He acknowledges, however, that even these vectors have their limits.
“So, the additional design constraint for some of these early gene circuits we’re building is not only to optimize for function, but also to minimize size,” he remarks. “That’s actually one of the reasons we came up with the synthetic promoter platform, because the size of the actual promoters makes a difference in how much you can package in [the vector].”
Senti Bio is hardly alone in designing gene circuits that need to tackle this issue.
In 2019, Tsinghua University’s Zhen Xie and colleagues, including researchers at Syngentech, described their efforts to develop oncolytic adenoviral approaches to cancer immunotherapy. Focusing not just on expression of immune effector molecules like GM-CSF, IL-2 and anti-PD-1 or anti-PD-L1 checkpoint inhibitors, the researchers also looked to control adenoviral replication in hepatocellular carcinoma (HCC), both in culture and in mice.
The researchers observed that non-replicating adenovirus constructs failed to inhibit tumor growth, whereas replication-competent constructs strongly inhibited tumor growth. For example, more than 80 percent of mice receiving the anti-PD-1 circuit were largely tumor-free for up to 60 days post-treatment.
They also noted that mice achieved vaccination against secondary challenge with HCC tumor cells after oncolytic virotherapy, “suggesting that immunological memory was induced to prevent tumor metastasis.”
Not everyone, however, is looking to individual genes for therapeutic answers. Some have extended their gaze to look beyond the gene’s linear structure.
Beyond the gene
Early efforts like the Human Genome Project tried to understand health and disease by focusing on the open reading frames of the exome, suggests Omega Therapeutics CSO Tom McCauley. And to some extent, that focus has borne fruit.
There is a growing body of evidence, however, that gene expression, and particularly cell-state-specific gene expression, is also controlled at a much higher level of chromatin structure, in the form of extended loops of DNA.
These loops—called insulated genomic domains (IGDs) by Omega, but also known as topologically associated domains (TADs)—are genomic stretches with increased internal contact but which tend to be isolated from neighboring regions.
“The loops can house single or multiple genes as well as their regulatory elements, and they are insulated at the base by two CTCF proteins, zinc-finger proteins that come together to form this insulated loop,” explains Omega CEO Mahesh Karande.
In mapping the more than 15,000 IGDs in human cells, Omega researchers have come to realize that much of the control of gene expression is dictated less by sequences proximal to a gene (e.g., promoters), but rather to enhancer and repressor binding sites much further away—however, any regulatory sequence must reside within an IGD to influence that gene.
“The position of TADs and boundary regions is remarkably stable between cell types or tissues, even displaying evolutionary conservation among species,” stated Max Delbrück Center for Molecular Medicine’s Irene Mota-Gómez and Dario Lupiáñez in a recent review. “The disruption of TADs is a prominent mechanism of human disease, leading to aberrant gene expression and causing congenital disease or cancer.”
Karande offers an example of such a link, describing one situation where a mutation created a novel CTCF binding site between two genes within a loop. This caused a second loop to form, altering gene expression and setting up a cancer pathology.
“When we disrupted the CTCF site, we basically saw the disease changing,” he continues. “We started seeing a return to the normal situation.”
This, he suggests, is a perfect example of the importance of both function and structure in health and disease.
“In this case, what happened was everything was functioning properly, but the structure changed when an additional CTCF site was created, and you saw the original loop bulging at one point,” he notes.
It is precisely this type of intervention that Omega is hoping to achieve with their Omega Controllers platform.
As they are expecting to make a series of announcements in 2020, the Omega team is vague on details about their platform, but offer broad indications of how it will work.
“First computationally and then empirically, we are able to figure out, within the IGD that is implicated in that particular disease phenotype, where the regulators sit and which regulators you could go at—CTCF itself, or depending on whether you are up-regulating or down-regulating, hit the enhancers differentially or you can hit a promoter,” Karande details. “Or you can do all of it, because these things tend to be additive.”
“While the core binding sequence elements of these CTCF sites are conserved across the genome, their flanking sequences and the context in which they sit is cell-type specific,” adds McCauley. “Having mapped that context across the cell types of interest for us in terms of specific diseases, we can target our controllers to specific cell types and have the effect we want to have only in specific tissues.”
The goal is then to use known epigenetic modulators to adjust gene expression in the desired manner, all of which will be encompassed in a yet-to-be-specified delivery system. To this latter aspect, Karande points to the achievements of sibling company Moderna, assured that delivery solutions will be forthcoming.
An additional feature that potentially works in Omega’s favor is the understanding that IGDs also tend to comprise genes functioning within the same metabolic pathways, explains Karande.
“The way nature has provided for it, those genes are put in the IGD because they literally act along a pathway for a particular disease,” he says. “That’s the beauty of our approach, because when we go and modulate the regulatory elements within the IGD, we are able to differentially modulate those genes.”
Again, without offering specifics, he suggests that in early experiments, the company was able to modulate multiple genes from within an IGD involved in inflammation and immunology, and produced outcomes currently achieved by a multibillion-dollar drug on the market today.
Aside from their own therapeutic plans, Karande sees opportunities to work with other companies to enhance the efficacy of other drugs, either directly or by laying the metabolic groundwork.
“Think of checkpoint inhibitors, which are miraculous drugs,” he offers. “You have 30 to 40 percent responders, which means 60 to 70 percent of people don’t respond to these drugs.”
“One of the research papers that came out of the Fred Hutchinson Institute earlier this year suggested that when a certain DUX4 gene gets expressed in patients who have cancer, I/Os [immuno-oncology therapies] don’t work,” he continues. “If we can go and shut that gene down temporarily—we’re not editing it, we’re not doing anything to the gene except turning it down so that the I/O can act—imagine that the 30 to 40 percent goes up to 40 to 50 or 60 percent. Imagine what you’re creating, in terms of patient benefits.”
It’s this idea of modulation rather than alteration that causes Karande to describe Omega’s approach as genomic medicine rather than genomic surgery.
Regardless of which approach is followed, most of these gene-replacement, -editing or -modulation therapies remain a few years from clinical testing. And yet, given the complexity of human health and disease, it is nice to know we will likely have options.