David R. Liu, PhD, a professor at Harvard University, the Broad Institute of MIT and Harvard, and the Howard Hughes Medical Institute

A research team led by gene editing pioneer David R. Liu, PhD, reports the application of base editing technology to develop a one-time treatment for spinal muscular atrophy (SMA), showing promising effectiveness in cell and mouse models.

Liu, a professor at Harvard University, the Broad Institute, and an investigator with the Howard Hughes Medical Institute, together with 16 co-authors, published their findings today in Science, in a report entitled, “Base editing rescue of spinal muscular atrophy in cells and in mice.”

The paper is the first one published in a peer-reviewed journal that details a base editing approach to treating SMA—the leading genetic cause of infant mortality. Two months ago, Benjamin P. Kleinstiver, PhD, assistant professor in the Center for Genomic Medicine at Massachusetts General Hospital (MGH) and Harvard Medical School, and 12 co-authors posted a related preprint on bioRxiv detailing successful results from their study to optimize and implement a base editing treatment for SMA.

In SMA, patients have either a loss of, or mutations in, the SMN1 gene. The result is a lack of its corresponding protein SMN, which is critical to cellular functions that include those required for embryo development.

Liu and co-authors detailed how they restored SMN protein abundance to normal levels and rescued disease phenotypes in cell and mouse models of SMA by developing one-time genome editing approaches targeting another gene, called SMN2.

SMN2 differs from SMN1 at just one position—a single C-to-T mutation in exon 7. The result is a defective truncated SMN protein that fails to fully compensate for the loss of SMN1, but which provides just enough SMN protein activity to allow babies to be born with SMA. In the most common SMA cases (type 1), those babies typically live less than two years without treatment.

Liu and colleagues developed a customized base editor (out of more than 40 tested) that directly converts SMN2 to the same sequence as SMN1 by reversing the C-to-T mutation that occurred during the duplication of SMN1 to SMN2. The investigators reported that their strategy converted SMN2 to SMN1 with near-ideal editing efficiency of 99% in cells, fully restoring SMN protein levels ~40-fold to normal levels, while generating minimal off-target edits.

“Because all SMA patients have SMN2—otherwise loss of SMN1 would not allow them to be born—this base editing approach to treating SMA also should be applicable to all SMA patients, regardless of the specific mutation that caused their SMN1 loss,” Liu said.

Different approach

“This work,” Liu continued, “represents a departure from the more common approach of identifying a specific mutation that affects a certain population of patients and then correcting that mutation, which can leave unserved those patients who suffer from the same disease but do not have that particular mutation.”

By contrast, the SMN2 base editing approach could potentially be applied to all SMA patients because it does not target their specific SMN1 mutation, but instead corrects the defective duplicated copy of SMN that they all possess.

“Such an approach of converting imperfectly duplicated or cryptic genes using precise gene editing technologies into healthy genes may also be a viable strategy to treat other genetic diseases,” Liu added.

The approach detailed by Liu and colleagues was one of dozens of genome editing approaches they explored for potential in treating SMA. The 79 (to be precise) approaches included either CRISPR-Cas9 nucleases or base editing strategies. All 79 targeted five regions of SMN2 to induce either post-transcriptional or post-translational regulatory changes in SMN2 that upregulate SMN protein production.

“We took a ‘no stone unturned’ approach to developing a one-time gene editing treatment for SMA,” Liu said.

Human cells

Kleinstiver has a close scientific kinship with Liu. The MGH faculty member trained with Keith Joung, MD, who has co-founded several biotech companies with Liu, including Editas, Beam Therapeutics, and Pairwise.

Benjamin P. Kleinstiver, PhD, Assistant Professor in the Center for Genomic Medicine at Massachusetts General Hospital (MGH) and Harvard Medical School

Kleinstiver told GEN that his group used different enzymes from Liu’s group but the same target base in its editing approach—a single nucleotide in exon 7 of SMN2 via an A•T to G•C base edit. That approach led to a 40- to 50-fold increase in unregulated SMN expression, his group reported in their bioRxiv preprint.

“In general, our studies are highly complementary, both converging on efficient and safe base editing approaches to up regulate SMN protein expression as a treatment for SMA. Their work is excellent and there is a lot of synergy between our strategies,” Kleinstiver said. (Kleinstiver’s manuscript is still under review for publication in an undisclosed journal.)

Kleinstiver’s group focused largely on human HEK293T cells and primary SMA patient-derived cells to assess on and off-target editing, as well as phenotypic changes in SMN expression, based on cells donated by five SMA patients and collected at MGH.

“This gave our work the benefit of editing the endogenous human locus (to determine impacts on gene expression, genome-wide safety, etc), as would be done in any eventual expanded preclinical or clinical work,” Kleinstiver told GEN Edge.

PAM challenge

The traditional CRISPR-Cas enzymes used to read and edit DNA has certain limitations, due to the requirement of Cas enzymes to recognize a short sequence called a protospacer adjacent motif (PAM), prior to gene editing.

“We directly addressed this challenge by engineering a version of a CRISPR enzyme called SpRY that can, for the first time, edit any DNA sequence,” Kleinstiver said. “By allowing unprecedented access to the entire genome, SpRY fundamentally reframes how we approach genome editing since we can now conceivably target and modify nearly any genetic mutation.”

SpRY has been leveraged by Kleinstiver’s group and others in applications that have included correction of disease-causing sequences in preclinical models of various diseases previously intractable with existing tools, functional genomics applications that require high resolution perturbations such as CRISPR screens, and development of new molecular tool. “Critically, with SpRY we can achieve a wide range of previously impossible edits, including those needed to correct mutations that cause a broad range of diseases,” Kleinstiver said.

Kleinstiver was co-corresponding author of his lab’s base editing SMA preprint along with an Instructor in his research group, Christiano Alves, PhD.

“Our SpRY base editors can reach >98% editing in SMA patient derived primary cells, efficiencies not previously demonstrated,” Alves said. “This single base editing strategy is also extensible in vivo via AAV vector delivery in a severe mouse model of SMA.”

Alves noted that safety assays identified only two significant off-target edits in SMA patient-derived fibroblasts across more than 168 off-target bases, both located in a part of the genome unlikely to result in damage—results that showed that the group’s base editing strategy was highly precise, efficient, and safe.

“Our manuscript establishes a proof-of-concept of how a single base editor enzyme built on SpRY is applicable to treat a variety of other disease-causing mutations, including neuromuscular diseases and beyond,” Alves concluded. “This roadmap for a simple editing strategy using one enzyme in patient derived cells, followed by rapid translatability into in vivo models, will expedite the development of new genetic medicines.

Liu observed that the Kleinstiver lab’s approach appears to be one tested by his lab, but which showed a ~2.5x lower editing efficiency than the approach eventually adopted by his lab. “This difference may explain their lower level of in vivo editing—four to six percent in CNS, compared to ~40% in the CNS of the mice we treated—as well as the lack of reported disease phenotype rescue.”

SMA and beyond

One of the co-lead authors of the Liu group’s new report is former postdoctoral fellow Mandana Arbab, PhD.

Mandana Arbab, PhD, assistant Professor at the Harvard Medical School Department of Neurobiology

She was recently appointed Assistant Professor at the Harvard Medical School Department of Neurobiology, and will be starting her own lab at Boston Children’s Hospital’s Rosamund Stone Zander Translational Neuroscience Center. The Center aims to help scientists and clinicians accelerate the translation of research findings to treat SMA and other pediatric neurological disorders.

“My lab will continue to optimize a base editing therapeutic for SMA, and similarly develop precision genome editing strategies for other genetic neurological diseases,” Arbab told GEN Edge.

An important area of research for her lab, Arbab said, will be finding ways to limit the expression of genome editing agents in vivo to improve the safety of a future base editing therapeutic, and to use genome editing tools to better understand the pathological mechanisms of genetic neurological diseases.

“Together with David and others, I have applied for NIH funding to continue the preclinical validation of a base editing treatment for SMA,” Arbab added.

Prior to joining the Liu lab, Arbab did her PhD in The Netherlands on induced pluripotent stem cell (iPSC) modeling of motor neuron diseases, particularly focused on SMA.

“This disease has interested me due to its many contradictions,” Arbab said. “The implicated gene is an essential housekeeping gene, yet only motor neurons are so affected as to cause disease. And while the genetics has been uncovered decades ago, to this day we don’t understand the cellular mechanisms of pathology. However, SMA is a monogenic disease [that] can be rescued by restoration of SMN expression, suggesting that SMA would be a great candidate for a gene-based therapy.”

When CRISPR-Cas9 editing was discovered a decade ago, Arbab saw the possibility for a gene editing approach for SMA by modifying SMN regulatory sequences.

“I spent the last few years of my PhD working on nuclease-based strategies to treat SMA in cells and in mice,” Arbab said. “Right as I was graduating, David’s lab published their first base editing paper. Motivated by the strong impact of single-nucleotide changes on SMN gene expression and the clinical potential of base editing, I applied to David’s lab with the hopes of finding a base editing treatment for SMA.”

In the years that followed, the development of the adenine base editor and other CRISPR and base editing optimizations “finally enabled us to make the desired changes to the SMN2 gene that [grad student] Zaneta [Matuszek] and I explored in this paper,” Arbab added. “Most importantly, we were now able to efficiently edit the C-to-T base change that differentiates insufficient SMN2 genes from healthy SMN1 genes to enable native endogenous regulation of SMN expression.”

Base knowledge

Base editing was first described in two key papers published in Nature that detailed work led by two then- postdocs in Liu’s lab. A 2016 paper described the work led by Alexis C. Komor, PhD, (now an assistant professor at University of California, San Diego) on the cytosine-to-thymine  base editor (CBE). A follow-up paper published in 2017 detailed the adenosine-to-guanosine base editor (ABE) in work led by Nicole M. Gaudelli, PhD (now Senior Director, Head of Gene Editing Platform Technologies at Beam Therapeutics). Liu co-founded Beam Therapeutics to commercialize the technology.

In 2019, Liu and colleagues at the Broad Institute published another paper in Nature  introducing “prime editing”. Like base editing, prime editing does not introduce double-strand breaks in the target sequence. Liu and postdoctoral fellow, Andrew Anzalone, MD, PhD, co-founded Prime Medicine. (Anzalone recently discussed the technology and the company on GEN’s “Close to the Edge” video interview series).

Given the considerable excitement and potential of prime editing, it’s not unreasonable to ask: Why wasn’t prime editing studied in the SMA research?

“Our paper’s experiments to choose an editing strategy were done several years ago, before the development of prime editing in late 2019, which is why we tested nucleases and base editors but not prime editors in this paper,” Liu explained.

When delivered with  AAV serotype 9 (AAV9) into SMA mice that lacked working Smn1 and had the defective human SMN2 gene, the AAV9 efficiently delivered the base editor into 43% of spinal motor neurons. In treated SMA mice, SMN2 was converted to SMN1 in 87% of the cells that received the base editor. The treated mice showed marked rescue of motor function when assessed via electrophysiology measurements and behavioral measures.

The lifespan of the SMA mice was extended from an average of 17 to 23 days, Arbab and colleagues reported, adding that the extension was likely due to the unusually short window for treatment in these mice, which allowed only approximately six days for correction to take place before the fate of motor neurons is sealed. Base editing takes about one to three weeks to occur following AAV delivery, explaining why lifespan extension was modest despite the efficient conversion of SMN2 to SMN1. In humans, the window for treatment that benefits SMA patients ranges from months to more than one year.

Simple enhancement

Reasoning that virtually all SMA patients would likely be on a drug marketed to treat the disease, Liu and colleagues hypothesized that co-administering a single dose of the Biogen/Ionis Pharmaceuticals SMA treatment Spinraza® (nusinersen), an SMN2-directed antisense oligonucleotide, along with the base editor AAV would extend the therapeutic window for gene correction in SMA mice, and also would more closely resemble the likely treatment state of SMA patients in a clinical trial.

The researchers found that one-time in vivo co-administration of Spinraza and base editor extended animal lifespan nearly seven-fold, from an average of 17 days untreated to an average of 111 days, along with substantial rescue of motor function in behavioral assays such as inverted screen hang time or the time required for the mice to right themselves from being on their backs.

“This work is important as it established a very narrow therapeutic window in this severe SMA mouse model, showing that the expression of base editors from an AAV delivered ICV (Intracerebroventricular injection) does not occur quickly enough for editing to occur prior to onset of pathology,” Kleinstiver said.

“The use of nusinersen potentiates the therapeutic window, providing time for the base editors to be expressed, edit the SMN2 gene, and restore SMN production prior to major deterioration. This resulted in a large extension in lifespan of the SMA mice.”

Spinraza won the FDA’s first approval for an SMA treatment in in 2016 and is now one of three drugs indicated for the disease. (The others are Evrysdi®, marketed by Roche and its Genentech subsidiary) and Zolgensma®,   a one-time gene therapy designed to replace the SMN1 gene.

Unlike the other drugs, which carry broad SMA indications in children and adults, Zolgensma is approved for a narrower patient population of children under age two with SMA with bi-allelic mutations in SMN1. Last year, Novartis acknowledged that two patients died of acute liver failure following treatment with its Zolgensma.

Evrysdi and Spinraza deliver transient results. And while Zolgensma is indicated for single-dose intravenous infusion, some patients may require repeat administration, the safety and effectiveness of which have not been evaluated, according to the gene therapy’s prescribing label. None of the three drugs have restored SMN protein to native levels under native endogenous regulation mechanisms, both of which have been shown to be potentially important for cellular or animal health.

“These limitations motivated us to develop a one-time gene editing treatment for SMA that might permanently restore normal SMN protein levels with native regulatory mechanisms, without requiring repeated dosing,” Liu said.

Previous articlemRNA-LNP Vaccine Development: Evaluation of Novel Ionizable Lipids
Next articleOne AI Tool Bests Another in Enzyme Prediction Match