Advances in Precision Editing Therapies for Alpha-1 Antitrypsin Deficiency

- AAT Introduction

Alpha-1 antitrypsin deficiency (AATD) is a disease most commonly due to mutations in the SERPINA1 gene, which encodes a serine protease inhibitor. AAT’s most critical function is as a neutrophil elastase inhibitor, but it has been shown to have numerous anti-inflammatory effects and to inhibit several other proteases (Table 1).¹ The condition affects 1 in 25 persons of European descent, with 95% of cases resulting from a G-to-A point mutation (PiZ) in SERPINA1.^2,3 AAT is mainly produced by hepatocytes, monocytes and macrophages and secreted into the serum and functions to inhibit neutrophil elastase in the lungs. In healthy individuals (PiMM), serum AAT concentration ranges from 20-53 μM.³ Heterozygous PiMZ carriers produce ~60% of normal AAT levels, while homozygous PiZZ patients have only ~15% of normal levels.⁴ A serum AAT level of < 11μM alone can cause emphysema at a relatively young age, even in patients who are not exposed cigarette smoke or other environmental insults. AAT heterozygotes (PiMZ) who smoke have a 2.3-fold greater risk of developing emphysema than PiMM individuals.⁵

In the liver, AATD primarily manifests through hepatocellular accumulation of the misfolded Z variant.⁶ The Z variant aggregates in the endoplasmic reticulum to form periodic acid-Schiff staining positive and diastase resistant (PAS-D) inclusions. Over time, these toxic polymers can contribute to hepatic injury and cirrhosis in approximately 10% of patients through a gain-of-function, which can be exacerbated by alcohol use, obesity, and underlying liver disease.^7,8 The only curative treatment for AATD-associated liver disease is liver transplant, which has a 5-year survival rate around 95%.⁹

As mentioned above, serum AAT in the lung inhibits a spectrum of serine proteases, most notably elastase but also proteinase-3, metalloproteases and TNFalpha converting enzyme (TACE), and has a variety of other anti-inflammatory effects (Figure 1).^2,3 Although Z-AAT retains partial inhibitory activity, it is markedly less effective than M-AAT.¹⁰ Consequently, insufficient elastase inhibition in AATD leads to alveolar wall destruction, and an increased risk of panacinar lung emphysema through a “loss of function” mechanism. In addition, polymerized Z-AAT acts as a neutrophil chemoattractant, further amplifying pulmonary inflammation while M-AAT has anti-inflammatory properties including the inhibition of IL-8.¹¹ The combination of chronic inflammation and unopposed protease activity predisposes individuals with AATD to progressive emphysematous lung disease.

AATD-associated emphysema presents clinically similarly to chronic obstructive pulmonary disease (COPD) and treatment strategies overlap, emphasizing both lifestyle and pharmacological therapies. Intravenous AAT infusion is the only AAT-specific therapy approved by the FDA for life-long use in PiZZ individuals with severe AATD, defined by serum AAT levels below 11 M and evidence of lung disease.^3,12,13 However, it only attenuates lung function decline and fails to reverse existing lung damage.

- BE/PE/RNA editing

CRISPR technologies have revolutionized modern genetic medicine. Traditional CRISPR-Cas9 editing relies on a nuclease-based mechanism, in which a single guide RNA (sgRNA) directs the Cas9 to a target genomic locus to generate a double-stranded break (DSB) (Figure 2).^14,15 The resulting DSB is repaired by the cell’s endogenous pathways, primarily classical non-homology end joining (c-NHEJ) or microhomology-mediated-end-joining (MMEJ). If a donor DNA template is provided, homology-directed repair (HDR) can occur instead, though this pathway is limited to cells in the S and G2 phases of the cell cycle.

The development of CRISPR-Cas9 technology laid the foundation for next-generation genome editing tools such as base editors and prime editors. Base editing enables precise single-nucleotide conversions, generally from A-to-G (adenine base editor) or C-to-T changes (cytosine base editor).^16,17 Adenine base editors (ABEs) rely on the use of a TadA deaminase protein tethered to a nickase or dead Cas9 (Figure 2). The Cas9 nickase will nick the target strand while the TadA deaminase will deaminate adenines in the non-target strand to become inosine, which is read as guanosine by the cell. Similarly, cytosine base editors (CBEs) use a cytosine deaminase to convert cytosine to uracil, which the cell reads as thymine. CBEs typically include an uracil glycosylase inhibitor (UGI) to prevent the removal of uracil by base excision repair enzymes.

While base editors offer a powerful means for gene correction, they can also introduce bystander edits due to their relatively broad editing windows, which encompass multiple adenines or cytosines. Bystander edits that yield silent mutations may be benign; however, those affecting critical residues can alter protein misfolding, stability or function. To mitigate this, researchers have refined both the protein effector - developing sequence-dependent, domain-inlaid, and narrow-window editors- and RNA components, including imperfect, or DNA-RNA hybrid gRNAs.^18-22

Unlike base editing, which enables only point edits, prime editing (PE) and related methods allow for substitution, insertions, and deletions without double-strand breaks.²³ In PE, Cas9 is fused to a reverse transcriptase domain and the sgRNA is replaced with a prime editor guide RNA (pegRNA) that contains a primer binding site (PBS) and reverse transcriptase template (RTT) (Figure 2). Following Cas9-mediated nicking of the non-target strand, the PBS can anneal with the exposed 3’ flap, and the RTT serves as the template for reverse transcription of the desired edit. PE has undergone multiple rounds of optimization including effector and guide improvement to support larger, more efficient, and more precise edits.^24-28

Editing at the RNA level offers a means to correct pathogenic mutations without altering the underlying genome.^29-32 The transient nature of RNAs require RNA editors to be readministered periodically to maintain therapeutic effect. However, this transient nature can also be advantageous, as it avoids permanent genomic alterations and reduces the risk of undesired effects such as bystander edits, indels, and off-target modifications commonly associated with genome editors.

Sharp and his colleagues were the first to report cirrhosis in patients with AATD.³³ They subsequently reported on less severe liver pathology and adult-onset cirrhosis among patients with AATD.^34,35 Larger series have shown that many cases of AAT liver abnormalities during infancy and childhood do not progress to severe disease.^36,37 The observation that Z-AAT accumulates in polymerized form within the ER of hepatocytes in patients with AAT liver disease has led to the conclusion that liver disease in AAT deficiency is due a toxic gain of function of the mutant PiZ protein.³⁸ Therefore, therapeutic interventions to prevent or reverse AAT liver disease depend on the ability to decrease the production of the mutant protein, either by knocking down gene expression or by editing the gene to reverse the Z mutation to the wild-type sequence. This has now been substantiated by clinical trial data with fazirsiran and siRNA that down-regulates AAT expression in the liver of PiZZ patients and improves liver transaminase levels in the serum.³⁹

In contrast, AAT lung disease is due the loss of the normal antiprotease function of AAT, leading to the unrestrained action of neutrophil elastase and other proteases and pro-inflammatory molecules. Multiple lines of evidence indicate that AAT replacement, i.e., augmentation therapy with wild-type AAT protein, has beneficial effects on the severity and progress of AAT lung disease, including prospective studies demonstrating effects on CT lung densitometry and frequency of exacerbations.^2,40-42 Thus, a gene editing strategy for AAT lung disease must either convert the mutant gene or mRNA to the wild-type sequence or insert a wild-type gene in a context where it will be expressed at physiologic levels.

- Preclinical studies

Preclinical studies for AATD have primarily focused on gene replacement, gene editing, and RNA editing approaches. The most widely used mouse models are the C57BL/6-PiZ and NSG-PiZ animals, both of which carry multiple copies of the human PiZ allele and recapitulate the hepatic manifestations of AATD.^43,44 More recently, a novel AAT-null;PiZ mouse strain was generated, in which the endogenous murine Serpina1 genes were knocked out while the human PiZ allele is retained.⁴⁵ This model enables the study of AATD-associated lung pathology, particularly following multiple repeated lipopolysaccharide (LPS) challenges.

CRISPR-based gene knock-in strategies offer a means to restore functional SERPINA1 by inserting the PiM allele adjacent to the PiZ mutation. The HDR template with the PiM allele was introduced using AAV8 while the Cas9 expression cassette was introduced with AAV9.⁴⁶ M-AAT in the serum increased to ~45 μg/mL in C57BL/6-PiZ animals at an editing rate of ~2%. Similarly, SaCas9-mediated HDR generated ~5% correction as detected by RNA-seq.⁴⁷ These strategies highlight the therapeutic potential for CRISPR mediated AAT therapy, though further optimization is needed to improve editing efficiency, and achieve clinically meaningful protein restoration.

The inefficiency of Cas9 nuclease to introduce the AAT gene prompted the use of other genome editors, such as base editors to introduce precise edits. Beam Therapeutics’ BEAM-302 consists of a SpCas9-ABE mRNA encapsulated in a lipid nanoparticle (LNP) for delivery. In preclinical studies, BEAM-302 was administered at 0.75 mg/kg or 1.5 mg/kg in both NSG-PiZ and C57BL6-PiZ animals.⁴⁸ Editing reached 25% in C57BL6-PiZ animals and 36% in NSG-PiZ animals, resulting in serum M-AAT up to ~1500 μg/mL (Table 2). Kim et. al. further demonstrated the versatility of LNP delivery for both hepatocyte and AT2 cell editing, achieving 40% and 10% correction, respectively.⁴⁹ Importantly, sufficient gene correction by ABE can also mitigate AATD emphysema. Gao et. al. used a single AAV8 to deliver a compact eNme2-ABE, resulting in reduced PAS-D+ pathology and improved lung physiology in C57BL6-PiZ animals and AAT-null;PiZ animals.⁴⁵ Prime Medicine leveraged an LNP encapsulated prime editor to correct the PiZ allele.⁵⁰ In their preclinical study, the editor achieved up to 72% precise correction in humanized mice and restored over 95% of serum AAT to the correct M-AAT isoform (Table 2). Finally, other efforts for AATD include Tessera Therapeutics’ RNA Gene Writer, which corrected 70-95% of the PiZ allele in hepatocytes, and CRISPR Therapeutics’ SyNTase editor, which achieved >99% mRNA correction (Table 2).^51,52

RNA editing provides an alternative approach for correcting the PiZ allele at the transcript level. Wave Life Sciences developed a GalNAc-conjugated oligonucleotide, WVE-006, which achieved 68-75% RNA editing in primary hepatocytes derived from PiZ transgenic mice.^53,54 In treated cells, ELISA measurements showed total AAT levels of 4,543 ng/mL, compared to 1,994 ng/mL in untreated cells, indicating a substantial restoration in AAT expression following RNA editing (Table 2).

- Ongoing clinical trials

The success of preclinical trials for AATD gene therapy has paved the way for clinical translation across multiple therapeutic platforms. Beam Therapeutics advanced its preclinical program into the clinic and initiated treatment of AATD patients with BEAM-302 (Table 3).⁵⁵ Part A of their phase I/II study enrolled 18–70-year-old PiZZ patients with total serum AAT < 11 uM, evidence of lung disease (FEV1 >40% predicted, FEV/FVC <70%), confirmed emphysema and no liver disease. A single 60 mg dose increased total AAT to 12.4 uM and decreased Z-AAT by 79% by day 28, with no adverse side effects reported. Ongoing efforts include evaluating two 60 mg doses given 8 weeks apart and expanding eligibility to patients with mild to moderate liver disease with or without lung disease. Similarly, Tessera Therapeutics and Regeneron’s AATD correction program also focuses on gene editing.⁵⁶

Building on this momentum, RNA-targeting strategies have also begun to demonstrate promising clinical activity. Wave Life Sciences launched the RestorAATion-2 trial to correct the mutant AAT transcript using WVE-006 (Table 3). In patients receiving multiple 200 mg doses, total AAT increased to 11.9 μM, with M-AAT comprising 64.4% of total AAT and Z-AAT decreasing by 60.3%. A single 400 mg dose produced similar improvements, raising total AAT to 12.8 μM, with M-AAT accounting for 47.2% of total AAT and Z-AAT declining by 49%.⁵³ No severe adverse effects were reported, highlighting the feasibility of transcript correction as a therapeutic approach.

Similarly, Korro Bio initiated the REWRITE program to deliver the RNA editing agent, KRRO-110, to AATD patients (Table 3).⁵⁷ Following an initial safety and tolerability assessment in healthy volunteers, PiZZ patients received either 0.6 mg/kg or 0.8 mg/kg KRRO-110. Turbidimetry demonstrated total serum AAT levels of 10 μM, while LC/MS analysis showed an increase of 2 μM in M-AAT. Future steps include implementing multiple-ascending dose studies of KRRO-110, which would further define the clinical potential of RNA editing in restoring AAT expression.

Discussion

Collectively, advancement in gene editing, RNA modulation, and vector engineering has transformed the therapeutic landscape for AATD, moving the field closer to a durable, genotype-directed cure. However, several challenges remain for both genome and RNA editing approaches. First, further optimization of both the delivery vehicle and the editing agent is warranted. For example, bystander edits remain a common limitation of base editing and could generate mutant AAT proteins of unknown clinical significance. Although several variants have been characterized in vitro, additional studies are needed to understand their potential immunogenicity, aggregation propensity, and functional interactions with downstream targets such as elastase. Notably, LC/MS data from Korro Bio’s KRRO-110 suggest an absence of bystander editing, highlighting the possibility that improved editor specificity could mitigate this concern.⁵⁷

Enhancing delivery systems also represents a key area for improvement. Increasing delivery efficiency to the target cell could reduce both the amount and frequency of dosing required to achieve therapeutic editing levels. This is particularly relevant for LNP-based delivery platforms, where repeated dosing must maintain potency despite innate immunity activation and adaptive responses that may develop over time.

Secondly, future studies should explore whether combining editing and traditional AAT augmentation could lead to a synergistic benefit. This would be particularly relevant for RNA editing, which – while potentially safer due to its transient nature- requires repeated dosing over a patient’s lifetime. Because both RNA editing and AAT augmentation involve ongoing administration, it will be important to determine whether repeated cycles of RNA editing could surpass the benefits observed with AAT augmentation alone. Such combination strategies may ultimately offer a more comprehensive approach by simultaneously reducing toxic Z-AAT accumulation and bolstering functional AAT levels. Finally, the toxicity of concomitantly administering augmentation and editing agents should be addressed.

Finally, the long-term implications of editing tools must be carefully considered. Although genome editing approaches are often designed as single-dose interventions, any bystander or off-target edits introduced at the DNA level are likely to be permanent. Such unintended edits could generate nonsense or missense variants that produce aberrant proteins, potentially triggering immune responses, or forming deleterious protein. Moreover, because genome editing typically requires a single dose, delayed adverse editing events may be difficult to detect, particularly over extended time frames as patients are lost to long-term follow-up.

The choice of delivery vehicle further influences the long-term safety profile of gene editing therapies. While non-viral platforms such as LNPs enable more transient editor expression, viral based vehicles such as AAVs can drive prolonged expression, potentially accumulating off-target risk. Accordingly, careful selection of promoter elements and viral serotypes to restrict tissue expression to relevant cell types, as well as the incorporation of strategies to limit editor activity after therapeutic correction has been achieved, will be important considerations for clinical translation.

In contrast to genome editing, RNA editing introduces reversible and transient modifications, which may offer an improved safety profile by avoiding permanent genomic alterations. However, this requires repeated dosing to maintain therapeutic efficacy. Cumulative exposure from repeated dosing may pose safety risks, such as continuous off-target editing across the transcriptome, or immune response to delivery vehicles or editing agents that affect editing efficiency.

In summary, the advance of genome and RNA editing strategies have enabled exciting preclinical and clinical studies for AATD. Ongoing clinical trials will unveil the long-term efficacy of safety of editing-based treatment.