Complete neutralizing antibody evasion by serodivergent non-mammalian AAVs enables gene therapy redosing

Figures

• 

  Figure 1. Design and selection of a divergent Dependoparvovirus capsid library (A) Maximum likelihood phylogenetic tree of the genus Dependoparvovirus. Relationships were inferred from the Rep78 sequence of 294 Dependoparvovirus isolates using IQ-TREE. Isolates were grouped into subclades based on sequence similarity. Gray boxes indicate clades selected for capsid matrix. (B) Representative Dependoparvovirus structure and genome with highlighted N-terminal domain. AAV9 structure predicted with AlphaFold3. 40 amino acid N-terminal domain highlighted in blue. Created with BioRender.com. (C) Principal-component analysis of Dependoparvovirus VP1 protein sequence (n = 435). Clustering was performed using K-means. Blue corresponds to AAV2-like isolates, gray to anseriform dependoparvovirus-like isolates, and green to other Dependoparvovirus isolates. Matrix candidates are marked in red. (D) Schematic for capsid matrix barcode screen. DNA sequences for 16 AAV capsids and 5 VP1 N-termini were combinatorially cloned, yielding 80 capsid sequences. Two 15-cm plates of HEK293 cells were triple-transfected per capsid, using unique barcoded ITR plasmid for each capsid. Media were pooled on days 3 and 7…

• 

  Figure 2. Library screening identifies AAV.div3A as top performing chimeric capsid (A) Capsid enrichment. RNA FC for the best-performing N-terminus of each capsid in vivo, averaged across liver, skeletal muscle, and heart tissues. Gastrocnemius was used as representative skeletal muscle tissue. Capsids were excluded from analysis if the input percentage was less than 0.015%. n = 3. (B) Capsid rank. Relative ranking for each capsid was determined by the RNA fold change of the best performing N-terminus of each capsid in vivo. Capsids were excluded from analysis if the input percentage was less than 0.015%. Biological replicates were averaged for data representation. Raw fold change values and in vitro data are shown in Figure S2. n = 3. (C) Vector yield. AAV titers were compared across four independent production runs for AAV.div3A and AAV9. For each production run, ten to twenty 15-cm plates of HEK293 cells were transfected with 12 μg pXX680, 10 μg pLH-RepCap, and 6 μg ITR plasmid per plate. Media and cells were harvested on days 3 and 7 post-transfection. AAV was then purified by PEG precipitation, followed by iodixanol ultracentrifugation and diafiltration (see STAR Methods). AAV preps…

• 

  Figure 3. AAV.div3A exhibits no seroprevalence or antigenic cross-reactivity (A) Antigenic cross-reactivity for AAV.div3A. Two female C57BL/6J mice were injected i.p. with 5e12 vg/kg for each respective capsid packaging ss-CBh-Luc-P2A-GFP. After 6 weeks, mice were sacrificed, and serum was pooled between biological replicates. Each serum sample was diluted 1:2 using PBS. The diluted serum was mixed in equal volume with each AAV vector, resulting in a final dilution of 1:4 and a final MOI of 1.25e4 vg/cell. Samples were incubated at 37°C for 1 h. U87 cells were then dosed in triplicate, and transduction was measured via luciferase assay at 72 h. Luminescence was then normalized to a no-serum control for each capsid and averaged across replicates. Red indicates neutralization of transduction, and blue indicates enhancement of transduction. Phylogenetic inference was performed with IQ-TREE. n = 3. (B) Schematic for donor neutralizing antibody assay. Serial dilution was performed on serum from 20 human donors. Diluted serum was mixed in equal volume with AAV vector packaging ss-CBh-Luc-P2A-GFP, resulting in a final MOI of 1.25e4 vg/cell. Samples were incubated at 37°C for 1 h and then added to…

• 

  Figure 4. AAV.div3A evades neutralization in vivo following passive immunization (A) Passive immunization study schematic. Two hundred microliters of serum, AAV9 NAb+ or AAV9 NAb−, was injected i.p. into four C57BL/6J mice each per serum donor (2 female, 2 male). After 72 h, mice were injected i.v. with 5e13 vg/kg AAV9 or AAV.div3A packaging ss-CBh-Luc-P2A-GFP. After 6 weeks, mice were sacrificed, and transduction was assayed in liver, heart, muscle (gastrocnemius), and brain tissue by luciferase assay. Created with BioRender.com. (B) AAV9 neutralizing antibody titers for NAb+ donors used in passive immunization study. ID50 values were calculated by fitting triplicate values to a logistic dose curve. Average R2 = 0.936. n = 3. (C and D) Transduction after passive immunization. Two mice were injected per serum sample for AAV.div3A (C) and AAV9 (D). Positive or negative serum samples were grouped for statistical analysis. Dotted line represents background luminescence from non-transduced tissue. Ungrouped donor level data are found in Figures S9B–S9F. n = 8 (positive serum) and n = 10 (negative serum). p values were determined by Kruskal-Wallis test followed by Dunn’s post hoc test. (E and…

• 

  Figure 5. AAV.div3A evades vector-induced immunity in vivo (A) Schematic for vector-induced immunity study. Female C57BL/6J mice were divided into immunized and unimmunized cohorts. Immunized mice were injected i.v. with 5e12 vg/kg AAV9 packaging a non-translated stuffer genome while unimmunized received a PBS vehicle injection. After 6 weeks, both cohorts received a subsequent i.v. dose of 5e12 vg/kg AAV9 or AAV.div3A packaging ss-CBh-Luc-P2A-GFP. Six weeks later, mice were sacrificed, and transduction was measured by luciferase assay. Created with BioRender.com. (B) Neutralization of transduction in immunized mice. Average percentage difference between immunized and unimmunized mouse transduction for each tissue type. Data were averaged across biological replicates and reported with standard error of the mean. n = 5. (C–F) Transduction in unimmunized and immunized mice for liver (C), heart (D), kidney (E), and muscle (gastrocnemius) (F). Dotted line represents background luminescence from non-transduced tissue. n = 5. p values were determined by two-way ANOVA followed with a Holm-Sidak post hoc test. ns: p > 0.05, ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001.

• 

  Figure 6. Rational engineering of AAV.div3A capsid yields myotropic variant (A) AlphaFold3 predicted structure of VR8 loops for AAV.div3A capsid mutants. AAV.div3A is shown in red, AAV.div3A-M1 is shown in gray on the left, and AAV.div3A-M4 is shown in gray on the right. (B) In vitro transduction. C2C12 cells were transduced with vector packaging ss-CBh-Luc-P2A-GFP at an MOI of 5e4 vg/cell. Transduction was measured after 72 h by luciferase assay. Full set of in vitro data for AAV.div3A capsid mutants is shown in Figures S10A–S10D. n = 5. p values were determined by one-way ANOVA followed with a Holm-Sidak post hoc test. (C–F) In vivo transduction. Female C57BL/6J mice were injected i.v. with vector packaging ss-CBh-Luc-P2A-GFP at a dose of 5e13 vg/kg. Mice were sacrificed after 6 weeks, and transduction was measured in diaphragm (C), heart (D), liver (E), and gastrocnemius (F) tissues by luciferase assay. Dotted line represents background luminescence from non-transduced tissue. Full set of in vivo data for AAV.div3A capsid mutants is shown in Figures S10E–S10L. n = 3. p values were determined by one-way ANOVA followed with a Holm-Sidak post hoc test. (G) Transduction of AAV.div3A-M1 was…

• 

  Figure 7. Therapeutic redosing of AAV.div3A-based capsids in Pompe disease model (A) GAA redosing study schematic. B6:129 GAA knockout mice were injected i.v. with 2e12 vg/kg AAV9 packaging ss-CBh-GAA. After 4 weeks, mice received a second i.v. injection of 2e13 vg/kg AAV9, AAV.div3A, or AAV.div3A-M1 packaging GAA or a PBS vehicle injection (3 female, 2 male per capsid). At week 10, mice were sacrificed. Tissue was assayed for GAA activity using a fluorometric 4-MUG cleavage assay. Created with BioRender.com. (B–E) GAA activity after redosing. GAA activity was assayed in diaphragm (B), heart (C), liver (D), and gastrocnemius (E) 6 weeks after receiving a second dose of AAV packaging ss-CBh-GAA. Data are represented as nmol substrate cleaved per hour per mg protein. n = 10 (n = 8 for 3A-M1). p values were determined by Kruskal-Wallis test followed by Dunn’s post hoc test. See also Figure S12.