Abstract
Adeno-associated virus (AAV)-based vectors are the most commonly used vectors for gene therapy. Wild-type AAV infections occur widely in humans and nonhuman primates (NHPs), and an accurate assessment of preexisting AAV antibodies is crucial for the efficient use of AAV-based gene therapies in preclinical and clinical studies. Cynomolgus macaques (Macaca fascicularis) are well-established preclinical large animal models for evaluating the efficacy and safety of AAV-mediated gene therapies intended for human use. We provide a retrospective evaluation comparing preexisting AAV-neutralizing or total antibody titers against serotypes AAV2, AAV5, AAV8, or AAV9 in cynomolgus macaque cohorts of Asian or Mauritian origin. We used an in vitro neutralizing antibody (NAB) assay to detect NAB titers or an in vitro Meso Scale Discovery-based assay for the quantification of total binding antibodies (TABs) in blood samples. Results were obtained to measure the serostatus of animals. In our analysis, the in vitro NAB assay revealed the lowest seroprevalence for AAV5 (13 ± 15% to 21 ± 6%) independent of origin. In the same assay, Asian animals were highly seropositive against AAV8, followed by AAV2 and AAV9 serotypes (88 ± 13%, 71 ± 10%, 69 ± 9%, respectively). Whereby, the prevalence of seropositivity was lower in animals of Mauritian origin with the highest seroprevalence for AAV9 (58 ± 7%), followed by AAV8 (53 ± 17%) and AAV2 (51 ± 20%) assessed by in vitro TAB assay. Notably, co-prevalences of antibody responses against AAV2, AAV8, and AAV9 serotypes resulted in 39.8% seropositivity (in vitro NAB assay) in NHPs of Asian and in about 32.6% (in vitro TAB assay) of Mauritian origin.
INTRODUCTION
Adeno-associated virus (AAV)-based vectors are the most commonly used vectors for gene transfer and replacement approaches in gene therapy.1–3 As small non-enveloped viruses with an icosahedral capsid and single-stranded DNA genome, AAVs belong to the parvovirus family and are widespread in humans and nonhuman primates (NHPs).4,5 Thirteen distinct serotype variants of AAV have been identified, each exhibiting a different tissue tropism and the ability to transduce a diverse array of both dividing and nondividing cells in mammalian hosts.5,6 AAV is replication-deficient and requires the presence of helper viruses, such as adenovirus or herpes simplex virus, to replicate and complete its life cycle.7,8 Infection with AAV does not cause any known disease.7,9
Over the past years, gene transfer using recombinant AAV as a vector has gained importance in the development of gene therapy drugs. There are numerous drug candidates under development in late-stage clinical trials, and seven AAV-based gene therapy drugs have been approved by the U.S. Food and Drug Administration and/or the European Medicines Agency. Examples are Luxturna™ (AAV2-based) to treat a rare form of inherited retinal disease10–12 ; Zolgensma™, an AAV9-based therapy for the treatment of spinal muscular atrophy (SMA)13–15 ; Hemgenix™, the first hemophilia B gene therapy based on AAV516,17; and Elevidys™, a gene therapy for patients with fatal X-linked Duchenne muscular dystrophy using an AAVrh74 vector.18,19
Humans are naturally exposed to wild-type AAVs. Consequently, they have a certain risk of developing neutralizing antibodies (NABs) that increase with age, which may lead to cross-reaction and neutralization of the gene therapy drug. This might lower therapeutic cargo delivery, leading to reduced therapeutic efficiency or even triggering an increased acute immune response.20–22 NABs, as well as non-NABs, can influence the biodistribution of the therapeutics.23,24 The susceptibility of AAV vectors to NABs varies, depending on the route of administration, the vector dose, and vector formulation. Especially when administered intravenously, anti-AAV NABs can develop rapidly at high titers and persist for several years.21,25,26 In clinical trials, patients are typically excluded from enrollment if they have preexisting NABs (titers greater than 1:5) against the serotype of the used AAV vector. 27 Depending on the trials, patients with total binding antibodies (TABs) may be excluded as well (i.e., treatment with Elevidys™ TAB titers ≥1:400), 28 since TABs may be involved in complement activation after dosing.
Clinical development of gene therapeutics requires a preclinical investigation of local and systemic toxicities, biodistribution, and persistence of transgene expression. 29 Mostly, this investigation is conducted in cynomolgus macaques, which demonstrate a high genetic similarity to humans and are often the only clinically relevant animal species.30–32 Cynomolgus macaques from captive-bred populations of different geographic locations are utilized, from mainland Asia (with subpopulations of Vietnam, Cambodia, or China) or the distant island of Mauritius.33,34
Like humans, cynomolgus macaques are naturally exposed to wild-type AAVs, and the occurrence of NABs can have similar consequences as in patients participating in clinical trials. The impact of anti-AAV immune responses has not been fully understood. Interestingly, the impact of preexisting AAV5 NABs on the efficacy of AAV5-based gene therapy revealed limited predictive value for the efficacy of gene transfer. 35 In a clinical trial for hemophilia B, some patients had preexisting AAV5 NABs before treatment, showed a typical immune response following AAV5-hFIX treatment, and gene expression was not adversely affected. These findings were corroborated by NHP data, showing successful liver transduction even with high AAV5 NAB titers. 35 In contrast, low levels of AAV2 or AAV8 NABs have been shown to impair liver transduction significantly.36–40
Prescreening of blood from individuals is performed to assess the prevalence of current antibody titers against the AAV serotypes used for gene therapy. We compiled literature with seroprevalences of humans and NHPs, showing information about AAV2, AAV5, AAV8, and AAV9 (Table 1). The data demonstrate that seroprevalences vary widely, and different assay formats are employed to determine AAV antibodies. While extensive data for humans are available, such information is limited for NHPs.
Summary of Selected Published Seroprevalences of AAV2, AAV5, AAV8, and AAV9 Serotypes in Humans and Nonhuman Primates
AAV, adeno-associated virus.
To complement information on seroprevalences of cynomolgus macaques, we report data collected from different cohorts of Asian or Mauritian origin. The prevalence of NABs or TABs against AAV2, AAV5, AAV8, or AAV9 was evaluated to select seronegative animals for preclinical investigations for different AAV-based gene therapies. Two different tests were used for detecting antibodies against AAVs: one method identifies seronegative animals on the basis of an in vitro neutralization assay to detect NABs.22,56 The second method is an in vitro Meso Scale Discovery (MSD)-based assay for the quantification of TABs, which represents a fast and cost-effective method. 24
RESEARCH DESIGN AND METHODS
Study design and experimental setup I: In vitro NAB analysis for AAV2, AAV5, AAV8, and AAV9
Animals and blood collections
Animal care and use was in accordance with applicable welfare regulations at an AAALAC International-accredited animal program. Serum samples from 220 purpose-bred cynomolgus macaques (Macaca fascicularis) of Asian (n = 108, i.e., 97 males and 11 females, 10–42 months, 1.5–5.8 kg) or Mauritian origin (n = 112, i.e., 56 males and 56 females, 23–36 months old, 1.5–9.2 kg) were analyzed with the in vitro NAB assay. All samples were sent frozen on dry ice to the analytical laboratory.
In vitro NAB assay
This assay was conducted at the Immunology Core Laboratory (Gene Therapy Program), Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States. NAB responses to AAV2, AAV5, AAV8, or AAV9 capsids were measured in serum using an in vitro HEK293 cell-based assay and serotype-specific LacZ expressing vectors as previously described. 22
The AAV2, AAV5, AAV8, or AAV9 vectors used contained the gene encoding β-galactosidase (LacZ) driven by a cytomegalovirus (CMV) promoter (AAV2.CMV.LacZ.bGH, AAV5.CMV.LacZ.bGH, AAV8.CMV.LacZ.bGH, or AAV9.CMV.LacZ.bGH) and were produced by the Vector Core Laboratory, University of Pennsylvania (Philadelphia, USA). The NAB assay method has been described in detail previously. 22 Briefly, 1 × 105 HEK293 cells were seeded into each well of a 96-well flat-bottom plate and cultured overnight at 37°C in a 5% CO2 incubator. On the following day, serum samples, previously heat-treated at 56°C for 35 min, were serially diluted in medium starting at 1:5. Diluted samples or controls were preincubated with the respective vector (multiplicity of infection of 1 × 104) at 37°C for 60 min. After incubation, preincubated mixtures were transferred to plated cells and incubated at 37°C in a 5% CO2 incubator for 20–24 h. Vector-transduced cells were then lysed, developed using a chemiluminescent substrate, and read on a spectrophotometer. The presence of the respective anti-AAV NABs in the test sample resulted in reduction of vector transduction, leading to a lower luminescence signal readout.
The NAB titer values were reported as the highest serum dilution, inhibiting AAV transduction (β-galactosidase expression) by ≥50%, compared with a naïve mouse serum control. The limit of detection was a 1:5 serum dilution, and variability ± one 2-fold serum dilution.
Softmax Pro 6.5.1 GxP for collection of luminescence values from a Molecular Devices M3 SpectraMax spectrophotometer and Microsoft Excel were used for data collection and analysis.
Study design and experimental setup II: In vitro TAB and NAB analysis for AAV2, AAV8, and AAV9
Animals and blood collections
Animal care and use was in accordance with applicable welfare regulations at an AAALAC International-accredited animal program. Serum samples from 289 purpose-bred cynomolgus macaques (M. fascicularis) of Mauritian origin (n = 264, i.e., 107 males and 157 females, 19–32 months old, 1.9–4.5 kg) were analyzed for TAB titers. In addition, serum samples from one cohort of Asian animals (n = 25 males, 29–30 months, 2.2–2.7 kg) were analyzed for TAB titers (Supplemental Data). All samples were sent frozen on dry ice to the analytical laboratory.
The following AAV production, titer determination, and assays were conducted at Eye Health & Research Beyond Borders, Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach an der Riss, Germany, and Reaction Biology Europe GmbH, Freiburg, Germany.
AAV production and titer determination
AAVs harboring anti-fluorescein isothiocyanate (FITC) under the control of the CMV promoter were produced following the method described previously. 57 Briefly, 16-layer CELLdiscs (Greiner Bio-One, Kremsmünster, Austria) were seeded with 1.5 × 104 HEK293H cells/cm² (acCELLerate, Hamburg, Germany). After 72 h, calcium phosphate triple transfection was performed using AAV helper-free system (Agilent, Santa Clara, CA), packaging plasmid containing rep2/cap2, 8 and 9, respectively, and transfer plasmids.
After 72 h, cells were detached using EDTA and harvested via centrifugation. Cell pellets were resuspended in lysis buffer (1 M NaCl, 50 mM Tris, 20 mM MgCl2, 0.001% [v/v] Pluronic F-68, pH 8.5) and lysed through two freeze/thaw cycles. Lysates were treated with salt-active nuclease (ArcticZymes Technologies, Tromsø, Norway), and AAV particles were purified through polyethylene glycol precipitation and iodixanol density gradient ultracentrifugation to enrich full capsids and remove cellular debris. Finally, the iodixanol fraction containing AAV particles was concentrated and buffer-exchanged into formulation buffer using 100 kDa Amicon-15 tubes (Merck Millipore, Darmstadt, Germany). For virus titration, viral DNA was extracted using the ViralXpress DNA/RNA extraction reagent (Merck Millipore), and absolute quantification of viral genomes was performed using droplet digital PCR. Briefly, viral DNA was serially diluted 1:5 (from 5.0 × 104 to 7.8 × 108), mixed with 12.1 μL of master-mix (Bio-Rad Laboratories, Hercules, CA) and 1.1 μL of a 20× primer–probe mix targeting the CMV promoter (forward primer: 5′-CCAAGTACGCCCCCTATTGAC-3′, reverse primer: 5′-CTGCCAAGTAGGAAAGTCCCATAAG-3′, and probe: 5′-CCGCCTGGCATTATG-3′, Thermo Scientific, Waltham, MA). Droplets containing DNA were generated using an automated droplet generator (Bio-Rad Laboratories), and viral genomes were amplified via PCR (95°C for 10 min, followed by 40 cycles of 94°C for 30 s and 60°C for 1 min, and 98°C for 10 min). Droplets containing amplified amplicons were quantified using QX200 Droplet Reader (Bio-Rad Laboratories).
Based on in-house experience and literature, iodixanol-purified AAV yields ∼90 − 95% full capsids, as confirmed by electron microscopy, SamuxMP Mass Photometry, and capsid enzyme-linked immunosorbent assay (ELISA)/viral titer ratios. The full/empty ratio was not determined for batches used in this study; instead, new viral batches were tested in TAB assays against intravenous immunoglobulin (IVIG) as reference for vector quality and assay validity.
In vitro TAB quantification assay
TABs against AAV2, AAV8, and AAV9 serotypes in serum were analyzed using an antigen-capture immunogenicity assay format on the MSD (Meso Scale Discovery®, Rockville, MD) platform as described previously. 24 MSD plates (L15XA-1) were coated at 4°C overnight with 5 × 108 AAV viral genomes/well. Blocking was performed using blocking solution (5% MSD Blocker B [R93BB-2]). After incubation with serum serial dilutions on coated AAV capsids, the bound IgG1–3 was detected with antihuman/NHP IgG antibody (MSD D20JL-6). Electrochemiluminescence was detected after addition of 2× MSD Read Buffer (R92TC-2) by MSD Imager.
Samples were serially diluted in AAV buffer from 1:2 to 1:2048. Background signals in relative counts (relative light unit [RLU]) of phosphate-buffered saline-coated wells for each serum (or IVIG) dilutions were subtracted from individual AAV-coated samples. Values <10 RLU were considered below detection. Half maximal inhibitory concentration (IC50) values were determined by nonlinear fitting in GraphPad Prism v.9.3.1 using log10-transformed dilutions and the “log(inhibitor) versus response variable slope” function (Supplementary Fig. S5).
In vitro NAB assay
The titer of NABs against AAV2, AAV8, and AAV9 serotypes in NHP serum was assessed using a transduction inhibition assay as described previously. 24 Briefly, 96-well plates were seeded with 2.5 × 104 HEK293H cells/well 24 h before transduction. Recombinant AAV vectors (expressing anti-FITC) were diluted in serum-free Dulbecco’s modified eagle medium (DMEM) and incubated with 2-fold serial dilutions of serum samples (from 1:2 to 1:1024 in serum-free DMEM) for 30 min at 37°C. Subsequently, serum–vector mixtures, corresponding to 2,500 viral genomes/cell, were added to plated cells and incubated for 72 h at 37°C with 5% CO2. All transductions were performed in triplicate. Supernatant was collected and transferred to new 96-well plates, with cell debris removed by centrifugation (200 g for 5 min). The amount of anti-FITC was determined by anti-FITC MSD assay, as previously described. 58
Transduction efficiency was measured as relative counts/well and normalized to a transduced control without serum. The cutoff for seropositivity was set as 50% inhibition of transduction at a 1:5 serum dilution. The NAB assay was used to confirm serostatus for TAB-negative animals (Fig. 2B).
Data evaluation and statistical analysis
Data were analyzed using Microsoft Excel and GraphPad Prism v.9.0.1. Different statistical methods were used, depending on the dataset and experimental setup; statistical methods are indicated in the figure legends. For comparison between two groups, the unpaired Student’s t-test was used. For comparison among several groups, statistical significance was assessed by analysis of variance followed by Student–Newman–Keul’s post hoc test for multiple comparisons. Data are reported as mean ± standard deviation. Categorical variables are presented as percentages, while continuous variables (such as gender or age) are depicted as individual data points. Values of p < 0.05 were considered statistically significant. For all figures, significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001.
Venn diagrams were created using the BioVenn web tool for visualization of overlapping sets. 59 Manual modifications were applied to represent dataset-specific characteristics.
A regression analysis and Pearson’s correlation analysis were carried out for NAB titer values and age (Supplementary Fig. S1) or TAB titer values and age (Supplementary Fig. S2). The confidence interval was set at 95%, and p < 0.05 (two-tailed) was considered as statistically significant. The p value, correlation coefficient according to Pearson (r), and its coefficient of determination (R2) are listed in the corresponding figures.
RESULTS
Experimental setup I: In vitro NAB analysis
Mauritian and Asian origin cynomolgus macaques show similarly low seroprevalence for NABs against AAV5
Anti-AAV5 NABs were analyzed in serum from 220 animals of Asian (n = 108) or Mauritian (n = 112) origin using a NAB assay. The results obtained for both populations demonstrate low and comparable seroprevalences. Asian animals showed seroprevalence of 21 ± 6%, and Mauritian animals showed seroprevalence of 13 ± 15% (Fig. 1A). The distribution across different dilutions (<1:5, 1:5, 1:10, >1:20) revealed similar results for both populations (Fig. 1B): A value of <1:5 was interpreted as a negative NAB response. Animals were considered positive at a titer of ≥1:5. A low titer (1:5) was seen in 10 ± 6% of Asian and 6 ± 4% of Mauritian origin animals. A medium titer (1:10) was found in 5 ± 3% of Asian and 3 ± 4% of Mauritian origin animals, and a high titer (>1:20) was detected in 6 ± 3% of Asian and 5 ± 7% of Mauritian origin animals. Next, we examined the impact of age on seroprevalences in Asian or Mauritian males or females. When plotted against age, there was no difference across groups, indicating that in our study, age is not the determining factor for seroprevalence (Fig. 1C). As in humans developing antibodies increase with age, we analyzed if age-dependent seroprevalences occur in NHPs. Based on Amato et al. 2002, cynomolgus macaques were split into the following categories: infant (0–30 days), juvenile (31 days to 2.5 years), peripubertal (>2.5–4 years), young adult (>4–9.5 years), adult (>9.5–20 years), and aged (>20 years). 60 We divided the animals into two possible groups: juvenile and peripubertal. Since no animals older than >4 years were included, results should be interpreted regarding the limited age ranges (Fig. 1D). Moreover, after regression and Pearson’s correlation analysis, no relationship was noted between NAB titer values and age (Supplementary Fig. S1A and B).
Serotype profiles between Asian and Mauritian (AAV5), and in Asian (AAV2, AAV8, and AAV9) cynomolgus macaques assessed by in vitro neutralizing antibody analysis (experimental setup I).
(Continued.)
Cynomolgus macaques of Asian origin show a high seroprevalence against AAV2, AAV8, and AAV9
Next to AAV5 NABs, AAV2, AAV8, and AAV9 NABs were screened in sera from 108 animals of Asian origin by NAB analysis. We identified a high percentage of seropositive animals for NABs specific to AAV8 (88 ± 13%), AAV2 (71 ± 10%), and AAV9 (69 ± 9%) serotypes with no significant difference across serotypes (Fig. 1E). Titer distribution analysis revealed that most animals show high (>1:20) antibody titers, with more animals showing high titers for AAV8 compared with AAV2 and AAV9 (Fig. 1F). Plotting over age did not reveal differences across groups, except a slightly lower age of AAV2-seropositive animals compared with seronegative animals (Fig. 1G). For AAV2, a significant correlation coefficient according to Pearson (Supplementary Fig. S1C) was noted, whereas for the other two AAV serotypes, no significant correlations were observed (Supplementary Fig. S1D and E). Splitting the Asian animals into age groups resulted in 96 juvenile and 12 peripubertal animals. Seroprevalence for AAV2 and AAV8 in juveniles was higher than in the older age group (Fig. 1H). These differences are not considered to be of relevance as age ranges were limited and require further investigation.
Co-prevalence of NAB responses against AAV2, AAV5, AAV8, and AAV9 serotypes is common
An important consideration for gene therapy studies based on AAVs is the co-prevalence of NABs against capsids of two or more different AAV serotypes. For 108 individual animals of Asian origin, results were available for AAV2, AAV5, AAV8, and AAV9 NABs (Fig. 1I, Table 2). A high percentage of animals were positive against AAV2, AAV8, and AAV9 NABs (39.8%). In addition, 15.7% were also positive for AAV5. Only 11.1% of all animals tested were negative for all four AAV serotypes. A positive rate of 11.1% was observed for AAV8, 7.4% for AAV8 and AAV9, or for AAV2 and AAV8, 2.8% for AAV2 and 1.9% for AAV9.
Cross-Tabulation of AAV2, AAV5, AAV8, and AAV9 Seroprevalences Ranked in Descending Order of Frequency in Asian Cynomolgus Macaques Assessed by In Vitro Neutralizing Antibody Serum Analysis (Experimental Setup I)
The analysis included 108 nonhuman primates of Asian origin with complete data for all serotypes (“+”: seropositive and “−”: seronegative).
Number of animals.
Percentage of affected animals (percentages may not total exactly 100% due to rounding).
AAV, adeno-associated virus.
Experimental setup II: In vitro TAB analysis followed by in vitro NAB analysis
Seropositivity against AAV2, AAV8, and AAV9 serotypes is similar in Mauritian cynomolgus macaques
The presence of antibodies, unrelated to neutralizing activity, is a hallmark of previous exposure and potentially even non-NABs could modify the biodistribution and biological activity of gene therapy using AAVs. Therefore, next to the abovementioned cell-based in vitro NAB assay, we also tested a ligand-binding assay for TABs against AAV2, AAV8, and AAV9 in sera of Mauritian cynomolgus macaques (Fig. 2). Additionally, 25 blood samples of one small cohort of male Asian cynomolgus macaques were tested for TABs against AAV8 (Supplementary Fig. S3).
Using the TAB analysis, seroprevalences for AAV2 (51 ± 20%), AAV8 (53 ± 17%), and AAV9 (58 ± 7%) were not statistically different (Fig. 2A). Most of the negative TAB animals were retested by in vitro NAB assay and confirmed as negative (Fig. 2B). When plotted against age, there was a statistically higher age in AAV2- and AAV8-seropositive animals (Fig. 2C). Analysis of the correlation coefficient according to Pearson showed a tendency of older animals having higher titers of AAV2 antibodies; however, this was not significant (Supplementary Fig. S2). Age stratification of Mauritian animals revealed higher AAV2 seroprevalence in juveniles than in peripubertal animals. For AAV8 and AAV9, only juvenile animals were available (Fig. 2D). Furthermore, in additional 25 Asian males, AAV8 seroprevalence was around 64% (Supplementary Fig. S3), and titers showed high variability at 30 months (Supplementary Fig. S2D), highlighting the importance of balanced age distribution in future studies to improve statistical reliability.
AAV2, AAV8, and AAV9 serotype profiles in Mauritian cynomolgus macaques analyzed by in vitro total binding antibody analysis (experimental setup II).
(Continued.)
Co-prevalence of TAB responses against AAV2, AAV8, and AAV9 serotypes in Mauritian cynomolgus macaques
Similar to NABs, co-prevalence for TABs against AAV2, AAV8, and AAV9 serotypes was investigated in 129 animals (Fig. 2E, Table 3), revealing 32.6% seropositivity and 27.9% seronegativity for all three AAVs. The highest percentage for double-positive individuals was noted for AAV8 and AAV9 (13.2%). Combinations with AAV2, such as AAV2 and AAV9 (7%), and AAV2 and AAV8 (6.2%) were lower. The lowest single positivity was observed for AAV8 (1.6%), followed by AAV2 (3.1%) and AAV9 (8.5%).
Cross-Tabulation of AAV2, AAV8, and AAV9 Seroprevalences Ranked in Descending Order of Frequency in Mauritian Cynomolgus Macaques Assessed by In Vitro Total Binding Antibody Serum Analysis (Experimental Setup II)
The analysis included 129 Mauritian cynomolgus macaques with complete data for all serotypes (“+”: seropositive and “−”: seronegative).
Number of animals.
Percentage of affected animals (percentages may not total exactly 100% due to rounding).
AAV, adeno-associated virus.
We compared TAB titers among seropositive Mauritian animals for AAV2, AAV8, and AAV9 (Supplementary Fig. S4B), as similar titers may suggest previous coinfection. AAV9 TAB titers were around 4-fold higher than AAV2 or AAV8. Likewise, we analyzed NAB titers of seropositive Asian animals (setup I), whereas AAV8 titers exceeded those of AAV2 and AAV9 (Supplementary Fig. S4A).
DISCUSSION
There are 13 well-characterized AAV serotypes, each with different binding receptors and varying capabilities to infect specific tissues or cell types.3–6 AAV2, AAV5, AAV9, and AAVrh74 serotypes are commonly used as recombinant vectors for gene therapy. Others, such as AAV8, are used in experimental settings to achieve higher efficiency of AAV-based gene therapy. Preexisting antibodies against capsids of AAV serotypes can reduce the efficiency of the therapy.20,61 This not only applies to humans in clinical trials but also to cynomolgus macaques used in preclinical investigations before novel gene therapies enter clinical development. The investigation reported herein aims to contribute further knowledge about seroprevalences of antibodies binding to AAV serotypes, AAV2, AAV5, AAV8, or AAV9.
Recent studies indicate seroprevalences against AAV serotypes vary widely, possibly due to differences in populations analyzed and the use of non-standardized assays with varying levels of sensitivity. While data are available for humans, robust information for NHPs is lacking; donor details, such as origin, age, and sex, are often missing. Therefore, a key strength of our article is the substantial sample size, which enhances the statistical power of the findings. We conducted a retrospective analysis of previously obtained data. Since the animals were enrolled in preclinical studies, it was not possible to collect additional blood samples from the breeder. Hence, experiments were carried out using different cohorts and assays. Unfortunately, there is no standard assay for the assessment of AAV-specific antibodies; laboratories use different assays. Our investigation employed two different types of in vitro assays: the first measured NABs by a neutralization assay, and the second detected TABs by an MSD-based assay independent of neutralizing capacity.
In setup I, NAB assays for AAV2, AAV5, AAV8, and AAV9 were conducted only on animal samples of Asian origin, with AAV5 additionally tested in Mauritian samples. Therefore, only AAV5 allowed direct comparison across origins using the same method. In setup II, TAB assays for AAV2, AAV8, and AAV9 were performed on Mauritian animals. Additionally, a small cohort of 25 Asian animals was included, supporting direct comparison and revealing similar seroprevalence. Generally, TAB seroprevalences (setup II) were lower compared with NAB (setup I). As Mauritian animals are geographically isolated, genetic or environmental factors may partly influence exposure and antibody responses. Moreover, the MSD TAB assay may result in some false positives and IgG negatives. False-positive sera in the IgG1–3 screen have occasionally been described as borderline outliers, showing RLU signals at the cutoff point between 10 and 100 RLU. False-negative animals occurred more frequently, lacking IgG but showing detectable IgA or IgM titers, or other neutralizing factors. 24 Therefore, sera near cutoff values were reanalyzed via NAB assay (setup II), revealing NAB-positive animals likely having IgA, IgM, or other factors mediating AAV neutralization (Fig. 2B). Combining TAB assays for initial screening with NAB assays for follow-up of low-titer or negative animals offers a practical and cost-effective approach, as the NAB assays are a more definitive assay due to functional biology that occurs. MSD-based TAB assays are broadly applicable to all AAV serotypes and recombinant capsids, independent of the packaged transgene. They deliver rapid, reproducible results using minimal serum, ideal for mouse studies and seroconversion experiments. The assay enables batch testing with patient serum, combining ELISA-like sensitivity with the precision of cellular NAB assays, providing a wide dynamic range for natural and recombinant AAVs. 24
However, differences between the experimental setups limit direct comparison. Further studies with older animals are required to clarify age-related seroprevalence patterns. The current findings offer reference points for selecting representative NHPs in preclinical research.
AAV5
Our work demonstrates a low prevalence of AAV5 NABs in Mauritian or Asian cynomolgus macaques with a seroprevalence of around 20%. This was generally lower than the 44% reported in rhesus macaques, likely due to differences in sample size. 53 We tested over 100 animals, compared with 16 in the published dataset. Smaller sample sizes are more susceptible to variability, which may partly explain the discrepancy. Additional factors, such as animal origin or assay sensitivity (using Huh7 cells vs. our HEK293 cells), may also contribute to this difference.
Seroprevalence for the AAV5 serotype in humans is reported to range from 3.2% 50 to 42%.41,43,45,46,52 Interestingly, seroprevalence varies by geography, 62 with 3–4% seroprevalence noted for France,43,50 40.2% for China, 63 and 42% in the Basque population. 43
AAV5 was originally isolated from human tissue, 64 and it is the most distinct from the other serotypes.65–67 AAV5 has tissue tropisms for lung, 68 central nervous system (CNS),69,70 and retina. 71 Thus, AAV5 might primarily infect tissues or cell types with less frequent immune interaction, potentially resulting in reduced antibody development.
AAV2
From all AAVs, AAV2 is considered the most studied serotype. 72 AAV2 binds strongly to human cell receptors, particularly heparan sulfate proteoglycans, which are prevalent in many cell surfaces. 73 This binding facilitates effective infection and persistence of AAV2 in tissues, increasing the probability that individuals will develop antibodies. Therefore, AAV2 has various tropisms for several tissues, including renal tissue,74,75 retina,76,77 liver,78,79 skeletal muscle,80,81 and CNS. 82
Our study detected antibodies against AAV2 in a range of 51–71%. A seroprevalence of 51 ± 20% was found in Mauritian origin animals using the TAB assay, and 71 ± 10% was found in Asian origin animals using the NAB assay. The difference in seroprevalence could be related to origin or individual colonies or—more likely—to the assay. Standardizing NAB assays across different laboratories is challenging because it is a biological assay with multiple parameters, making it variable and dependent on different AAV productions and vector quality control, as well as different cells and readouts. Our reanalysis of TAB-negative animals by NAB (with setup II) confirmed a high percentage of animals being negative, especially for animals with a negative IC50 in the TAB Assay. Our data match a 69.5% seroprevalence reported in cynomolgus macaques. 51 For humans, the reported prevalence of antibodies against AAV2 varies, ranging from 18% to 76%, depending on location and assay used.24,41–50,83
AAV8
Seroprevalences and titers are higher in species from which the virus was isolated, with considerable cross-species neutralizing activity. AAV8 was isolated from liver tissues of NHPs. 84 AAV8 tissue targets include the liver38,85–87 and CNS.88,89
In humans, AAV8 seroprevalence ranges from 16% to 52%.41–50 In our study, we identified 88 ± 13% of Asian animals seropositive for AAV8 using the NAB assay. Using the TAB assay, AAV8 seroprevalence of 53 ± 17% was found in Mauritian and 64% in an additional cohort of 25 Asian animals. These discrepancies could potentially be attributed to the degree of variation in the standardization of the (biological) NAB or TAB assays or to the differences between the individual colonies, which may account for the variability observed in our results. An ELISA-based method and NAB assay for AAV1, AAV8, and AAV9 showed strong correlation in 50 rhesus macaque and 20 human sera. Thus, indicating that either method could be used equivalently or together to identify seronegative animals for preclinical studies. 90
Our data are consistent with reported prevalence values in cynomolgus macaques of 59–77%.51,53,55
AAV9
AAV9 has a broader tissue tropism compared with other AAV serotypes, for example, to eye, 91 lung, 92 kidney,93,94 liver, 95 heart,95–97 and skeletal muscle, 97 is able to cross the blood–brain barrier, and can transfect cells of the CNS.98–102
We found a seroprevalence of 69 ± 9% in Asian and 58 ± 7% in Mauritian origin cynomolgus macaques. These values are comparable with previous reports of 63–68% seroprevalence in cynomolgus macaques.51,53 The extensive AAV9 tropism potential increases immune exposure and antibody production. Humans have a lower seroprevalence against AAV9 antibodies in the range of 17–38%.43–45,48,50,52 In a recent study among 69 adult patients with SMA type 2 and 3, only three patients (4.3%) had relevant AAV9 NAB titers. The prevalence did not increase with age, suggesting gene therapy with intravenously administered recombinant AAV9 vectors could be a feasible option for treating patients with SMA and applicable to other inherited neurological diseases. 103
Co-prevalence of antibody responses against AAV2, AAV8, and AAV9
While antibodies against AAV5 are less prevalent, antibodies against AAV2, AAV8, and AAV9 are more common; hence, many individuals may demonstrate co-prevalence of antibodies against capsids of multiple AAV serotypes. Co-prevalence can occur when an individual has been infected with different AAV serotypes over their lifetime, resulting in the presence of antibodies against these serotypes. Our data demonstrate around 40% of NHPs of Asian origin presenting co-prevalence of NAB responses to AAV2, AAV8, and AAV9, while 32.6% of Mauritian origin animals had TABs to the same serotypes. Asian animals showed higher NAB titers for AAV8, indicating a stronger immune response, while similar AAV2 and AAV9 titers suggest possible previous coinfection. In Mauritian animals, AAV9 TAB titers were around 4-fold higher than AAV2 or AAV8, suggesting an enhanced immune response. Different titers might imply cross-reactivity of anti-AAV antibodies. AAV2, AAV8, and AAV9 are phylogenetically related.104,105 The capsid proteins (VP1 to VP3) of these serotypes show significant structure and sequence similarities, which are relevant for the development of cross-reactive antibodies. Assuming similar capsid proteins, AAV2, AAV8, and AAV9 were reported to share the same cell receptor, the 37/67-kDa laminin receptor. 106 In a 10-year study with 25 chimpanzees, natural AAV-infection induced cross-reactive responses to AAV serotypes, as animals with AAV8 NABs had high levels of cross-reactive antibodies against other serotypes (AAV1, AAVrh10, AAV5, and AAV9), while AAV8-negative animals lacked such neutralizing responses. 107
The studied animals are geographically separated and might have different genetic and environmental factors, impacting exposure to and infection with AAVs and antibody responses. However, key factors contributing to seroprevalences need to be further investigated, especially across extended age groups. Depending on the AAV serotypes and seroprevalences we observed, broader NHP screening, especially for AAV2, AAV8, and AAV9, may be required to identify appropriate preclinical candidates. Different strategies to overcome antibodies include designing novel AAV variants 108 or eliminating circulating IgGs, which correlate with the level of NABs, by enzymatic degradation using Streptococcus pyogenes enzymes to enable effective AAV gene transfer despite preexisting antibodies. 109
AUTHORS’ CONTRIBUTIONS
B.P. provided conceptualization, data curation, performed the data analysis, statistical analysis, and wrote the original draft of the article. U.M. performed data curation and data analysis. D.B., S.M., and F.F.-T. performed experiments and experimental analysis. B.P., U.M., D.B., F.F.-T., A.D., B.K., S.G.-W., S.M., and L.M. contributed to editing and review of the data and findings. All authors read, reviewed, and approved the final article.
Footnotes
ACKNOWLEDGMENTS
The authors thank Jessica A. Chichester, Immunology Core Laboratory, Gene Therapy Program, Perelman School of Medicine, University of Pennsylvania, United States, for performing the in vitro NAB assay experiments. They also thank Svenja Leyse and Stefanie Klaas, Labcorp Early Development Services GmbH, Germany, for technical support.
AUTHOR DISCLOSURE STATEMENT
B.P., U.M., A.D., B.K., S.G.-W., and L.M. are employees of Labcorp. D.B. and S.M. are employees of Boehringer Ingelheim. F.F.-T. was an employee of Reaction Biology Europe.
FUNDING INFORMATION
This research received no external funding.
Supplemental Material
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
