Hexons from adenovirus serotypes 5 and 48 differentially protect adenovirus vectors from neutralization by mouse and human serum

Abstract

Adenovirus vectors are widely used in gene therapy clinical trials, and preclinical studies with these vectors are often conducted in mice. It is therefore critical to understand whether mouse studies adequately predict the behavior of adenovirus vectors in humans. The most commonly-used adenovirus vectors are derived from adenovirus serotype 5 (Ad5). The Ad5 hexon protein can bind coagulation factor X (FX), and binding of FX has a major impact on vector interactions with other blood proteins. In mouse serum, FX protects Ad5 vectors from neutralization by natural antibodies and complement. In the current study, we similarly find that human FX inhibits neutralization of Ad5 vectors by human serum, and this finding is consistent among individual human sera. We show that human IgM and human IgG can each induce complement-mediated neutralization when Ad5 vectors are not protected by FX. Although mouse and human serum had similar effects on Ad5 vectors, we found that this was not true for a chimeric Ad5 vector that incorporated hexon regions from adenovirus serotype 48. Interestingly, this hexon-chimeric vector was neutralized by human serum, but not by mouse serum. These findings indicate that studies in mouse serum accurately predict the behavior of Ad5 vectors in human serum, but mouse serum is not an accurate model system for all adenovirus vectors.

Fig 1. FX inhibits neutralization of Ad5 in human serum.

Luciferase-expressing Ad5 vectors were incubated at 37°C with either C57BL/6 mouse serum or individual human sera at a serum concentration of 80%. Samples were then diluted and assayed for infectivity on 293 cells. (A) Infectivity of vectors after incubation with 12 human sera that have low antibody-mediated neutralizing activity. Sera were heat-inactivated at 56 °C to destroy complement. Ad5.0 is a control Ad5 vector with a wild-type capsid, and Ad5.7 is a modified vector that is unable to bind coagulation factors due to mutation of the FX binding site in hexon. (B) Neutralization of vectors by a human serum that has high antibody-mediated neutralizing activity (sample 18, heat-inactivated). (C) Effect of non-heat-inactivated serum on infectivity of Ad5.0 and Ad5.7 vectors. (D) Impact of blocking FX (with 80 μg/mL X-bp) on neutralization of Ad5.0 by non-heat-inactivated serum. T-tests were used to evaluate differences, followed by the sequential Holm-Sidak procedure for multiple comparisons to maintain the experimentwise type I error rate at ≤ 0.05. * Groups are significantly different (p ≤ 0.05). Each bar represents the mean of 3–4 replicates ± S.D.

Fig 2. Human IgM binds to Ad5, but is insufficient for neutralization.

(A) Binding of human IgM to Ad5.0-coated plates was assessed by ELISA. Sources of IgM were: IgM derived from pooled human serum, IgM from individual human sera 1 and 2, or IgM from sera of two IgM myeloma patients. (B) Infectivity of Ad5.0 vector after incubation with human IgG (from serum 18) or IgM (from sera 1 and 2) at either 150 or 500 μg/mL. Statistical significance was evaluated by one-way ANOVA, followed by the Holm-Sidak post-hoc test. * Groups are significantly different (p ≤ 0.05). Each bar represents the mean of 3 replicates ± S.D.

Fig 3. Human IgM and IgG enhance neutralization of vectors via complement, and FX protects vectors against this neutralization.

Vectors were incubated with serum from Rag1^-/- mice, with or without the addition of human antibodies, and infectivity was assessed. (A) Infectivity of Ad5.0 (control vector) and Ad5.7 (non-FX-binding vector) after incubation with 150 μg/mL of human IgM (purified from sera 1 and 2) in the presence of mouse complement (serum from Rag1^-/- mice). (B) Infectivity of vectors after incubation with 50 μg/mL of human IgG (purified from serum 18) in the presence of active mouse complement, or after inactivating complement with 10 mM EDTA. The 100% reference lines indicate vector infectivity in medium without serum. Statistical significance was evaluated by one-way ANOVA, followed by the Holm-Sidak post-hoc test. * Groups are significantly different (p ≤ 0.05). Each bar represents the mean of 3 replicates ± S.D.

Fig 4. A hexon-chimeric vector with Ad48 hexon is neutralized by human serum, but not by mouse serum.

The chime ric Ad5:H48 vector is a luciferase-expressing Ad5 vector in which the seven hypervariable regions of Ad5 hexon have been replaced by the corresponding regions of Ad48 hexon. (A) Representative surface plasmon resonance sensorgrams showing binding of FX to immobilized Ad5.0, Ad48 and Ad5:H48 (black lines). For Ad5.0, binding curves were fitted using a 1:1 binding model (red lines). The kinetic binding affinity (K_D) of human FX for Ad5.0 was 0.82 nM, and the affinity of mouse FX for Ad5.0 was 0.42 nM. FX did not detectably bind to immobilized Ad5:H48 or Ad48. (B) Infectivity of Ad5:H48 vector following incubation with C57BL/6 mouse serum or individual human sera. To assess the role of complement, neutralization was also assessed with sera that had been heated to inactivate complement. (C) Infectivity of Ad5:H48 vector following incubation with serum in the presence or absence of EDTA to inactivate complement. The 100% reference lines indicate vector infectivity in medium without serum. Statistical significance was evaluated by t-test to compare the paired control and experimental groups, followed by the sequential Holm-Sidak procedure for multiple comparisons to control the experimentwise type I error rate at ≤ 0.05. * Groups are significantly different (p ≤ 0.05). Each bar represents the mean of 3 replicates ± S.D.

Fig 5. Interactions of human immunoglobulins and mouse complement with Ad5:H48.

(A) Binding of human IgM to Ad5:H48 was assessed by ELISA, using the same sources of IgM as in Fig 2A. (B) Binding of mouse IgM to Ad5:H48. (C) Ad5:H48 and Ad5.7 were incubated with mouse complement (serum from Rag1^-/- mice), with or without addition of 300 μg/mL of mouse IgM or human IgM, and infectivity was assessed. (D) Infectivity of Ad5:H48 and Ad5.7 after incubation with human IgG (purified from serum 18) in the presence of active mouse complement, or after inactivating complement with 10 mM EDTA. The 100% reference lines indicate vector infectivity in medium without serum. Statistical significance was evaluated by one-way ANOVA, followed by the Holm-Sidak post-hoc test. * Groups are significantly different (p ≤ 0.05). Each bar represents the mean of 3 replicates ± S.D.