Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Aug 20;32(3):101327.
doi: 10.1016/j.omtm.2024.101327. eCollection 2024 Sep 12.

Predictive power of deleterious single amino acid changes to infer on AAV2 and AAV2-13 capsids fitness

Tiziana La Bella  1   2 , Bérangère Bertin  1   2 , Ante Mihaljevic  1   2 , Justine Nozi  1   2 , Patrice Vidal  1   2 , Sandrine Imbeaud  3 , Jean-Charles Nault  3   4 , Jessica Zucman-Rossi  3   5 , Giuseppe Ronzitti  1   2
Affiliations

Predictive power of deleterious single amino acid changes to infer on AAV2 and AAV2-13 capsids fitness

Tiziana La Bella et al. Mol Ther Methods Clin Dev. .

Abstract

Adeno-associated virus (AAV) is the most widely used vector for in vivo gene transfer. A major limitation of capsid engineering is the incomplete understanding of the consequences of multiple amino acid variations on AAV capsid stability resulting in high frequency of non-viable capsids. In this context, the study of natural AAV variants can provide valuable insights into capsid regions that exhibit greater tolerance to mutations. Here, the characterization of AAV2 variants and the analysis of two public capsid libraries highlighted common features associated with deleterious mutations, suggesting that the impact of mutations on capsid viability is strictly dependent on their 3D location within the capsid structure. We developed a novel prediction method to infer the fitness of AAV2 variants containing multiple amino acid variations with 98% sensitivity, 98% accuracy, and 95% specificity. This novel approach might streamline the development of AAV vector libraries enriched in viable capsids, thus accelerating the identification of therapeutic candidates among engineered capsids.

Keywords: AAV vector; AAV viability; adeno-associated virus; capsid engineering; deleterious mutation; fitness; hyper variable regions; prediction model; wild-type variant.

PubMed Disclaimer

Conflict of interest statement

T.L.B., P.V., J.N., G.R., S.I., J.-C.N., and J.Z.-R. are authors in patents related to capsids development. G.R. is associate editor of the journal Molecular Therapy Methods & Clinical Development.

Figures

None
Graphical abstract
Figure 1
Figure 1
Manufacturability and mutations landscape of WT AAV variants (A) Production of AAV variants in 50 mL of HEK293T cells growing in suspension. Titers of production are expressed as vector genomes per mL (vg/mL) and normalized as fold change compared with the AAV2 control capsid. Each variant was independently produced twice. The mean value of the titers and SD are indicated per each variant. Black and red dotted lines represent the level of AAV2 production and the production threshold, respectively. Variants labeled in red are characterized by large indels or frameshift mutations. Phylogenetic tree on VP1 coding protein region of WT AAV variants is represented on the bottom. The tree was constructed using the neighbor joining method based on the evolutionary distances computed by the Poisson correction method. The representation is toggled for topology scale. (B) Landscape of mutations identified in WT AAV variants. AAV2 sequence is used as reference, common (black) and unique (red) mutations are listed below each position. The start codons of VP proteins are highlighted in dark gray. Variants with indels and frameshift mutations are not represented. (C) Median viability s’ score for all mutations belonging to viable (n = 40) and non-viable capsids (n = 9). Statistical analysis was performed using Wilcoxon rank-sum test. (D) Median viability s’ score for common and unique mutations belonging to viable (n = 40) and non-viable capsids (n = 9). Statistical analysis was performed using Wilcoxon rank-sum test. (E) Frequency distribution represented as density of the s’ scores of common and unique mutations in WT AAV variants. Rug plot below each density plot shows the distribution of s’ data. The vertical dashed line represents the threshold of potentially deleterious mutations. Mutations with s’ > 4 are not represented in the graph. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Figure 2
Figure 2
Representative features of deleterious mutations (A) Scatter plot showing the s’ scores associated to all mutations in WT AAV variants along the VP sequence. Mutations are categorized in common (black) and unique (red) variations. Unique mutations belonging to non-viable capsids are represented as triangles. Mutations below the s’ threshold (horizontal dashed line) are labeled with the names of the corresponding AAV variants. VP positions and major amino acid features (residue accessibility, HVRs position, hydrophobicity, and polarity of the amino acids) are represented below the graph. (B) Median s’ scores of mutations in Ogden and colleagues’ dataset according to the accessibility of the residues, (C) localization within or outside HVRs, (D) hydrophobicity of the mutated amino acid, and (E) polarity of the R chain of the mutated residue. Median, interquartile range, minimum, and maximum values are represented in the boxplot. The violin plot shows the distribution of the data points. Outliers are indicated as individual dots. Statistical analysis was performed using Wilcoxon rank-sum test. (F) Density distribution of s’ scores of mutations in WT AAVs according to the accessibility of the residues, and (G) the localization within or outside the HVRs. The s’ score distributions of common and unique mutations are shown according to analyzed features. Rug plot below each density plot shows the distribution of s’ data. The vertical dashed line represents the threshold of potentially deleterious mutations. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Figure 3
Figure 3
Type of amino acid change and impact on capsid viability according to residue accessibility (A) Analysis of the major type of amino acid changes (variation in hydrophobicity) and deletions in accessible and buried residues. Mutations in the Ogden dataset were categorized in neutral and deleterious according to the s’ value, then the frequency of the different types of variations was analyzed in each group. Statistical analysis was performed using χ2 test with Monte Carlo simulation. (B) s’ score of mutations in the Ogden dataset according to major hydrophobic changes, deletions, and stop codon integrations in accessible and buried residues. Median, interquartile range, minimum, and maximum values are represented in the boxplot. The violin plot shows the distribution of the data points. Potentially deleterious mutations identified in WT AAV variants are represented as individual red dots. Statistical analysis was performed using Wilcoxon rank-sum test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Figure 4
Figure 4
Interplay between accessibility and HVR localization in viability prediction (A) Frequency of deleterious mutations in the Ogden dataset according to their localization in accessible and buried residues within or outside HVRs. Mutations were categorized in neutral and deleterious according to the s’ value threshold. Statistical analysis was performed using χ2 test with Monte Carlo simulation. (B) Median s’ score of unique mutations of WT AAVs in accessible and buried residues within or outside HVRs. Median, interquartile range, minimum, and maximum values are represented in the boxplot. Horizontal dashed line represents the threshold of potentially deleterious mutations. Statistical analysis was performed using Wilcoxon rank-sum test. (C) Frequency of accessible and buried residues along the VP sequence calculated per window of 20 amino acids. Buried residues were divided in 2 groups: buried within HVR and buried outside HVR. The black line over the graph represents the frequency of deleterious mutations calculated in each 20 amino acid window. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Figure 5
Figure 5
Experimental validation of potentially deleterious mutations on capsid viability Schematic representation of common (black) and unique (red) mutations with associated s’ score in VP sequence of hybrid capsids: (A) AAV2-2208, (B) AAV2-2128, and (C) AAV2-2141. Potentially deleterious mutations are represented below the s’ threshold (horizontal dashed line) and labeled with the name of the corresponding variation. The impact of potentially deleterious mutations was evaluated by reverse mutation of the amino acid residues. Luciferase activity assay was used to assess the transduction efficiency of the reverted mutants related to hybrid capsids: (D) AAV2-2208, (E) AAV2-2128, and (F) AAV2-2141. The corresponding WT AAV variant of each hybrid capsid was tested in parallel as control. Horizontal dotted line indicates the luciferase activity level of AAV2 reference capsid. Statistical analysis was performed using one-way ANOVA with Dunnett’s multiple comparisons test. The corresponding WT AAV variant was used as control group for mean comparison. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Figure 6
Figure 6
Validation of the prediction method on public datasets (A) Schematic representation of AAV2 capsid amino acid sequences with mutations described in two independent public datasets., (B) Median s’ score in viable and non-viable capsids classified according to experimental data. Each point represents the lowest s’ (s'min) value among all mutations of a capsid to underline the presence of deleterious mutations in the sequence. Statistical analysis was performed using Wilcoxon rank-sum test. (C) ROC curves representing the performance of the prediction method in the collection of WT variants (GNT series) and in public datasets. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Atchison R.W., Casto B.C., Hammon W.M. Adenovirus-Associated Defective Virus Particles. Science. 1965;149:754–756. doi: 10.1126/science.149.3685.754. - DOI - PubMed
    1. Balakrishnan B., Jayandharan G.R. Basic biology of adeno-associated virus (AAV) vectors used in gene therapy. Curr. Gene Ther. 2014;14:86–100. doi: 10.2174/1566523214666140302193709. - DOI - PubMed
    1. Cao M., You H., Hermonat P.L. The X gene of adeno-associated virus 2 (AAV2) is involved in viral DNA replication. PLoS One. 2014;9:e104596. doi: 10.1371/journal.pone.0104596. - DOI - PMC - PubMed
    1. Ogden P.J., Kelsic E.D., Sinai S., Church G.M. Comprehensive AAV capsid fitness landscape reveals a viral gene and enables machine-guided design. Science. 2019;366:1139–1143. doi: 10.1126/science.aaw2900. - DOI - PMC - PubMed
    1. Sonntag F., Schmidt K., Kleinschmidt J.A. A viral assembly factor promotes AAV2 capsid formation in the nucleolus. Proc. Natl. Acad. Sci. USA. 2010;107:10220–10225. doi: 10.1073/pnas.1001673107. - DOI - PMC - PubMed

LinkOut - more resources