scAAVengr, a transcriptome-based pipeline for quantitative ranking of engineered AAVs with single-cell resolution

  1. Bilge E Öztürk
  2. Molly E Johnson
  3. Michael Kleyman
  4. Serhan Turunç
  5. Jing He
  6. Sara Jabalameli
  7. Zhouhuan Xi
  8. Meike Visel
  9. Valérie L Dufour
  10. Simone Iwabe
  11. Luis Felipe L Pompeo Marinho
  12. Gustavo D Aguirre
  13. José-Alain Sahel
  14. David V Schaffer
  15. Andreas R Pfenning
  16. John G Flannery
  17. William A Beltran
  18. William R Stauffer
  19. Leah C Byrne  Is a corresponding author
  1. Department of Ophthalmology, University of Pittsburgh, United States
  2. Computational Biology, School of Computer Science, Carnegie Mellon University, United States
  3. Department of Neurobiology, University of Pittsburgh, United States
  4. Eye Center of Xiangya Hospital, Hunan Key Laboratory of Ophthalmology, Central South University, China
  5. Helen Wills Neuroscience Institute, University of California, Berkeley, United States
  6. Division of Experimental Retinal Therapies, Department of Clinical Sciences & Advanced Medicine, School of Veterinary Medicine, University of Pennsylvania, United States
  7. Chemical Engineering, University of California, Berkeley, United States
  8. Vision Science, Herbert Wertheim School of Optometry, University of California Berkeley, United States
  9. Department of Bioengineering, University of Pittsburgh, United States
8 figures, 1 table and 5 additional files

Figures

Figure 1 with 4 supplements
scAAVengr pipeline.

(A) Overview of scAAVengr experimental workflow. An AAV library, consisting of variants packaged with a GFP transgene fused to unique barcodes (AAV-barcodes), was packaged, pooled, quantified by …

Figure 1—figure supplement 1
Directed evolution performed in dogs.

(A) Highly diverse (~1E + 7) libraries of AAV capsid variants were packaged such that each virus contained a genome encoding its own capsid. Libraries were pooled and injected intravitreally in …

Figure 1—figure supplement 2
Description of libraries used in the study.

The directed evolution plasmid library was used to package the directed evolution AAV library. Following screening in dogs, a subset of top-performing AAVs was repackaged as the DE subset library. A …

Figure 1—figure supplement 3
Injection of GFP-barcoded AAV library of top AAV variants in dogs.

(A) Intravitreal injection of GFP-barcode library in three dogs resulted in transgene expression 3 weeks after injection. (B) Imaging of GFP in a cross-section through the superior quadrant of the …

Figure 1—figure supplement 4
Performance of AAV variants following intravitreal injection of DE subset library in dog.

AAV-barcodes amplified from the ONL and RPE were subjected to Illumina sequencing to quantify the representation of each of the variants. Heat maps show the performance of variants, ranked on the …

Figure 2 with 1 supplement
Clustering and quantification of AAV-infected retinal cells.

(A) AAV-infected marmoset retinal cells. Maps of clustered cells from superior, inferior or macular retina show AAV infection. Cells of similar type cluster together. The cell type of each cluster …

Figure 2—figure supplement 1
Illustration of barcoding strategies.

For directed evolution secondary screening experiments, performed in bulk tissue from dogs, 20 vectors, along with an AAV2 control, were packaged individually with a ubiquitous CAG promoter driving …

Figure 3 with 1 supplement
Quantitative comparison of variant infection across retinal cell types.

(A) Percent of cells infected by AAV serotypes in marmoset and cynomolgus macaque retina. Heat maps show the percent of identified cells infected by each serotype in the screen, corrected by the AAV …

Figure 3—source data 1

Marmoset 1-Superior-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data1-v2.csv
Figure 3—source data 2

Marmoset 1-Inferior-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data2-v2.csv
Figure 3—source data 3

Marmoset 1-Macula cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data3-v2.csv
Figure 3—source data 4

Marmoset 1-Superior-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data4-v2.csv
Figure 3—source data 5

Marmoset 1-Inferior-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data5-v2.csv
Figure 3—source data 6

Marmoset 1-Macula-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data6-v2.csv
Figure 3—source data 7

Marmoset 2-Superior-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data7-v2.csv
Figure 3—source data 8

Marmoset 2-Inferior-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data8-v2.csv
Figure 3—source data 9

Marmoset 2-Macula cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data9-v2.csv
Figure 3—source data 10

Marmoset 2-Superior-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data10-v2.csv
Figure 3—source data 11

Marmoset 2-Inferior-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data11-v2.csv
Figure 3—source data 12

Marmoset 2-Macula-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data12-v2.csv
Figure 3—source data 13

Cyno-Peripheral-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data13-v2.csv
Figure 3—source data 14

Cyno-Macula cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data14-v2.csv
Figure 3—source data 15

Cyno-Peripheral-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data15-v2.csv
Figure 3—source data 16

Cyno-Macula-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig3-data16-v2.csv
Figure 3—figure supplement 1
Numbers of AAV-infected cells.

Heat maps show the number of cells infected by each serotype of virus, across retinal regions, in marmosets and cynomolgus macaque. Percent of cells infected is shown in Figure 3. Numbers are …

Serotype performance across retinal regions.

(A) Scatter plots reveal that K912 is the overall best performing canine variant across retinal regions, while NHP26 is the best performing primate DE variant. Plots show the number of transcripts …

Dynamics of infection from multiple AAV serotypes.

(A) Upset plots show that multiple AAV serotypes can infect the same retinal cell, although the majority of retinal cells are infected by the top performing variants. Plots are shown across multiple …

Figure 6 with 1 supplement
K912 expression in primate retina.

(A–G) GFP expression in a cynomolgus macaque injected with ~2.6E + 12 vg of K912-scCAG-GFP. (A) GFP expression in a flatmounted cynomolgus macaque retina 2.5 months after injection. (B) GFP …

Figure 6—figure supplement 1
Expression of K912-CAG-GFP in cynomolgus macaque retina.

Intravitreal injection of K912-scCAG-GFP results in GFP expression in the central macaque retina. Expression results in bright expression in a perifoveal ring, and in punctate regions around retinal …

scAAVengr quantified the performance of AAV variants in mouse brain, heart, and liver following systemic injection of libraries.

(A) Maps of clustered AAV-infected cells from brain, heart, and liver. The cell type of each cluster is indicated by color. AAV-infected cells are shown in black. Maps are shown for cells processed …

Quantitative comparison of variant infection across cell types in mouse brain, heart and liver.

(A) Percent of cells infected by AAV serotypes. Heat maps show the percent of identified cells infected by each serotype in the screen, corrected by the AAV dilution factor, for each retinal cell …

Figure 8—source data 1

Brain-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data1-v2.csv
Figure 8—source data 2

Brain-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data2-v2.csv
Figure 8—source data 3

Brain FACS-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data3-v2.csv
Figure 8—source data 4

Brain FACS-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data4-v2.csv
Figure 8—source data 5

Heart-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data5-v2.csv
Figure 8—source data 6

Heart-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data6-v2.csv
Figure 8—source data 7

Heart FACS-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data7-v2.csv
Figure 8—source data 8

Heart FACS-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data8-v2.csv
Figure 8—source data 9

Liver-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data9-v2.csv
Figure 8—source data 10

Liver-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data10-v2.csv
Figure 8—source data 11

Liver FACS-Percent cells.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data11-v2.csv
Figure 8—source data 12

Liver FACS-Transcripts.

https://cdn.elifesciences.org/articles/64175/elife-64175-fig8-data12-v2.csv

Tables

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Escherichia coli)NEB 5-alphaNEBC2987HCompetent cells
Strain, strain background (Escherichia coli)MegaX DH10B T1ThermoFisherC640003Electrocompetent cells
Strain, strain background (Mus musculus)C57Bl/6 JJackson LaboratoriesStock No: 000664RRID:IMSR_JAX:000664
Cell line (Homo-sapiens)293AAVCell BiolabsAAV-100RRID:CVCL_KA64
Cell line (Homo-sapiens)HEK293TATCCCRL-1573RRID:CVCL_0045
AntibodyLectin PNA (Peanut agglutinin)Molecular ProbesL32459RRID:AB_2315178(1:200)
AntibodyAnti-GFP, (rabbit polyclonal)Thermofisher ScientificA11122RRID:AB_221569(1:250)
Recombinant DNA reagentpX601-AAV-CMV::NLS-SaCas9-NLS-3xHA-bGHpA;U6::BsaI-sgRNAAddgenePlasmid #61591RRID:Addgene_61591A single vector AAV-Cas9 system containing SaCas9 and its sgRNA
Recombinant DNA reagentscCAG-eGFP-Barcode-bghPolyAThis paperByrne Lab, see materials and methods, under GFP barcoded AAV library construction.
Recombinant DNA reagentAAV librariesReferences: ~ 588 peptide insertion library (Müller et al., 2003), AAV2-Loopswap library (Koerber et al., 2008) AAV2-ErrorProne library (Koerber et al., 2006) SCHEMA library (Ojala et al., 2018).
Commercial assay or kitQuickTiter AAV Quantitation KitCell biolabsVPK-145AAV quantification kit
Commercial assay or kitQiagen DNeasy Blood and Tissue KitQiagenCat. No. / ID: 69504DNA extraction kit
Commercial assay or kitAllPrep DNA/RNA Micro KitQiagenCat. No. / ID: 80284DNA/RNA extraction kit
Commercial assay or kitNeural Tissue Dissociation Kit for postnatal neuronsMACS Miltenyi130-094-802Retina dissociation kit
Commercial assay or kitAdult Brain Tissue Dissociation KitMACS Miltenyi130-107-677Brain dissociation kit
Commercial assay or kitMulti Tissue Dissociation Kit 2MACS Miltenyi130-110-203Heart dissociation kit
Commercial assay or kitLiver Dissociation KitMACS Miltenyi130-105-807Liver dissociation kit
Commercial assay or kitChromiumNext GEMSingle Cell 3Reagent Kits v310x GenomicsPN-1000075, PN-1000073, PN-120262
Commercial assay or kitChromiumNext GEMSingle Cell 3Reagent Kits v3.1 (Dual Index)10x GenomicsPN-1000268, PN-1000120, PN-1000215
Commercial assay or kitTargeted Gene Expression Reagent Kit10x GenomicsPN-1000248, PN-1000249
Chemical compound, drugCyclosporineGENGRAF6 mg/kg
Chemical compound, drugMeloxicamVivlodex0.2 mg/kg
Chemical compound, drugTriamcinolone Acetonide (Kenalog 40)Bristol-Myers Squibb
Software, algorithmSTARsoloA.Dobin et al., STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).RRID:SCR_021542v2.7
Software, algorithmCell Ranger10x GenomicsRRID:SCR_017344v3
Software, algorithmDropletUtilsA.T. L. Lun et al., EmptyDrops: distinguishing cells from empty droplets in droplet-based single-cell RNA sequencing data. Genome Biol 20, 63 (2019).v1.4.3
Software, algorithmSoupXM.D. Young, S. Behjati, SoupX removes ambient RNA contamination from droplet based single cell RNA sequencing data. bioRxiv, (2020).RRID:SCR_019193v0.3.1
Software, algorithmSCDSA.S. Bais, D. Kostka, scds: computational annotation of doublets in single-cell RNA sequencing data.Bioinformatics 36, 1150–1158 (2020).RRID:SCR_021541v1.0.0
Software, algorithmScranLun ATL, McCarthy DJ, Marioni JC. A step-by-step workflow for low-level analysis of single-cell RNA-seq data with Bioconductor. F1000Research 5: 2122 (2016).RRID:SCR_016944v1.12.1
Software, algorithmALRAG.C. Linderman, J. Zhao, Y. Kluger, Zero-preserving imputation of scRNA-seq data using low-rank approximation. bioRxiv, (2018).RRID:SCR_021540v1.0
Software, algorithmScanpyF.A. Wolf, P. Angerer, F. J. Theis, SCANPY: large-scale single-cell gene expression data analysis. GenomeBiol 19, 15 (2018).RRID:SCR_018139v1.4.4.post1
Software, algorithmScanoramaHie B, Bryson B, Berger B. Efficient integration of heterogeneous single-cell transcriptomes using Scanorama. Nature Biotechnology 37: 685–691 (2019).RRID:SCR_021539v1.2
Software, algorithmSalmonR.Patro, G. Duggal, M. I. Love, R. A. Irizarry, C. Kingsford, Salmon provides fast and bias-aware quantification of transcript expression. Nat Methods 14, 417–419 (2017).RRID:SCR_017036v0.9.1
Software, algorithmCRISPResso2Clement K, Rees H, Canver MC, Gehrke JM, Farouni R, Hsu JY, Cole MA, Liu DR, Joung JK, Bauer DE, Pinello L. CRISPResso2 provides accurate and rapid genome editing sequence analysis. Nature Biotechnology 37: 224–226 (2019).RRID:SCR_021538v2.0.34

Additional files

Supplementary file 1

Table with Summary of injections performed in dogs and primates.

AAV selection rounds in canines 2b and 5b were repeated selections of the previous rounds, which did not result in the amplification of AAV variants.

https://cdn.elifesciences.org/articles/64175/elife-64175-supp1-v2.docx
Supplementary file 2

Statistical Analysis.

A. Friedman’s test was conducted to determine differences across AAV variants. The test was run separately for each cell type as well as total cells combined. Marmoset and cynomolgus macaque samples were both used in the analysis (n = 8). Significant P-values < 0.05 are shown in bold red. S2.1 p-values resulting from a Friedman’s test using percent cells infected as the data points. Significant P-values < 0.05 are shown in bold red. S2.2 p-values resulting from a Friedman’s test using average transcripts per infected cell as the data points. Significant P-values < 0.05 are shown in bold red.

https://cdn.elifesciences.org/articles/64175/elife-64175-supp2-v2.docx
Supplementary file 3

p-values from a Wilcoxon signed-rank test.

A one-sided Wilcoxon signed-rank test was used for a pairwise comparison between K912 or NHP26 and the other variants. P-values were corrected using Benjamini-Hochberg correction method. Significant P-values < 0.05 are shown in bold red. S3.1 p-values resulting from a one-sided Wilcoxon signed-rank test using percent cells infected as the data points, comparing NHP12 and other variants. S3.2 p-values resulting from a one-sided Wilcoxon signed-rank test using percent cells infected as the data points, comparing NHP26 and other variants. S3.3 p-values resulting from a one-sided Wilcoxon signed-rank test using average transcripts per infected cell as the data points, comparing K912 and other variants. S3.4 p-values resulting from a one-sided Wilcoxon signed-rank test using average transcripts per infected cell as the data points, comparing NHP26 and other variants.

https://cdn.elifesciences.org/articles/64175/elife-64175-supp3-v2.docx
Supplementary file 4

List of primers used in the study.

https://cdn.elifesciences.org/articles/64175/elife-64175-supp4-v2.docx
Transparent reporting form
https://cdn.elifesciences.org/articles/64175/elife-64175-transrepform1-v2.pdf

Download links