Pulmonary Targeting of Adeno-associated Viral Vectors by Next-generation Sequencing-guided Screening of Random Capsid Displayed Peptide Libraries

Abstract

Vectors mediating strong, durable, and tissue-specific transgene expression are mandatory for safe and effective gene therapy. In settings requiring systemic vector administration, the availability of suited vectors is extremely limited. Here, we present a strategy to select vectors with true specificity for a target tissue from random peptide libraries displayed on adeno-associated virus (AAV) by screening the library under circulation conditions in a murine model. Guiding the in vivo screening by next-generation sequencing, we were able to monitor the selection kinetics and to determine the right time point to discontinue the screening process. The establishment of different rating scores enabled us to identify the most specifically enriched AAV capsid candidates. As proof of concept, a capsid variant was selected that specifically and very efficiently delivers genes to the endothelium of the pulmonary vasculature after intravenous administration. This technical approach of selecting target-specific vectors in vivo is applicable to any given tissue of interest and therefore has broad implications in translational research and medicine.

Figure 1

Next-generation sequencing (NGS)-guided in vivo selection of a random X₇-peptide adeno-associated virus (AAV) display library for identification of efficient and specific capsid targeting peptides. A random AAV display peptide library was injected intravenously in mice. Two (first selection round) or 6 (selection rounds 2–5) days after library administration, total DNA was isolated from the tissue of interest. The viral DNA containing the random library oligonucleotides of particles enriched in the target tissue was amplified by nested polymerase chain reaction (PCR) and cloned into library plasmid backbones to generate a secondary AAV display peptide library by transfection of AAV producer cells. The secondary library was used for a subsequent round of selection. NGS was used to analyze the selected oligonucleotide sequences after PCR-amplification from total DNA, isolated from murine lungs. Median of the changes in relative abundance of the 20 most frequent sequences (T20 Median) between the individual rounds was determined. As this median value sharply declined between rounds four and five, the selection process was stopped, and PCR-amplified virus DNA from the target and off-target organs was analyzed. Specificity and efficacy of dominant capsid variants was scored and AAV2-ESGHGYF was selected for further validation.

Figure 2

Scoring of dominant capsid variants for specificity and efficacy. After five selection rounds, the targeting peptides accounting for >99% of all reads from the lung in the next-generation sequencing analysis were scored. Peptides appear in order of their relative abundance in the lung from left to right, ranging from PRSADLA (32.74%) to GGDLSRA (0.02%). Three of the displayed peptides deviate in length. (a) The enrichment score E describes for the target tissue the change in relative abundance from selections round four to five within a range from 0 to 1. Peptides with the same relative abundance in both rounds yield a value of 0.5 (dashed line), highly enriched sequences yield a value close to 1. (b) The general specificity score GS is a combination of the organ-specific S scores and describes the target specificity of a peptide. Peptides with the same relative abundance in lung and all three control organs score at 0.125 (dashed line). With increasing specificity, the score increases toward 1. (c) The combined score C is generated by multiplying E and GS scores and describes the peptide performance regarding specificity and efficacy with an ideal value of 1. A peptide that doubles its relative abundance from round four to five and shows only half of the relative abundance observed in the lung in all three off-target organs scores at 0.198 (dashed line). ESGHGYF was the only analyzed peptide exceeding these criteria.

Figure 3

Luminescence imaging of mice after intravenous injection of AAV-vectors carrying a luciferase reporter gene. (a) Images were taken 14 days after i.v. administration of 5 × 10¹⁰ gp/mouse of vectors displaying the peptide ESGHGYF selected for lung tropism, the random control peptide CVGSPCG, or wild-type AAV2 capsids, respectively. Mice were imaged from the back (left panel), from the side (intermediate panel) and from the front (right panel). Images show representative examples of >3 animals per group. (b) Sagittal, coronal, and transaxial sections (left panel) and three-dimensional reconstruction (right panel) of the luminescence images of a mouse injected with AAV2-ESGHGYF vector (as in a), obtained by measuring different wavelengths of the emitted light, confirming the lung as source of luminescence. (c) Single organ of an ESGHGYF-injected mouse (as in a), measured ex vivo immediately after tissue explantation. (d) ESGHGYF-displaying vector was administered and long-term transgene expression was analyzed by bioluminescence imaging at 14-time points during a 244 days period (n = 2 animals). Upper panel: Imaged mice, rainbow scale as in a. lower panel: Values of luminescence/ region of interest.

Figure 4

Determination of luciferase activity from tissue lysates. Luciferase transgene expression analysis in tissue lysates 14 days after vector administration. Data are shown as bars (mean) with plotted individual data points (n = 3 animals per group). (a) Organ distribution of luciferase transgene expression mediated by the three vectors as detailed in Figure 3a. (b) Direct comparison of transgene expression mediated by vectors carrying wild-type AAV2 capsids (AAV2-WT), CVGSPCG-displaying control capsids or ESGHGYF-displaying lung-targeted capsids in three organs in which transgene expression could be detected by bioluminescence imaging (heart, liver, and lung).

Figure 5

Biodistribution of recombinant adeno-associated virus (AAV) vectors. Amounts of vectors were determined by quantitative real-time PCR of vector genomes. All experiments were performed after i.v. administration of 5 × 10¹⁰ gp/mouse. Data are shown as bars (mean) with plotted individual data points (n = 3–4 animals per group). (a) Distribution of wild-type AAV2 (left) and AAV2-ESGHGYF (middle) in seven analyzed organs, 2 minutes after vector administration, reflecting tissue homing of injected particles. Amount of circulating particles recovered from left ventricular blood (right). (b) Distribution of genomes delivered by AAV2-ESGHGYF or control vectors (wild-type AAV2 and AAV2-CVGSPCG), respectively, 14 days after vector administration. (c) Separate depiction of vector genome quantities in the three organs containing the highest amount of vector genomes (liver, spleen, and lung) comparing the three capsid variants used for gene delivery.

Figure 6

AAV2-ESGHGYF-mediated specific gene delivery to pulmonary vasculature endothelial cells. Paraffin-embedded tissue sections of mouse lungs and control organs were stained by immunohistochemistry 14 days after i.v. vector administration of 1 × 10¹¹ gp/mouse (AAV2-ESGHGYF or wild-type AAV2 as control). Each panel a and b displays a representative area of the tissue of interest, shown in different magnifications. Size of scale bars: 100 µm at 20× magnification; 50 µm at 40× magnification; and 20 µm at 100× magnification. (a) Representative area of the lung stained with an antibody against the transgene product GFP. Red arrow: Example of endothelial cells of larger pulmonary vessel; Blue arrow: Example of endothelial cells of pulmonary microvasculature. (b) Representative area of the liver stained with an antibody against GFP. Black arrows: examples of transduced hepatocytes (c) serial sections of a representative area of the lung of mice injected with rAAV2-ESGHGYF, stained with antibodies against the endothelial marker CD31 and against the transgene product GFP.