Identification of adeno-associated virus variants for gene transfer into human neural cell types by parallel capsid screening

Human brain cells generated by in vitro cell programming provide exciting prospects for disease modeling, drug discovery and cell therapy. These applications frequently require efficient and clinically compliant tools for genetic modification of the cells. Recombinant adeno-associated viruses (AAVs) fulfill these prerequisites for a number of reasons, including the availability of a myriad of AAV capsid variants with distinct cell type specificity (also called tropism). Here, we harnessed a customizable parallel screening approach to assess a panel of natural or synthetic AAV capsid variants for their efficacy in lineage-related human neural cell types. We identified common lead candidates suited for the transduction of directly converted, early-stage induced neural stem cells (iNSCs), induced pluripotent stem cell (iPSC)-derived later-stage, radial glia-like neural progenitors, as well as differentiated astrocytic and mixed neuroglial cultures. We then selected a subset of these candidates for functional validation in iNSCs and iPSC-derived astrocytes, using shRNA-induced downregulation of the citrate transporter SLC25A1 and overexpression of the transcription factor NGN2 for proofs-of-concept. Our study provides a comparative overview of the susceptibility of different human cell programming-derived brain cell types to AAV transduction and a critical discussion of the assets and limitations of this specific AAV capsid screening approach.


Neurons
Three different types of iPSC-derived neuronal cultures were subjected to virus screening. LtNES were differentiated into neuroglial cultures by withdrawing the growth factors EGF and bFGF. Cells were differentiated in 1:1 DMEM/F12 : Neurobasal with 0.5x N2 supplement, 0.5x B27 supplement, 0.9 mg/ml D(+)-glucose, 10 µg/ml insulin and 1x Anti-Anti (Thermo Fisher Scientific), and cryo-preserved on day 14 of differentiation. For experiments, pre-differentiated cells were thawed in ltNES differentiation medium supplemented with 10 ng/ml BDNF and 10 ng/ml GDNF, seeded at a density of 2.5x10 4 cells per well for virus screening and 5.5x10 4 cells/cm 2 for cell type characterization, and further matured up until day 27 of differentiation. Alternatively, iPSCs were differentiated into neurons via overexpression of the transcription factor(s) NGN2 or ASCL1 plus DLX2 according to a previously published protocol. 24 On day 8 of forward programming, cells were replated and seeded at a density

RGL-NPCs and astrocytes
In order to derive highly purified astrocyte cultures, RGL-NPCs were seeded at a density of 9x10 4 cells/cm 2 on Matrigel-coated dishes (1:30). Differentiation was initiated by switching to N2 medium supplemented with 100 µg/ml apo-transferrin, 10 ng/ml human LIF and 10 ng/ml BMP-4 (Thermo Fisher Scientific). For virus screening, cells were replated on day 28 of astrocyte differentiation, by seeding 0.7x10 4 astrocytes per well. Virus screening was performed on days 29-31 of differentiation in 200 µl medium. For functional experiments, astrocytes were seeded at a density of 9x10 4 cells/cm 2 on day 14 of astrocyte differentiation and AAVs were transduced one week later. Upon AAV transduction, medium was changed to N2 medium (1x DMEM/F12 with 1x N2 supplement, 1.6 mg/ml D(+)glucose, 50 µg/ml insulin, 100 µg/ml apo-transferrin and 1x Pen/Strep) supplemented with 1x B27 supplement without vitamin A, 10 ng/ml BDNF and 10 µM Rock-inhibitor Y-27632 (Cell Guidance Systems), in order to support the survival of astrocytes as well as eventually emerging neurons.

iPSC-derived microglia
To generate human iPSdMiG, iPSCs were cultured on Geltrex-coated 6-well plates (180 µg/ml) in iPSbrew (Stemcell Technologies, Vancouver, Canada) medium and passaged using 0.5 mM EDTA (Sigma-Aldrich). To induce differentiation, embryoid bodies were generated by detaching intact iPSC colonies using 1 mg/ml collagenase (Thermo Fisher Scientific) dissolved in DMEM/F12. Detached colonies were cultured in suspension for four days before being seeded on poly-L-ornithine-plus fibronectin (Sigma-Aldrich)-coated culture plates. The differentiation of iPSdMiG was carried out according to a proprietary protocol of the LIFE & BRAIN GmbH (patent application number EP20162230). In short, neuroepithelial and hemogenic endothelial precursors were differentiated into neural and immature microglial cells within the same culture paradigm. After approximately six weeks of differentiation, mature microglia were released into the supernatant of the multilineage differentiation culture, from which they could be repeatedly harvested and plated in poly-L-lysine (Sigma-Aldrich)-coated cell culture plates in order to obtain homogeneous, adherent cultures. For AAV screening, 2.3x10 3 harvested iPSdMiG were seeded per well in 200 µl of their respective medium. Supplementary Table S1: Overview of peptide modifications inserted into the capsids of diverse AAV serotypes.

Supplementary Tables
a The peptide insertion sites are represented with flanking residues in the one-letter amino acids code.
Depending on the serotype, more than one insertion site has been targeted. Integrins printed in bold are the five family members with highest expression levels in iNSCs. Bar graphs represent means ± standard error with N = 3-9 independent samples per group. (b) qPCR-based expression profiling of the five integrins with highest gene counts in iNSCs across different human cell programming-derived cell types. Left panels in each graph depict the individual data of all cell types studied here, whilst the panels on the right depict the means of all well or poorly transduced types (tropism high or low, respectively). Wilcoxon signed rank tests were performed on the pooled datasets.