Material and Methods

Cell culture

Spodoptera frugiperda Sf9 cells (Invitrogen, Cat#: 11496-015) were routinely sub-cultured every 2-3 days at 0.6-1 × 106 cell.mL-1 in serum-free Sf900-IITM SFM medium (Thermo Fisher Scientific) when the cell concentration reached 2-4 × 106cell.mL-1. Cells were cultured in shake flasks (Corning) using 10% (w/v) working volume and maintained at 27 °C in an Inova 44R shaking incubator (orbital motion diameter of 2.54 cm; Eppendorf) at 100 rpm.

Baculovirus amplification and storage

Two recombinant baculoviruses (rBAC) were used for AAV production, one incorporating a GFP transgene flanked by inverted terminal repeats of AAV serotype 2 (AAV2) and under control of the cytomegalovirus promoter (hereby named rBAC-GFP, kindly provided by Généthon) and a second rBAC carrying AAV2 rep and cap genes (hereby named rBAC-REP/CAP), produced in-house using Addgene plasmid #65214 (Pais et al., 2019; Smith et al., 2009). Amplification of baculovirus stocks and storage was performed as described previously (Virgolini et al., 2022). Baculovirus titers were determined using the MTT assay as described elsewhere (Mena et al., 2003; Roldão et al., 2009).

AAV production

AAV production was carried out in a 0.5 L stirred tank bioreactor (Sartorius BIOSTAT Qplus) by infecting Sf9 cells at 2 × 106 cell.mL-1 with rBAC-REP/CAP and rBAC-GFP, both at a multiplicity of infection (MOI) of 0.05 pfu/cell. Each vessel was equipped with one Rushton impeller; gas flow of 0.01 VVM was supplied through a ring sparger. pO2(partial pressure of oxygen) setpoint of 30% of air saturation was set, which was achieved through varying the agitation rate (70 - 300 rpm) and percentage of O2 in the gas flow (0 - 100 % of O2).
Cultures were maintained for up to 96 hours post-infection (hpi). Samples for assessment of cell concentration, viability, intracellular AAV titer and metabolite concentration were taken every 24 hours. For scRNA-seq analysis, samples were taken at 0, 10 and 24 hpi.

Analytics

Cell concentration and viability

Cell concentration and viability were determined with the trypan blue dye exclusion method (Tennant, 1964) using the Cedex HiRes Analyzer (Roche).

Intracellular AAV viral genomes quantification

Cell culture samples were collected and intracellular AAV titer quantified as described elsewhere (Pais et al., 2019). Intracellular AAV viral genomes (VG) were quantified by real-time quantitative PCR (RT-qPCR), as established previously (Virgolini et al., 2022).

Single-cell RNA sequencing

Single-cell isolation and sample processing for scRNA-seq was performed, using the BD RhapsodyTM Express Single-Cell Analysis System (BD Biosciences) according to manufacturer’s instructions. In short, cultured cells were centrifuged (300 × g, 4 °C, 5 min), washed and strained (30 µm mesh size – CellTrics®) prior dilution to the recommended cell concentration to target 6,000 single cells. Cells were then captured in nanowell-containing cartridges, lysed, and the released mRNA isolated using poly(dT)-coated magnetic beads.
Sequencing libraries were prepared using the mRNA Whole Transcriptome Analysis (WTA) Library Preparation Protocol (BD Biosciences), as recommended by the manufacturer, and using unique index primers for each library. The quality of each library was assessed using a high-sensitivity D5000 kit (Agilent) and quantification was done by Qubit analysis (Thermo Fisher Scientific). Finally, libraries were pooled to achieve 40,000 reads per cell and sent for sequencing (Illumina NovaSeq6000) elsewhere, spiked with 20% PhiX.

Single-cell RNA sequencing data analysis

UMI count matrix generation

Raw FASTQ files were processed using the BD Rhapsody™ WTA Analysis Pipeline (Seven Bridges Genomics), using default parameters. A customised reference genome, which included reference sequences ofS. frugiperda (RefSeq assembly accession: GCF_011064685.1)( Xiao et al., 2020), Autographa californica multiple nucleopolyhedrovirus (ViralProj14023)(Ayres et al., 1994) and AAV transgene sequences (rep , cap and gfp ), was supplied for mapping using STAR version 2.5.2b (Dobin et al., 2013). Finally, recursive substitution error correction (RSEC)-adjusted molecule count matrices were generated and used for downstream analysis.

Sample pre-processing

Downstream analysis was performed in R (v4.2.1) using the Seurat package (v4.1.1) (Hao et al., 2021); default parameters were used to create the Seurat objects. Low-quality cells (≥ 5% mitochondrial UMI counts per cell) were removed.

Cell cycle scoring

To determine the cell cycle phase of each cell, a score indicating the likelihood of cells being in either S or G2/M phase was assigned, based on the supplied reference gene list for respective phases from Seurat (according to mouse reference genes published from Tirosh et al., 2016). The list of mouse cell cycle genes was associated to the S. frugiperda genome using a protein BLAST search (e-value cut-off 0.01). Then, identified sequences were blasted back to the S. frugiperdagenome. Genes validated after both steps were considered for cell cycle association.

Seurat analysis

To evaluate cell heterogeneity along infection, data sets (from 0, 10 and 24 hpi samples) were merged prior to log-normalization. Next, 2,000 variable genes of each sample were identified using the “vst” method. Data was scaled, regressing effects caused by the total number of UMIs and dimensionality reduced using principal component analysis (PCA) using all variable genes. A specified number of principal components was chosen for subsequent UMAP analysis. Relative gene expression, differential expression analysis and gene correlation analysis were performed using the data slot of the Seurat object. Gene markers for clusters were identified using the FindMarkers function and the MAST test (Finak et al., 2015). Genes with an absolute expression change of at least 1.5-fold and a BH-adjusted p -value ≤ 0.05 were deemed significant.

Trajectory analysis and enrichment analysis

To understand the changes in cellular response along infection, we utilized the 10 hpi timepoint to conduct trajectory analysis within the Monocle3 package v.1.2.9 (Trapnell et al., 2014). First, the processed 10 hpi Seurat object was transformed to a Monocle object using the appropriate function within the SeuratWrappers package v.0.3.0. Next, clusters were identified using the Leiden community detection and a resolution of 0.002. A trajectory is learned, and cells ordered according to pseudotime. Finally, genes correlated with the progression of cells along the trajectory were identified. Genes were deemed significant if the q- value was < 0.01 and the average expression of the respective gene in the Seurat object was > 0.5.
Overrepresentation of gene ontology (GO) terms within the gene list found to have a significant difference along pseudotime were identified using ClusterProfiler v4.0.5 (Yu et al., 2012) and a GO term reference list established previously (Virgolini et al., 2022). Terms with an adjusted p -value < 0.5 were deemed significant.