Abstract

   Adaptive laboratory evolution has been used to improve production of
   influenza hemagglutinin (HA)-displaying virus-like particles (VLPs) in
   insect cells. However, little is known about the underlying biological
   mechanisms promoting higher HA-VLP expression in such adapted cell
   lines. In this article, we present a study of gene expression patterns
   associated with high-producer insect High Five cells adapted to neutral
   pH, in comparison to non-adapted cells, during expression of influenza
   HA-VLPs. RNA-seq shows a decrease in the amount of reads mapping to
   host cell genomes along infection, and an increase in those mapping to
   baculovirus and transgenes. A total of 1742 host cell genes were found
   differentially expressed between adapted and non-adapted cells
   throughout infection, 474 of those being either up- or down-regulated
   at both time points evaluated (12 and 24 h post-infection).
   Interestingly, while host cell genes were found up- and down-regulated
   in an approximately 1:1 ratio, all differentially expressed baculovirus
   genes were found to be down-regulated in infected adapted cells.
   Pathway analysis of differentially expressed genes revealed enrichment
   of ribosome biosynthesis and carbohydrate, amino acid, and lipid
   metabolism. In addition, oxidative phosphorylation and protein folding,
   sorting and degradation pathways were also found to be overrepresented.
   These findings contribute to our knowledge of biological mechanisms of
   insect cells during baculovirus-mediated transient expression and will
   assist the identification of potential engineering targets to increase
   recombinant protein production in the future.

   Keywords: insect cells, baculovirus expression system, influenza VLP,
   RNA sequencing, pathway analysis

1. Introduction

   The insect cell baculovirus expression vector system (IC-BEVS) relies
   on infection of insect cells with a recombinant baculovirus genetically
   modified to include a nucleic acid sequence encoding a gene of
   interest, commonly under the transcriptional control of very late,
   strong baculovirus promoters such as polh and p10 [[36]1]. The IC-BEVS
   system permits the expression recombinant protein at high titers and,
   importantly, glycosylation patterns similar to those of higher
   eukaryotes [[37]2]. This expression system is widely used in the
   production of enzymes, membrane proteins, viral capsids, and envelope
   proteins for use as vaccines or for analytical purposes [[38]3]. The
   success of the human papilloma virus vaccine (Cervarix^®) and the
   hemagglutinin-based influenza vaccine (Flublok^®) illustrates the
   utility of the system for biopharmaceutical manufacturing.

   In recent years, academic and industrial research groups have sought to
   improve knowledge of the molecular characteristics underpinning the
   efficient production of biopharmaceuticals using IC-BEVS [[39]4].
   Understanding the host cell’s metabolic regulation during expression of
   a foreign gene also facilitates the design and upscaling of a
   production process [[40]5,[41]6]. Next-generation sequencing
   technologies have accelerated the development of better expression
   systems to produce recombinant protein [[42]7]. Transcriptomics was
   applied to analyze the transcriptional changes of both Autographa
   californica multiple nucleopolyhedrovirus (AcMNPV) [[43]8] and
   alphanodavirus-free High Five cells (Tnms42) [[44]9] during protein
   expression using IC-BEVS. It has been shown that the number of viral
   transcripts increased significantly after the first 6 h of infection,
   concomitantly with a decrease in expression of host cell transcripts
   due to global shut-off of the host protein synthesis [[45]10]. More
   recently, comparative transcriptome analysis was conducted to study the
   differences in Tnms42 cell response upon expression of intracellular or
   secreted protein products using IC-BEVS, identifying key proteins as
   promising targets for achieving higher yields of protein secretion
   [[46]11].

   While genetic engineering of insect cells and/or baculovirus has been
   used to improve production yields [[47]12], strategies such as shifts
   in culture parameters to non-physiological values have been shown to
   impact the growth performance and recombinant protein production
   [[48]13]. The use of atypical culture conditions in insect cells has
   also been recently addressed through adaptive laboratory evolution
   approaches, i.e., by adapting cells to grow at such non-standard
   culture conditions, allowing higher recombinant protein yields both in
   stable cell lines [[49]14] and IC-BEVS [[50]15]. In the latter,
   adaptation of insect High Five cells to grow at neutral pH allowed a
   threefold improvement in cell-specific production rate of influenza
   virus-like particles (VLPs). However, relatively little is known about
   the mechanisms underlying the higher recombinant protein productivity
   achieved with this adapted cell line.

   In this study, the transcriptome of High Five cells adapted to neutral
   pH (established in [[51]15]), producing influenza VLPs using IC-BEVS,
   were assessed by RNA-seq and compared to those of parental, non-adapted
   insect High Five cells in order to gain an understanding of the
   mechanisms behind the higher productivity of the adapted cell line.

2. Materials and Methods

2.1. Cell Lines and Culture Media

   Insect High Five cells (Invitrogen), hereon referred to as non-adapted
   cells, and High Five cells adapted to neutral pH [[52]15], hereon
   referred to as adapted cells, were routinely sub-cultured to 0.3–0.5 ×
   10^6 cell.mL^−1 every 2–3 days when cell concentration reached 2–3 ×
   10^6 cell.mL^−1. Non-adapted cells were cultured in Insect-XPRESS^TM
   medium (Sartorius Stedim Biotech, Göttingen, Germany); adapted cells
   were cultured in cell culture medium composed of a 1:1 mixture of
   Insect-XPRESS^TM and chemically defined solution as previously reported
   [[53]14]. Both cell lines were cultured in 125–500 mL shake flasks (10%
   working volume) and maintained at 27 °C in a Inova 44R shaking
   incubator (Eppendorf, Hamburg, Germany) set to 100 RPM and with an
   orbital motion diameter of 2.54 cm.

2.2. Baculovirus Amplification and Storage

   Recombinant baculoviruses enclosing influenza capsid M1 from
   A/California/06/2009 H1N1 strain and hemagglutinin (HA) from
   A/Brisbane/59/2007 strain genes were kindly provided by Redbiotec AG
   (Schlieren, Switzerland). Amplification of baculovirus stocks was
   performed as described elsewhere [[54]16].

2.3. Production of Influenza HA-VLPs

   Influenza HA-VLPs were produced in 500 mL shake flasks (10% working
   volume) by infecting non-adapted cells and adapted cells at a cell
   concentration at infection (CCI) of 2 × 10^6 cell.mL^−1, with
   baculovirus using a multiplicity of infection (MOI) of 1 pfu/cell (n =
   3, number of replicates). Samples were taken daily for the
   determination of cell concentration and viability, metabolite
   concentration, and detection/relative quantification of M1 and HA
   proteins; for RNA-seq, samples were taken before infection, and at 12
   and 24 h post-infection (hpi) (further details in [55]Supplementary
   Figure S1).

2.4. Analytics

2.4.1. Cell Concentration and Viability

   Cell counting was performed in a Fuchs–Rosenthal hemocytometer chamber
   (Brand, Wertheim, Germany) and viability was assessed using the
   trypan-blue exclusion method [[56]17].

2.4.2. Metabolites Concentration

   For metabolite quantification, cell culture samples were centrifuged
   (300× g, 4 °C, 5 min) and supernatant collected and stored at −20 °C.
   Metabolite quantification was performed using Cedex Bio Analyzer 7100
   (Roche Diagnostics, Mannheim, Germany).

2.4.3. Baculovirus Titration

   Baculovirus titers were determined using the MTT assay as described
   elsewhere [[57]18,[58]19].

2.4.4. Western Blot

   For M1 and HA detection/relative quantification, cell culture samples
   were centrifuged (300× g, 4 °C, 5 min) and supernatant was collected
   and stored at 4 °C. Western blot analysis was performed as reported
   elsewhere [[59]14]. Briefly, for HA identification, a mouse monoclonal
   antibody (IRR, Manassas, VA, USA, FR-494—mouse monoclonal antibody to
   recombinant H1 HA from influenza A/Brisbane/59/2007 (H1N1)) was used at
   a dilution of 1:2000, and M1 protein was identified using a goat
   polyclonal antibody (Abcam, Cambridge, UK, Cat# ab20910) at a dilution
   of 1:2000. Secondary anti-mouse or anti-goat IgG antibodies conjugated
   with alkaline phosphatase were used at a dilution of 1:2000 for
   identification of HA and M1, respectively. Densitometry analysis of
   Western blot membranes (to assess relative productivity) was performed
   using the FIJI software [[60]20].

2.5. RNA Sequencing and Data Analysis

2.5.1. RNA Isolation and Library Preparation

   For RNA sequencing analysis, cell culture samples were centrifuged
   (300× g, 4 °C, 5 min) and pellets collected for RNA extraction using 1
   mL of Trizol (Invitrogen) and the Direct-zol RNA mini prep kit (Zymo
   Research) according to manufacturer’s instructions. RNA purity and
   quality were assessed by spectrophotometry (mySPEC equipment, VWR) and
   fragment analysis (Agilent).

   Library preparation and sequencing were performed elsewhere (Genewiz,
   Leipzig, Germany). In short, Poly (A) selection on total RNA (NEBNext^®
   Poly (A) mRNA Magnetic Isolation Module) was performed prior to
   strand-specific library preparation (NEBNext^® Ultra™ II Directional
   RNA Library Prep Kit for Illumina^®). Sequencing libraries were
   quality-checked using Qubit (Invitrogen) and a fragment analyzer
   (Agilent), and loaded on the Illumina NovaSeq 6000 system configured to
   yield a minimum of 25 million 2 × 150 bp Paired-End (PE) reads per
   sample.

2.5.2. Data Processing, Alignment and Counting

   Trimmomatic v0.36 was used to remove adapters and to perform quality
   trimming [[61]21] of the raw RNA-seq reads. The reads were subsequently
   aligned to a hybrid reference, comprising the insect Trichoplusia ni
   cell genome (Tnl; RefSeq assembly accession: GCF_003590095.1) [[62]22],
   the baculovirus, i.e., AcMNPV (RefSeq assembly accession:
   GCF_000838485.1, ViralProj14023) [[63]23], and the transgene (M1 and
   HA) sequences using STAR v2.7.3a [[64]24]. Those reads mapping to
   annotated genes were counted using HTSeq [[65]25].

2.5.3. Differential Expression Analysis

   The edgeR Bioconductor package [[66]26] was used to determine the
   number of differentially expressed genes between infected adapted cells
   and non-adapted cells. Count data were normalized to account for
   variation in the number of sequenced reads in each sample using the TMM
   method [[67]27]. To assess differential gene expression, the Fisher’s
   exact test was used to identify statistically significant differences
   in gene expression between selected groups [[68]28]. Genes with
   expression changes of at least 1.5-fold and with a false discovery rate
   (FDR)-adjusted p-value < 0.05 were considered to be differentially
   expressed.

2.5.4. Functional Annotation and Pathway Enrichment Analysis

   For gene annotation, the amino acid sequence of protein-coding genes
   with at least one read aligned was used as a query. Blastp search was
   applied in the NCBI nr protein database using Blast2GO OmicsBox
   software [[69]29]. No taxonomy filter was applied, and the E-value
   cutoff was set to 1.0 × 10^−3.

   Blast2GO was used to perform pathway enrichment analysis with Fisher’s
   exact test and the Gene Set Enrichment Analysis (GSEA) method
   [[70]30].These tests were used to identify statistically significant
   over-represented biological processes for each differentially expressed
   gene list. FDR was applied as multiple test correction method with a
   cut-off of 0.05 [[71]31].

3. Results

3.1. Production of Influenza HA-VLPs Using Adapted Insect High Five Cells

   Adapted and non-adapted cells were infected at the optimum conditions
   (CCI of 2 × 10^6 cell.mL^−1 and MOI of 1 pfu.cell^−1) previously
   identified in our laboratory [[72]14], and infection kinetics and HA
   expression were assessed throughout. Adapted cells maintained higher
   cell concentration and viability upon infection, with the onset of cell
   viability drop having a 24 h delay in comparison with non-adapted cells
   ([73]Figure 1A). The specific consumption or production rates for
   glucose (Glc), glutamine (Gln), and lactate (Lac) were similar in both
   cell lines, with the highest variation observed for glucose ([74]Figure
   1B). Importantly, M1 and HA proteins were identified by Western blot
   ([75]Supplementary Figure S2), with relative band intensities
   suggesting that expression of HA was approximately 2.5-fold higher in
   adapted cells while M1 expression remained similar in both cell lines
   ([76]Figure 1C), in agreement with previously reported data [[77]15].

Figure 1.

   [78]Figure 1
   [79]Open in a new tab

   HA-VLP production in non-adapted and adapted High Five insect cells
   using the baculovirus expression vector system. (A) Viable cell
   concentration (full circles) and cell viability (empty circles) after
   infection. (B) Specific glucose (rGlc) and glutamine (rGln) consumption
   and lactate (rLac) production (hence shown as negative) rates,
   estimated by linearization of metabolite concentration and integral of
   total cells, during infection. (C) Fold-change (adapted/non-adapted) in
   expression of HA and M1 proteins assessed by densitometry analysis of
   Western blot images (i.e., relative band intensity). Color code: yellow
   represents non-adapted cells, green represents adapted cells. Data are
   expressed as mean ± standard deviation of three culture replicates (n =
   3).

3.2. Gene Expression Profiling of Infected Adapted vs. Non-Adapted Cells

   To reveal the differences at the transcriptional level between adapted
   and non-adapted cells upon infection, quality-filtered reads were
   mapped to the reference genome and sequences. A total of 10,850 and
   10,987 reads (at 12 hpi), and 10,082 and 10,236 reads (at 24 hpi) were
   mapped in samples from adapted and non-adapted cells, respectively;
   from these, 10,389 (at 12 hpi) and 9620 (at 24 hpi) reads are
   coincident in both cell lines.

   A reduction in the proportion of reads mapping to the host cell genome
   was observed between 12 and 24 hpi, with a concomitant increase in
   reads mapping to baculovirus and transgenes; at 24 hpi, 51–60% of the
   total reads were mapped to baculovirus genome and 9–14% of the total
   reads were mapped to transgenes (HA + M1) ([80]Figure 2A). Differential
   gene expression analysis revealed that a total of 1742 host cell genes
   were differentially expressed between adapted and non-adapted cells
   throughout infection, 474 of those being differentially expressed at
   both time points evaluated ([81]Figure 2B). The top 20 most
   differentially expressed host cell genes at both 12 and 24 hpi are
   detailed in [82]Table 1 and [83]Table 2. The most up-regulated genes
   encode for transmembrane transports (e.g., organic cationic transports
   protein-like, 93-fold at 24 hpi) and proteins involved in proteolysis
   (e.g., xxa-Pro aminopeptidase 1, 756-fold at 24 hpi), and lipase
   activity (e.g., lipase member H-like, 112-fold at 12 hpi), whereas the
   most down-regulated genes encode for proteins involved in programmed
   cell death (e.g., homeobox protein abdominal-A homolog isoform X1,
   120-fold at 24 hpi), lipid transport (e.g., apolipophorins, 50-fold at
   24 hpi), innate immune response (e.g., protein toll-like, 37-fold at 12
   hpi), and oxidoreductase (e.g., cytochrome P450 9e2-like, 163-fold at
   12 hpi). Protein-coding genes involved in the regulation of
   transcription and signaling were found both up- and down-regulated.
   While host cell genes were found up- and down-regulated in an approx.
   1:1 ratio regardless of infection time, all baculovirus differentially
   expressed genes were found down-regulated in infected adapted cells
   ([84]Figure 2C).

Figure 2.

   [85]Figure 2
   [86]Open in a new tab

   Differential gene expression analysis. (A) Percentage of reads mapping
   to host cell (circles), baculovirus (triangles) and transgenes
   (diamonds) throughout infection. (B) Venn diagram showing the number of
   host-cell and baculovirus genes being differentially expressed (|log2
   (Fold-Change| > 0.58 and FDR < 0.05) between adapted and non-adapted
   cells, at 12 hpi, 24 hpi, and both timepoints simultaneously. (C)
   Volcano plot showing the distribution of differentially expressed genes
   with higher (in red) or lower (in blue) expression in adapted cells,
   compared to non-adapted cells. Number in boxes: number of
   differentially expressed genes.

Table 1.

   List of TOP 20 up-regulated genes in adapted cells vs. non-adapted
   cells.
   Gene ID Gene Name hpi FC logCPM Biological Process (P) or Molecular
   Function (M)
   LOC113500835 xaa-Pro aminopeptidase 1 12 273.6 3.1 hydrolase activity
   (M)
   24 756.1 1.6
   LOC113493104 potassium channel subfamily K member 18-like 12 102.1 1.7
   transmembrane transport (P)
   24 433.4 0.8
   LOC113505620 serine protease HP21 precursor 12 104.1 −0.1 proteolysis
   (P)
   24 116.2 −0.9
   LOC113496461 lipase member H-like 12 111.6 −1.6 lipid metabolic process
   (P)
   24 79.4 −1.3
   LOC113506781 organic cation transporter protein-like 12 91.9 −0.3
   transmembrane transport (P)
   24 92.8 −1.1
   LOC113504337 GATA zinc finger domain-containing protein 4-like 12 79.8
   −0.5 regulation of transcription (P)
   24 74.3 −1.3
   LOC113500515 alkylglycerol monooxygenase-like 12 97.8 1.0 lipid
   metabolic process (P)
   24 46.6 −0.3
   LOC113501396 transcription factor glial cells missing 2-like 12 108.6
   −1.7 regulation of transcription (P)
   24 29.1 −2.2
   LOC113501938 monocarboxylate transporter 1-like 12 86.2 −1.8
   transmembrane transport (P)
   24 36.5 −1.9
   LOC113494762 octopamine receptor Oamb isoform X1 12 65.3 1.1 regulation
   of transcription (P)
   24 18.4 0.7
   LOC113495982 uncharacterized protein 12 7.7 −1.0 n.a.
   24 74.8 −1.3
   LOC113501619 organic cation transporter protein-like 12 43.4 −2.5
   transmembrane transport (P)
   24 24.1 −2.3
   LOC113506861 uncharacterized protein 12 37.4 −2.6 n.a.
   24 24.0 −2.3
   LOC113492056 glucose dehydrogenase [FAD, quinone]-like 12 34.6 −2.7
   oxidoreductase activity (M)
   24 26.4 −2.2
   LOC113495496 thyrotropin-releasing hormone receptor-like isoform X1 12
   3.8 −0.2 signaling (P)
   24 51.7 1–6
   LOC113500144 uncharacterized protein 12 39.2 −0.2 n.a.
   24 14.2 −1.6
   LOC113506931 proton-coupled amino acid transporter-like protein
   pathetic 12 36.3 −1.4 transmembrane transport (P)
   24 6.1 −1.2
   LOC113500825 acid sphingomyelinase-like phosphodiesterase 3a 12 4.5
   −0.9 hydrolase activity (M)
   24 34.2 −2.1
   LOC113498260 mitochondrial import receptor subunit TOM40 homolog 12
   29.8 0.3 transmembrane transport (P)
   24 7.7 −0.6
   LOC113500192 glutamate receptor ionotropic, kainate 2-like 12 18.8 3.0
   regulation of transcription (P)
   24 18.4 2.2
   [87]Open in a new tab

   hpi: hours post-infection; FC: Fold-Change (adapted/non-adapted); CPM:
   Copy Per Million; n.a.: not available.

Table 2.

   List of TOP-20 down-regulated genes in non-adapted vs. adapted cells.
   Gene ID Gene name hpi FC logCPM Biological Process (P) or Molecular
   Function (M)
   LOC113496224 cuticle protein 16.5-like 12 168.3 −1.0 structural
   molecule activity (M)
   24 84.3 −1.4
   LOC113492802 juvenile hormone esterase-like 12 11.5 −0.1 hydrolase
   activity (M)
   24 165.3 −0.5
   LOC113495404 cytochrome P450 9e2-like 12 163.0 −1.1 oxidoreductase
   activity (M)
   24 13.6 −1.3
   LOC113496454 homeobox protein abdominal-A homolog isoform X1 12 41.7
   0.7 programmed cell death (P)
   24 120.1 −0.9
   LOC113508341 uncharacterized protein 12 31.2 1.2 n.a.
   24 83.5 0.2
   LOC113498352 apolipophorins isoform X2 12 16.2 1.3 lipid transport (P)
   24 49.9 0.2
   LOC113493958 solute carrier family 15 member 1-like 12 36.2 −1.3
   transmembrane transport (P)
   24 26.3 −2.3
   LOC113496890 Acanthoscurrin-1-like 12 25.3 −1.7 n.a.
   24 34.5 −2.1
   LOC113501789 uncharacterized protein 12 19.4 2.7 anatomical structure
   development (P)
   24 31.2 1.3
   LOC113499613 gloverin-like 12 21.3 2.9 n.a.
   24 26.4 1.5
   LOC113503497 chondroitin proteoglycan 2-like 12 14.2 −0.7 chitin
   binding (M)
   24 31.3 −1.0
   LOC113496839 protein toll-like 12 36.6 0.3 immune system process (P)
   24 8.8 −1.3
   LOC113492804 esterase FE4-like 12 29.1 1.3 hydrolase activity (M)
   24 15.1 0.5
   LOC113494776 myb-like protein D 12 7.0 −2.3 n.a.
   24 34.2 −2.1
   LOC113494175 alpha-tocopherol transfer protein-like 12 12.7 −1.0 n.a.
   24 25.3 −2.3
   LOC113503288 acetylcholine receptor subunit alpha-like 1 12 23.4 −0.3
   signaling (P)
   24 14.4 −0.8
   LOC113507593 probable E3 ubiquitin-protein ligase bre1 isoform X1 12
   6.1 −1.0 n.a.
   24 30.2 −1.0
   LOC113503500 odorant receptor 67c-like 12 4.7 −1.8 signaling (P)
   24 31.3 −2.2
   LOC113505115 irregular chiasm C-roughest protein-like isoform X1 12
   16.5 −0.3 programmed cell death (P)
   24 15.3 −1.7
   LOC113506962 neural retina-specific leucine zipper protein-like 12 25.1
   2.3 regulation of transcription (P)
   24 5.5 1.1
   [88]Open in a new tab

   hpi: hours post-infection; FC: Fold-Change (non-adapted/adapted); CPM:
   Copy Per Million; n.a.: not available.

3.3. Pathway Enrichment Analysis

   To further understand the biological mechanisms behind adapted cells’
   higher productivity, pathway enrichment analysis was performed. KEGG
   analysis using Fisher’s exact test revealed that 14 pathways are
   enriched, two being up-regulated in adapted cells (i.e., neuroactive
   ligand-receptor interaction, ribosome) while the remaining 12 were
   down-regulated (including metabolism of xenobiotic and endogenous
   compounds, carbohydrates, amino acids, vitamins and lipids) ([89]Figure
   3A). Additional pathways were found to be enriched using the GSEA
   method, including those playing a role in protein processing (including
   proteasome and ubiquitin mediated proteolysis) and oxidative
   phosphorylation ([90]Supplementary Figure S3). For the pathways found
   to be enriched, heatmaps with hierarchical clustering of all genes
   (differentially expressed and not) were generated. Results suggest that
   there is no common trend in adapted and non-adapted samples clustering
   as infection progresses, as reflected in the different dendrograms
   obtained—see two examples in [91]Figure 3B.

Figure 3.

   [92]Figure 3
   [93]Open in a new tab

   Pathway enrichment analysis. (A) Pathway enrichment analysis of
   differentially expressed genes using Fisher’s exact test. Bar plots
   indicate enriched terms at 12 and 24 hpi (i.e., longer bars denote
   pathways more significantly enriched). Color code: red identifies
   pathways found up-regulated and blue identifies pathways found
   down-regulated in adapted cells, over non-adapted cells. (B) Heat map
   of tyrosine metabolism pathway and genes encoding for ribosome
   subunits; the z-score (gradient color code) is defined for all the
   genes found to be involved in these specific pathways (differentially
   expressed or not). BI denotes before infection.

4. Discussion

   In this work, we assessed the gene expression profile of insect High
   Five cells adapted to neutral pH during production of influenza HA-VLPs
   using IC-BEVS, and compared it to that of non-adapted cells.

   During infection, the number of reads mapping to the host cell genome
   decreased, whereas those aligned to baculovirus (AcMNPV) and transgene
   (M1 and HA) sequences increased. Such a trend is a consequence of the
   global takeover of the cellular transcription machinery by
   baculoviruses towards overproduction of viral proteins during infection
   [[94]4]. The percentages of reads aligned to host cells and virus
   genome throughout the course of infection is similar to those reported
   previously in other studies [[95]9,[96]11]. Differential expression
   analysis revealed that all baculovirus-derived genes were found to be
   down-regulated in adapted cells. Together with distinct onset cell
   viability drop, this suggests that adapted and non-adapted cells have
   different susceptibility to infection.

   Adapted and non-adapted cells respond differently to baculovirus
   infection as reflected by the major differences observed in their gene
   expression profiles. For instance, genes involved in lipid metabolism
   processes (i.e., biosynthesis, hydrolysis, and transport) were found to
   be either up- or down-regulated in adapted cells at both 12 and 24 hpi.
   Insect cells are known to have a limited lipid metabolism, reflected in
   their limited capacity in synthesizing, desaturating and elongating
   fatty acids [[97]32], and lipid deprivation is linked to cell
   degeneration and impairment in the production of baculovirus [[98]33].
   Moreover, lipids such as cholesterol are especially important during
   the production of enveloped viral particles in mammalian culture
   systems [[99]34] and IC-BEVS [[100]35], such as the case of the
   influenza HA-VLPs herein produced. Therefore, distinct lipid metabolism
   could be associated with different productivity of adapted and
   non-adapted cells.

   Pathway enrichment analysis revealed that genes encoding ribosome
   subunits were up-regulated in adapted cells at 24 hpi. Since
   baculovirus is known to promote the shutdown of host cell protein
   synthesis upon infection for overproduction of viral proteins
   [[101]10,[102]36], this result suggests that the host cell translation
   machinery was less impaired in adapted cells, thereby leading to higher
   protein biosynthesis (including of HA-VLPs). The oxidative
   phosphorylation pathway was also found to be up-regulated in adapted
   cells at 24 hpi, allowing lower substrate consumption when compared to
   non-adapted cells. In contrast, pathways associated with drug,
   carbohydrate, and amino acid metabolism were down-regulated in adapted
   cells. Interestingly, the set of genes driving such enrichment is
   shared between most pathways; these code for proteins such as
   UPD-glucuronosyltransferases-like and glutathione S-transferase-like
   proteins, enzymes associated with cellular protection and resistance to
   oxidative stress [[103]37]. In addition to these, genes involved in
   oxidative metabolism were also down-regulated in adapted cells, some of
   which have been already identified as differentially expressed in
   insect cells resistant to harmine and fungi (e.g., cytochromes 450)
   [[104]38,[105]39]. Pathways associated with baculovirus infection such
   as immune response, protein processing in the endoplasmic reticulum,
   proteosome and ubiquitin mediated proteolysis, which are known to be
   up-regulated upon infection [[106]4,[107]40], were also found
   down-regulated in adapted cells. Taken together, these results suggest
   that adapted cells cope better with the stress induced by baculovirus
   infection when compared to non-adapted cells.

   The major AcMNPV fusion protein, GP64, plays an essential role in
   mediating virus–receptor binding, internalization, and membrane fusion
   during virus entry into both mammalian and insect cells [[108]41]. The
   fusogenicity of GP64 is low-pH-dependent, and fusion of the viral
   envelope with the endosomal membrane is triggered in the acidic
   endosomal lumen [[109]42]. Virus fusion was shown to be impaired at
   relatively high-pH conditions in mammalian cells [[110]43]; thus,
   baculovirus entry could be less efficient in adapted cells. In this
   regard, the baculovirus gene ACNVgp64 was found to be significantly
   down-regulated in adapted cells at 12 hpi (1.7-fold) and 24 hpi
   (2.0-fold). While the amount of differentially expressed host cell
   genes in adapted cells was equally distributed between those being up-
   or down-regulated, all the differentially expressed baculovirus-derived
   genes herein identified were shown to be down-regulated in adapted
   cells, hence showing that virus transcripts are produced earlier (or at
   higher quantity) in non-adapted cells. Whether this outcome is a
   consequence of less efficient virus entry in adapted cells still
   remains unknown, but such a fact could be behind the different cell
   growth kinetics observed for adapted cells, i.e., a slight increase in
   cell concentration and delayed onset of cell viability drop. The
   apparent lower burden caused to adapted cells at initial stages of
   infection may have been the key to achieving prolonged infection and
   consequently higher productivity.

5. Conclusions

   In this study, comparative transcriptome analysis revealed significant
   differences between adapted and non-adapted insect High Five cells
   during production of influenza HA-VLPs using IC-BEVS. Differential gene
   expression analysis showed baculovirus genes being down-regulated in
   adapted cells, revealing less susceptibility to infection. Several
   pathways were found enriched and differently regulated, such as those
   associated with protein synthesis, metabolism of xenobiotic and
   endogenous compounds, carbohydrates and amino acids. The gene
   expression signatures herein identified can be exploited for rational
   genetic engineering of insect cells and/or baculovirus to further
   improve production yields. Furthermore, single-cell RNA sequencing
   could help us to further understand the molecular signatures playing a
   role in systems’ productivity, as well as to conclude on adapted cell
   population heterogeneity.

Acknowledgments