The AAV-DJ/8 Helper Free Promoterless Expression System (Cell Biolabs, San Diego, CA, United States) was applied to construct AAV vector. The pAAV-cfos-NLuc-ACC plasmid (unpublished) was constructed by Professor Zhou Lab (UAB, United States). The blue luminescence-emitting Nanoluciferase (NLuc) reporter gene cloned from pNL-CMV-NLuc (Promega #N1091) and synthesized ACC genes (unpublished), total of ∼3.9 kb, were inserted into the pAAV-MCS vector (Cell Biolabs). The plasmid pAAV-CMV-eYFP was constructed by cloning CMV promoter cloned from pcDNA3.1-PsChR2-eYFP (Addgene #69057) (Govorunova et al., 2013; Ernst et al., 2019) and eYFP reporter gene from pcDNA3.1-PsChR2-eYFP, totoal of 1.1 kb, into the pAAV-MCS. The PCR primers are NLuc-forward: 5′- ACT​CAC​TAT​AGG​GAG​ACC​CAC​ACC​ATG​GTC​TTC​ACA​CTC​G-3′, NLuc-reverse: 5′- TGC​CTG​ATC​CCG​CCA​GAA​TGC​GTT​CGC​AC-3′, eYFP forward: 5′-GGT​AGT​AGC​AAT​GCT​AGT​GAG​CAA​GGG​C-3′, and eYFP reverse: 5′-ACC​TTC​GAA​CCG​CGG​GCC​CTC​TAG​ATT​ACT​TGT​ACA​GCT​CGT​CC-3′. The pHelper plasmid and pAAV-DJ/8 Rep-Cap plasmid (Cell Biolabs) were used to produce AAV-DJ8 serotype.

The embryonic human kidney (HEK) 293AAV cell line (Cell Biolabs, #AAV-100), which was cloned and selected from parental 293 cell line (transformed with human adenovirus type 5 DNA), was used to produce AAV in adherent culture. The HEK 293F cell line (Gibco, Buffalo, NY, United States) that was adapted to serum-free medium was used to produce AAV in suspensive culture. The human triple-negative breast cancer (TNBC) cell line MDA-MB-468 (GenTarget, San Diego, CA, United States) was used to test the in vitro transfection capability of AAV and was also used to establish the tumor xenograft animal model to validate the in vivo infection.

All basal media, nutrient supplements and other reagents used in this study were purchased from Thermo Fisher Scientific (Waltham, MA, United States) or Gibco (Grand Island, NY, United States) unless otherwise specified. The HEK 293AAV cells were cultivated in DMEM (high glucose) supplemented with 10% fetal bovine serum (FBS, v/v), 0.1 mM MEM Non-Essential Amino Acids (NEAA), 2 mM L-glutamine, and 100 IU penicillin and 100 μg/ml streptomycin (i.e., 1% Pen-Strep) in T75 flasks. The HEK 293F cells were maintained in chemically defined (CD) FreeStyle 293 Expression Medium (Gibco) supplemented with 4 mM GlutaMAX in 125 or 250-ml shaker flasks on an orbital shaker at 135 rpm. The TNBC MDA-MB-468 cells were maintained in DMEM/F12 medium supplemented with 4 g/L glucose, 4 mM L-glutamine, 10% FBS (v/v) and 1% P/S in T-flasks. All the cell cultures were incubated in a humidified incubator (Caron, Marietta, OH) with set points of 37°C and 5% CO₂.

The adherent AAV production was performed in 150-mm petri dish. Specifically, the 30 ml of complete DMEM medium (same formulation as seed culture) was seeded with HEK 293AAV at viable cell density (VCD) of 0.1 × 10⁶ cells/ml. When the confluence reached 95%, spent medium was removed and replaced with fresh complete medium. The cells were transfected with pAAV-cfos-NLuc-ACC, pHelper, and pAAV-DJ/8 Rep-Cap plasmids with transfection reagent of 25 mM of Ca²⁺, lipofectamine 3,000 (Gibco), or liposomes synthesized in house. 1) Calcium transfection: The 0.33 ml of 2.5 M CaCl₂, 30 µg of each plasmid (total of 120 µg), and 0.1 µM of TE buffer were mixed to reach final volume of 1.5 ml. The Ca²⁺/DNA mixture was added to 1.5 ml of 2X HBS buffer slowly and dropwise to make 3 ml of Ca²⁺/DNA/HBS mixture, which was added to the 30-ml HEK 293AAV culture. The transfected cells were incubated for 4–6 h, followed with medium exchange and incubation for 48–60 h, and harvested for AAV lysis. 2) Lipofectamine transfection: 10 µg of each plasmid (total of 30 µg) and 60 µL of P3000 reagent were added into 300 µL of Opti-MEM medium to prepare the DNA-P3000 mixture. The 45 µL of Lipofectamine 3,000 was added into 300 µL of Opti-MEM medium to prepare Lipofectamine 3,000 mixture. The 600-µL complete DNA-Lipofectamine 3,000 mixture was incubated for 15 min at room temperature and used to transfect the HEK 293AAV cells. Medium exchange was performed at 6 h post transfection. The transfected HKE 293AAV cells were harvested and centrifuged at 48–60 h for further AAV clarification. 3) Cationic liposomes transfection: We synthesized cationic liposomes following similar synthetic procedure of neutral liposomes (Si et al., 2021) with modifications: mixing 7 µmole 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), 7 µmole 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) and 10 ml of chloroform and evaporating at 60°C and 50 rpm for 1 h. The generated liposomes were analyzed using NanoSight, which showed concentration of 3.2 × 10¹¹ particles/ml and mean size of 92.6 ± 0.4 nm. Total of 30 µg three plasmids (10 µg each plasmid) were mixed with 900 µL of cationic liposomes, incubated on ice for 30 min, and mixed with Opti-MEM medium with volume ratio of 1:1. The HEK 293AAV cells in petri-dish were transfected with the liposomes-DNA complex and incubated for 48–60 h before AAV harvest.

The suspensive AAV production was performed in shaker flasks or spinner flasks. The Freestyle 293 medium was inoculated with HEK 293F cells at seeding VCD of 0.1–0.3 × 10⁶ cells/ml and incubated for 20–24 h before transfection. Different plasmid:cell ratios of 1.6, 3.3, and 6.6, and three transfection reagents of Polyethylenimine HCI (PEI) MAX, TurboFect and our synthesized liposomes were evaluated and compared in suspensive AAV production. The following formulation was designed for 30-ml HEK 293F culture. 1) PEI transfection: Total of 30 µg plasmids (10 µg each plasmid) were mixed with 120 µL of 1.6 mg/ml PEI MAX and incubated at 37°C for 10 min, which was used to transfect HEK 293F cells. 2) TurboFect transfection: The same formulation of TurboFect-DNA transfection complex as PEI was prepared for suspension transfection. 3) Cationic liposomes transfection: The same transfection reagent formulation as in adherent AAV production was used to transfect HEK 293F cells.

In the end of AAV production, cell pellets were collected with centrifugation of the culture broth at 1,000 g and 4°C for 10 min. The cell pellets were re-suspended using lysis buffer (50 mM TrisCl, 150 mM NaCl, 2 mM MgCl₂, pH 8.0) with volume of 1/30 of production culture volume. Then cell lysis was processed through three cycles of freezing/thawing (30-min freezing in ethanol/dry ice and 15-min in 37°C water bath thawing each cycle). The 50 U/ml benzonase and 5 μg/ml RNase A were added into AAV lysate and incubated in shaker at 37°C and 150 rpm for 60 min. Then 0.5% sodium deoxycholate was added to further treat the lysis solution for 30 min. The AAV supernatant was collected after centrifugation at 3,000×g and 4°C for 20 min, and filtered using 0.45 and 0.22 μm PES membrane to remove cell debris.

To extract ssDNA from AAV, 1 µL of 2000 U/ml DNase I (BioLabs), 2 µL of 10X DNase I reaction buffer, 12 µL of nuclease free water, and 5 µL of AAV sample were added into a 0.2-ml PCR tube. The mixture was incubated at 37°C for 30 min and followed with DNase I inactivation at 70°C for 10 min. The sample was further digested with 1 µL of 2 mg/ml Proteinase K by incubation at 50°C for 1 h and deactivation at 95°C for 20 min. Finally, the extracted ssDNA was diluted by 10 folds with nuclease free water and titrated using RT-PCR with primers of NLuc forward: 5′-ATT​GTC​CTG​AGC​GGT​GAA​A-3′, NLuc reverse: 5′-CAC​AGG​GTA​CAC​CAC​CTT​AAA-3′, eYFP forward: 5′-GCA​CAA​GCT​GGA​GTA​CAA​CTA-3′, and eYFP reverse: 5′-TGT​TGT​GGC​GGA​TCT​TGA​A-3′. The raw AAV or purified AAV was buffer exchanged in 1x PBS, 5% Sorbitol, and 350 mmol/L NaCl, and stored at −80°C for long term.

The animal study conforms to the National Institutes of Health Guide for the Care and Use of Laboratory Animals published (Publication No. 85‐23), with the approved animal protocol of IACUC-21949 by the Institutional Biosafety Committee at University of Alabama at Birmingham.

In in vitro evaluation, the MDA-MB-468 cancer cells were seeded in 2 ml of complete DMEM/F12 medium in 6-well plate with seeding VCD of 0.05 × 10⁶ cells/ml. After 24-h incubation, the cells were infected with the raw AAV carrying NLuc at multiplicity of infection (MOI) of 10,000 in triplication with mock infection with PBS as control. Two days after infection, the fresh growth medium containing 30 µM of substate ViviRen was used to replace spent medium. The In Vivo Imaging System (IVIS) Lumina Series III (PerkinElmer, Waltham, MA, United Staes) was applied to take the bioluminescence images of infected cells at emission wavelength of 460 nm.

To evaluate the in vivo infection capability of produced AAV, 5 × 10⁶ human TNBC MDA-MB-468 cells were subcutaneously (s.c.) injected into the 6-week-old NSG (NOD scid gamma) female mice (Jackson Laboratory, Bar Harbor, ME) to establish xenograft model. Total of 1 × 10¹⁰ gc AAV was intravenously (i.v.) injected and mice were maintained for 7 days to allow NLuc expression, followed by i.p. injection of 37 µg ViviRen substrate. The live-animal image was captured using IVIS system at excitation/emission wavelength of 660/710 nm. The bioluminescence intensity was recorded with exposure time of 5 s to monitor peak excitation.

The replication numbers (n) used in this study was >4 and all experimental data were presented as mean ± standard error of the mean (SEM). Two-tailed Student’s t tests were used to determine the probability of significance between groups and comparison was performed using a one-way ANOVA followed by post-hoc (Dunnett’s) analysis. The statistical significance of **p value <0.05 was considered for all tests.

The overview of the developed AAV biomanufacturing process is detailed in Figure 1, including both adherent production, suspensive production, purification, and storage. In adherent operation mode, the HEK 293AAV cells were used to seed multiple petri-dishes for AAV production. The calcium solution that showed the highest transfection efficiency and AAV productivity was applied to deliver the three helper-free expression system plasmids, including the pAAV plasmid expressing NLuc reporter gene (Figure 2A) or eYFP reporter gene (Figure 2B), to HEK 293AAV cells. The key parameters that we identified in calcium transfection were the pH of transfection complex, medium exchange before transfection, and medium exchange post transfection. In suspensive operation mode, the HEK 293F cells were seeded in CD medium in shaker flasks or spinner flasks for AAV production. The synthesized cationic liposomes identified as the best transfection reagent was used to deliver plasmids to 293F cells. The key parameters that we identified in suspensive production were the transfection cell density and ratio among liposomes, plasmid and cell number.

The cell pellets harvested in the end of AAV production were processed to release AAV through freezing/thawing in lysis buffer. The supplement of sodium deoxycholate, benzonase, and RNase A could further treat AAV lysis and remove DNA/RNA impurities. The raw AAV supernatant post centrifugation was filtered using 0.45 and 0.22 μm PES filters to remove cell debris. The well-developed iodixanol gradient ultracentrifugation could be used to purify AAV product, which could achieve high purity and recovery rate despite the low purification capacity and long processing procedure (Zolotukhin et al., 1999; Strobel et al., 2015). The ion-exchange chromatography column (Brument et al., 2002; Lock et al., 2012; Potter et al., 2014; Qu et al., 2015) or affinity chromatography column (Wu and Ataai, 2000; Grieger et al., 2006; Arakawa et al., 2007; Potter et al., 2014) can be used to purify some serotypes of AAVs. The buffer exchange and AAV concentration was performed using 3 or 10 kDa MWCO concentrator. The purified AAV was stored at −80°C freezer for long-term storage.

In this study, we focused on the process development and optimization of AAV-DJ/8 production and validation of the infection bioactivity. The purification and storage were also briefly tested and discussed.

During the AAV bioprocessing optimization, we first evaluated the effects of production mode (i.e., adherent and suspensive), transfection reagents and host cells on AAV packing and production. As described in Figure 3A, the calcium-facilitated transfection of pAAV-cfos-NLuc-ACC, pHelper, and pAAV-DJ/8 Rep-Cap plasmids generated the highest volumetric productivity of 3.87 × 10⁹ gc/ml using HEK 293AAV cells in petri dishes. The transfections with lipofectamine 3,000 and in-house synthesized cationic liposomes resulted in AAV productivity of 8.37 × 10⁸ gc/ml and 5.29 × 10⁸ gc/ml, respectively. The titer of raw AAV lysis were 5.45 × 10¹⁰ gc/ml, 1.98 × 10¹⁰ gc/ml, and 1.09 × 10¹⁰ gc/ml for these three transfection reagents. The comparison among three host cells showed that the HEK 293AAV cells had higher AAV productivity of 7.86 × 10⁹ gc/ml and titer of 4.89 × 10¹¹ gc/ml than HEK 293A with productivity of 6.25 × 10⁹ gc/ml and titer of 1.67 × 10¹¹ gc/ml and HEK 293F cells with productivity of 3.01 × 10⁹ gc/ml and titer of 1.34 × 10¹¹ gc/ml using calcium transfection in petri dish culture (Figure 3B). Three transfection reagents of PEI MAX, TurboFect and liposomes were investigated in suspensive AAV production. As shown in Figure 3C, the liposomes generated the highest AAV productivity of 5.78 × 10⁹ gc/ml and titer of 9.84 × 10¹⁰ gc/ml by HEK 293F in shaker flasks. The PEI produced AAV with productivity of 5.38 × 10⁸ gc/ml and titer of 5.38 × 10⁹ gc/ml, and TurboFect generated productivity of 1.70 × 10⁸ gc/ml and titer of 2.83 × 10⁷ gc/ml.

This study showed that calcium had higher transfection efficiency than other reagents in adherent AAV production, but it’s hard to use calcium to perform suspension transfection and medium exchange pre-and post-transfection. The cationic liposomes demonstrated high transfection efficiency in suspensive AAV production and the low cost of liposomes could significantly reduce the AAV production cost as compared to PEI and TurboFect. Moreover, the AAV productivity using HEK 293F and liposomes was only ∼25% lower than that using HEK 293AAV and calcium. Therefore, the suspensive AAV production with HEK 293F and cationic liposomes was performed to optimize AAV production.

To further optimize the suspensive AAV production, we tested the effects of transfection cell density, ratio between plasmid DNA and transfected cells, and size of insert gene in the AAV-DJ/8 Helper Free Promoterless Expression System. In this study, the suspensive HEK 293F was transfected with synthesized liposomes. First, the transfection VCD of 0.4 × 10⁶ cells/ml of HEK 293F with pAAV carrying 3.9-kb insert gene produced the highest productivity of 5.55 × 10⁹ gc/ml and titer of 3.33 × 10¹⁰ gc/ml, which was significantly higher than that of VCD of 0.2, 0.5, and 0.6 × 10⁶ cells/ml with productivity of 0.29–2.20 × 10⁹ gc/ml and titer of 0.18–1.32 × 10¹⁰ gc/ml (Figure 4A). Second, different ratios of plasmid DNA and transfected cells were examined, including 1.6, 3.3, and 6.6 µg:10⁶ cells. The AAV production showed that ratio of 1.6 generated higher productivity (2.20 × 10⁹ gc/ml) and titer (1.32 × 10¹⁰ gc/ml) (Figure 4B). Third, the AAV production using pAAV carrying 3.9 kb of insert gene was higher than that using pAAV carrying 1.1 kb of insert, i.e., productivity of 7.64 vs. 4.56 × 10⁹ gc/ml and titer of 9.84 vs. 1.37 × 10¹⁰ gc/ml (Figure 4C). The maximum insert of AAV-DJ/8 expression system is 3.9 kb, including promoter and gene of interest, so we tested a large insert cfos-NLuc-ACC with size of 3.9 kb and small insert CMV-eYFP with size of 1.1 kb in this study.

HEK 293F cells have been widely used to transiently produce antibody or other recombinant protein. Literature has reported that the plasmid transfection and protein expression level are affected by the transfection VCD and ratio of plasmid and cells (Grieger et al., 2016). This study showed that the transfection cell density and plasmid amount can be optimized to improve the suspensive AAV production. Interestingly, we found that the size of insert gene had no obvious correlation to AAV production in suspensive culture. The identified optimal transfection conditions, i.e. transfection VCD of 0.4 × 10⁶ cells/ml, plasmid and HEK 293F cell ratio of 1.6, and synthesized cationic liposomes, were used in suspensive AAV production process.

The scalability of our developed suspensive AAV production process was evaluated in both shaker flask and spinner flask, which showed consistent AAV productivity. Specifically, the identified process parameters in 30-ml shaker flask suspension production (Table 1), i.e. transfection VCD of 0.4 × 10⁶ cells/ml of HEK 293F cells, liposomes as transfection reagent of 6.9-kb pAAV-Luc, and DNA:cells ratio of 1.6 µg:106 cells, were applied in the scalability evaluation in 240-ml of production culture in shaker flask at 37°C and agitation 100 rpm and 450-ml production culture in spinner flask at 37°C and agitation 70 rpm. As presented in Figure 5, the 240-ml production culture in 1-L shaker flask and 450-ml culture in 1-L spinner flask produced similar level of AAV with volumetric productivity of 5.04 × 10⁹ gc/ml and 4.99 × 10⁹ gc/ml and titer of 3.02 × 10¹⁰ gc/ml and 2.84 × 10¹⁰ gc/ml, respectively. During the liposomes-mediated plasmid transfection in shaker and spinner flasks, the cultures were kept in static status for half hour after adding transfection complex. Medium exchange is not necessary pre- and post-transfections. The successful scale-up of AAV production in spinner flask indicated the possibility to produce large-scale AAV in stirred-tank bioreactor.

In this study, the raw or purified AAV was concentrated and exchanged to formulation buffer (1x PBS, 5% sorbitol, 350 mmol/L NaCl). It is found that AAV is stable with this storage condition and the functional titer change is <5% between pre-freezing and post-thawing.

In addition to genome copy-based titration, it’s very important to evaluate the bilogical activity of AAV carrying the NLuc reporter gene. In this study, we transfected the TNBC MDA-MB-468 cells using AAV in 6-well plate. As shown in Figure 6A, the high-level expression of NLuc gene was detected in the cancer cells with IVIS imaging, but no bioluminecent signal in the cells with mock infection (i.e., no AAV). This result confirmed the in vitro transfection and bioactivity of the produced AAV. The kinetic profile of in vivo NLuc bioluminecence, which was excited at wavelength of 470 nm, demosntrated the peak signal at ∼15 min post substrate induction followed with signal reduction by 67% wuthin 60 min. We further tested the in vivo infection efficiency by administrating AAV into the TNBC MDA-MB-468 xengraft models via tail vein. As described in Figure 6B, we detected the functional expression of NLuc gene in TNBC xenograft, which demonstrated the effective in vivo infection and gene expression that was delivered by AAV. Moreover, the application of cancer-specific promoter cfos resulted in gene expression in tumor xenograft. To overcome the challenge of possible low resolution of live-animal IVIS imaging, we will further validate the AAV biodistribution through RT-PCR analysis of the important organs to detect the mRNA level of delivered gene in future.

This study compared and optimized both adherent production and suspensive production of AAV as summarized in Table 1. The results showed that these two operation modes could generate AAV with similar productivity and titer, which is doable for small-scale production. However, the large-amount AAV needs scalable suspensive production in stirred-tank container or bioreactor with optimal transfection conditions. For example, the agitation speed could affect the transfection efficiency significantly and might need optimization after adding transfection complex. The in-house synthesized cationic liposomes have size distribution of 92.6 ± 0.4 nm and titer of 3.2 × 10¹¹ particles/ml. The particle size of the cationic liposomes could be further optimized in order to improve the suspensive transfection efficiency. Although the AAV-DJ/8 with an engineered capsid to achieve high in vivo infection efficiency was used as model system, the developed bioprocess could be used to generate different serotypes.

This study aimed to optimize the AAV production bioprocessing, so we did not focus on purification strategies. However, we performed a quick test and evaluation of two chromatography purification methods following published protocols to purify AAV-DJ/8. The ion-exchange purification using liquid chromatography system (Bio-Rad, Hercules, CA, United States) equipped with Cytiva HiTrap™ Q Sepharose XL IEX column (Cytiva, Marlborough, MA, United States) (Brument et al., 2002; Lock et al., 2012; Potter et al., 2014; Qu et al., 2015) was performed, but the recovery rate was lower than 10% and the purity was low. The affinity purification with Cytiva HiTrap™ AVB Sepharose column (Wu and Ataai, 2000; Grieger et al., 2006; Arakawa et al., 2007; Potter et al., 2014) was also tested, and the recovery rate was less than 1% or close to 0%. The failure to use AVB column to isolate AAV-DJ/8 was also reported by Andari and Grimm (El Andari and Grimm, 2021) and Nass et al. (Nass et al., 2018) due to the serotypes evolution. The traditional iodixanol gradient ultracentrifugation with Beckman Coulter L-100K ultracentrifuge (Beckman Coulter, Brea, CA) (Zolotukhin et al., 1999; Strobel et al., 2015) achieved the highest recovery rate and purity although the operation took a long time and was hard to scale up. We will further develop and optimize the chromatography-based AAV purification protocol in future study.