Riboswitch-mediated Attenuation of Transgene Cytotoxicity Increases Adeno-associated Virus Vector Yields in HEK-293 Cells

Abstract

Cytotoxicity of transgenes carried by adeno-associated virus (AAV) vectors might be desired, for instance, in oncolytic virotherapy or occur unexpectedly in exploratory research when studying sparsely characterized genes. To date, most AAV-based studies use constitutively active promoters (e.g., the CMV promoter) to drive transgene expression, which often hampers efficient AAV production due to cytotoxic, antiproliferative, or unknown transgene effects interfering with producer cell performance. Therefore, we explored artificial riboswitches as novel tools to control transgene expression during AAV production in mammalian cells. Our results demonstrate that the guanine-responsive GuaM8HDV aptazyme efficiently attenuates transgene expression and associated detrimental effects, thereby boosting AAV vector yields up to 23-fold after a single addition of guanine. Importantly, riboswitch-harboring vectors preserved their ability to express functional transgene at high levels in the absence of ligand, as demonstrated in a mouse model of AAV-TGFβ1-induced pulmonary fibrosis. Thus, our study provides the first application-ready biotechnological system-based on aptazymes, which should enable high viral vector yields largely independent of the transgene used. Moreover, the RNA-intrinsic, small-molecule regulatable mode of action of riboswitches provides key advantages over conventional transcription factor-based regulatory systems. Therefore, such riboswitch vectors might be ultimately applied to temporally control therapeutic transgene expression in vivo.

Figure 1

3′-GuaM8HDV enables efficient suppression of GFP expression in HEK-293 AAV producer cells. (a) Scheme of AAV-GuaM8HDV construct design and riboswitch-mediated transgene suppression. The riboswitch was inserted either upstream (5′) or downstream (3′) of the transgene, whose expression is driven by a CMV promoter. The transgene cassette is flanked by AAV2 ITR sequences, which define the DNA section that is packaged into the AAV particle. While transgene expression in absence of the ligand (guanine) might lead to toxic effects that can decrease AAV vector yield (upper panel), guanine addition triggers self-cleavage of the riboswitch, which attenuates transgene expression, thereby increasing AAV vector yield (lower panel). (b) Normalized GFP expression measured by flow cytometry and (c) fluorescence microscopic analysis of HEK-293 cells 24 h after transfection with either 5′-, 3′- or 5′3′-GuaM8HDV-harboring pAAV-GFP or a riboswitch-free pAAV-GFP control construct and addition of increasing concentrations of guanine. n = 4 biological replicates, mean ± SD. Bar = 400 µm. (d) Normalized GFP expression measured by flow cytometry and (e) fluorescence microscopic analysis of HEK-293 cells 24, 48, and 72 hours after transfection with the 3′-GuaM8HDV-harboring pAAV-GFP construct and addition of increasing concentrations of guanine. Ctrl = 0 µmol/l guanine. n = 3 biological replicates, mean ± SD. *P < 0.05; **P < 0.01. Bar = 400 µm.

Figure 2

3′-GuaM8HDV attenuates transgene-mediated impairment of producer cell integrity. (a) Western blot analysis of BAX (23 kDa) expression in HEK-293 cell lysates 16 hours after transfection with either the 3′-GuaM8HDV-harboring pAAV-BAX construct or a riboswitch-free pAAV-BAX control construct and addition of indicated amounts of guanine. anti-Vinculin (116 kDa) staining was used as a loading control. (b) ELISA measurement of TNFα protein levels in the supernatant of HEK-293 cells, 24 hours after transfection with either the 3′-GuaM8HDV-harboring pAAV-TNFα construct or a riboswitch-free pAAV-TNFα control construct and addition of indicated amounts of guanine. n = 6 biological replicates, mean ± SD. (c,d) Cytotoxicity analysis by lactate dehydrogenase (LDH) measurement in the supernatant of HEK-293 cells 24 hours after transfection with either (c) the 3′-GuaM8HDV-harboring pAAV-BAX construct or a riboswitch-free pAAV-BAX control construct or (d) the 3′-GuaM8HDV-harboring pAAV-TNFα construct or a riboswitch-free pAAV-TNFα control construct and addition of indicated amounts of guanine. LDH levels measured in triton-lysed cells served as a positive control and were set 100%. n = 5 biological replicates, mean ± SD. (e) ELISA measurement of TGFβ1 protein levels in the supernatant and (f) qPCR-based measurement of PAI-1 gene expression in total RNA samples of HEK-293 cells, 24 hours after transfection with either the 3′-GuaM8HDV-harboring pAAV-TGFβ1 construct or a riboswitch-free pAAV-TGFβ1 control construct and addition of indicated amounts of guanine. n = 4 biological replicates, mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001 relative to the ctrl 0 µmol/l sample or as indicated. ^###P < 0.001 relative to untreated cells. (g) Western blot analysis of LOXL2 (87 kDa) expression in the supernatant of HEK-293 cells 48 hours after transfection with either the 3′-GuaM8HDV-harboring pAAV-LOXL2 construct or a riboswitch-free pAAV-LOXL2 control construct and addition of indicated amounts of guanine. Ponceau S staining was used as a loading control (see Supplementary Figure S3).

Figure 3

3′-GuaM8HDV-mediated suppression of toxic transgene expression increases AAV yields. Benzonase-resistant AAV vector genomes (VG) were quantified by qPCR in HEK-293 cell lysate, 72 hours after transfection with either the 3′-GuaM8HDV pAAV-BAX, -TNFα, -TGFβ1, or -LOXL2 construct or respective riboswitch-free control constructs (and further plasmids necessary for the production of AAV6.2, AAV8, or AAV9—see Materials and Methods section for details) in presence of indicated concentrations of guanine, which were added during the medium exchange step about 5 hours after transfection. n = 3 biological replicates, mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4

The 3′-GuaM8HDV riboswitch system enables high-titer vector production. AAV6.2 vectors carrying various transgene constructs were produced as described in Figure 3 in the presence or absence of 150 µmol/l guanine (as indicated). Seventy-two hours after transfection, benzonase-resistant AAV vector genomes (VG) were quantified by qPCR in HEK-293 cell lysate. The percentage of vector titers relative to the riboswitch-free GFP control construct is depicted on each bar. The AAV-GFP yield range is indicated by the striped background. 3′-Gua = 3′GuaM8HDV. n = 6 biological replicates, mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, relative to GFP without guanine.

Figure 5

3′-GuaM8HDV-harboring AAV vectors are functional in vivo. Mice received a single application of either 2.7 × 10¹¹ vg of a riboswitch-free AAV6.2-CMV-TGFβ1 control vector, the same dose or a 1.5-fold higher dose (4.0 × 10¹¹ vg) of the 3′-GuaM8HDV-containing AAV6.2-CMV-TGFβ1 vector or PBS via intratracheal application. Analyses were conducted 21 days after application. (a) ELISA measurement of TGFβ1 protein levels in bronchoalveolar lavage (BAL) samples. (b) Total immune cell counts measured in BAL samples. (c) Wet lung weight. (d) Masson-trichrome staining of formalin-fixed paraffin-embedded lung tissue sections and (e) corresponding Ashcroft score. (f) Lung function (lung compliance) analysis. n = 5 animals per group, mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, relative to PBS-treated animals or as indicated. Bar = 200 µm.