Applications and Limitations of Equilibrium Density Gradient Analytical Ultracentrifugation for the Quantitative Characterization of Adeno-Associated Virus Vectors

Adeno-associated virus (AAV) vectors are produced as a mixture of the desired particle (full particle, FP), which is filled with the designed DNA, product-related impurities such as particle without DNA (empty particle, EP), and aggregates. Cesium chloride or iodixanol equilibrium density gradient ultracentrifugation (DGE-UC) has been used for the purification of AAV vectors. DGE-UC can separate FP from impurities based on the difference in their buoyant densities. Here, we report the applications and limitations of equilibrium density gradient analytical ultracentrifugation (DGE-AUC) using a modern AUC instrument that employs DGE-UC principles for the characterization and quantitation of AAV vectors. We evaluated the quantitative ability of DGE-AUC in comparison with sedimentation velocity AUC (SV-AUC) or band sedimentation AUC (BS-AUC) using AAVs with different DNA lengths and different serotypes. DGE-AUC enabled the accurate quantification of the ratio of FP to EP when the AAV vector primarily contains these particles. Furthermore, we developed a new workflow to identify the components of separated peaks in addition to FP and EP. Ultraviolet absorption spectra obtained by multiwavelength detection can also support peak assignment following component identification. DGE-AUC experiments for AAV vectors have limitations with regard to minor components with low absorption at the detected wavelength or those with a density similar to that of major components of AAV vectors. DGE-AUC is the only analytical method that can evaluate particle density heterogeneity; therefore, SV-AUC or BS-AUC and DGE-AUC are complementary methods for reliable assessment of the purity of AAV vectors.


Cell culture
Suspended HEK293T cells were used for AAV8 vector production.Suspended HEK293T cells were maintained with BalanCD HEK293 (FUJIFILM Irvine Scientific, Inc., Santa Ana, CA) with 1% penicillin-streptomycin.Cells were grown as adherent cultures in 5% CO 2 at 37C.

AAV8 vector preparation
AAV8 vector was generated using the triple plasmid, co-transfection.Briefly, pAAV-Rep&Cap (serotype 8), pAd helper, and transgene (CMV-EGFP) plasmids (Vector Builder, Vector ID: VB010000-9394npt) were co-transfected into suspended HEK293T cells cultured in a bioreactor at a ratio of 1:1:1.AAV8 vector from the transfected cells and the medium was harvested 72 h posttransfection and purified through affinity chromatography using AAVX column (Thermo Fisher Scientific, Waltham, MA) followed by one-cycle CsCl-DGE-UC to separate FP from EP. AAV8 vector purified through affinity chromatography (>2 mL) was dissolved in 2.5 M CsCl / PBS solution with 0.001% poloxamer-188.This final volume of 12 mL solution was loaded into a 13.2 mL ultraclear ultracentrifuge tube and ultracentrifuged at 34,000 rpm at 20ºC for 72 h using SW 41 Ti rotor and Optima XE-90 (Beckman Coulter).After centrifugation, the band corresponding to FP was collected using Piston Gradient Fractionator (BioComp, Fredericton, Canada) and dialyzed using

Two-cycle CsCl-DGE-UC purification
To separate and fractionate low and high buoyant density FPs, two-cycle CsCl-DGE-UC were conducted.The purified AAV8 vector was (>2 mL) dissolved in 2.5 M CsCl / PBS solution with 0.001% poloxamer-188.This final volume of 12 mL solution was loaded into a 13.2 mL ultra-clear ultracentrifuge tube and ultracentrifuged at 34,000 rpm at 20ºC for 72 h using SW 41 Ti rotor and Optima XE-90 (Beckman Coulter).The band corresponding to FP was collected and dissolved in 2.5 M CsCl / PBS solution with 0.001% poloxamer-188.Then, this solution was subjected to another round of ultracentrifugation at 24,000 rpm for 72 h.After centrifugation, the bands corresponding to low and high buoyant density FPs were collected using Piston Gradient Fractionator (BioComp) and dialyzed using Slide-A-Lyzer 10K (Thermo Fisher Scientific).500 mM EDTA, 55 L of 1 PBS with 0.001% poloxamer-188, and 5 L of Proteinase K (20  ×  mg/mL) were added to obtain a final volume of 100 L. Next, the mixture was incubated at 55ºC for  60 min, after which the mixture was heated at 95ºC for 20 min, followed by centrifugation to collect the lysate.Then, the ssDNA was purified according to the protocol mentioned in the QIAquick PCR Purification Kit (QIAGEN, Hilden, Germany) and used as the final collected sample.CGE was measured using a PA800Plus system (Sciex, Framingham, MA).The prepared samples were injected via electrokinetic injection.Detection was performed using a 488-nm laser excitation fluorescence with an emission filter of 520 nm.

CGE for VP components
AAV8 samples for CGE measurement were prepared mostly in accordance with a previously reported procedure. 1AAV8 solutions with a volume of 10 L (5.0 10 10 viral genomes) were denatured  × and buffer-exchanged following the protocol and the final collected sample was diluted with 50 L  of deionized water for injection.CGE measurement was performed using a PA800Plus system (Sciex).
Prepared samples were injected with water plug sample stacking.Detection was performed at 214 nm using a photo diode array detector.To determine the VP stoichiometry, peak areas of the CGE electropherogram detected at 214 nm were divided by the molar extinction coefficient of each VP at 214 nm, reflecting the UV absorbance of peptide bonds and amino acids under denatured conditions.rotor speed is 42,000 rpm (black) and CsCl conc. is 2.70 M and rotor speed is 60,000 rpm (red).Left axis showed the absorbance at 230 nm of the DGE-AUC equilibrium profile and right axis showed the calculated density of CsCl/PBS solution at 42,000 rpm (light green) and 60,000 rpm (dark green) using Equation 3. The gray and orange dashed lines showed the peak top positions and corresponding calculated densities in the DGE-AUC equilibrium profiles at 42,000 rpm and at 60,000 rpm, respectively.

Figures-
Figures -Figure S1.Time required for the CsCl density gradient to reach equilibrium in DGE-AUC.-Figure S2.Examination of the validity of the DGE-AUC equilibrium profile.-Figure S3.Geometry of sector-shape centerpiece (AUC double sector centerpiece with gasket and venting holes).-Figure S4.DGE-AUC experiment of AAV5ZsGreen1.-Figure S5.Illustration of two-cycle CsCl-DGE-UC.-Figure S6.Linear correlation between the s-value against encapsidated full-length DNA of the several AAV8 vectors obtained by BS-AUC.-Figure S7.Linear correlation of physicochemical properties of CsCl/PBS solution with 0.001% poloxamer-188.-Figure S8.The relationship between the simulated and experimentally determined density gradient.-Figure S9.DGE-AUC experiments of AAV8 vectors.-Figure S10.Optimization of the DGE-AUC conditions for characterizing AAV vectors.-Figure S11.Comparison the DGE-AUC equilibrium profile under different rotor speed conditions for characterizing AAV vectors.
10 L AAV8 solutions (1.0 10 12 viral genome) were treated with DNAase and ProteaseK to  × extract the ssDNA from the capsid using the following protocol.To prepare a final volume of 30 L  AAV8 samples, 20 L of nuclease-free water, 3 L of 10 DNase buffer, 1.5 L of Benzonase,   ×  and 5.5 L of 1 PBS with 0.001% poloxamer-188 were mixed.After this Dnase-treated samples  × were incubated at 37ºC for 30 min, the samples were transferred to a new tube, to which 10 L of  S-5