Development of an icIEF assay for monitoring AAV capsid proteins and application to gene therapy products

Abstract

Adeno-associated virus (AAV) gene therapy vectors, which contain a DNA transgene packaged into a protein capsid, have shown tremendous therapeutic potential in recent years. Methods traditionally used in quality control labs, such as high-performance liquid chromatography (HPLC) and capillary electrophoresis (CE), do not provide a complete understanding of capsid viral protein (VP) charge heterogeneity. In the present study, we developed simple, one-step sample preparation and charge-based VP separation using imaged capillary isoelectric focusing (icIEF) for monitoring AAV products. The robustness of the method was confirmed through a design of experiments (DoE) exercise. An orthogonal reverse-phase (RP) HPLC method coupled with mass spectrometry was developed to separate and identify charge species. Additionally, capsid point mutants demonstrate the capability of the method to resolve deamidation at a single site on the viral proteins. Finally, case studies using two different AAV serotype vectors establish the icIEF method as stability indicating and demonstrate that increases in acidic species measured by icIEF correlate with increased deamidation, which, we show, results in decreased transduction efficiency. The addition of a rapid and robust icIEF method to the AAV capsid analytical toolkit enables development and consistent manufacturing of well-characterized gene therapy products.

Keywords: AAV; LC-MS; deamidation; gene therapy research; icIEF charge heterogeneity determination.

Graphical abstract

Figure 1

An icIEF method is developed for charge variant determination of VPs of various AAV serotypes Shown in the electropherogram overlay are four serotype samples of AAV9, AAV-a, AAV6, and AAV3b. Charged species of each AAV sample are resolved well. Based on the theoretically estimated pI value and relative abundance of the three VPs, VP3 was tentatively assigned to the largest center peak of each serotype, and VP1 and VP2 were assigned to serotype AAV9 based on MS data presented later. The final method consists of the following major parameters: 4% broad-range ampholytes (pH 3–10), 0.2% methylcellulose, two internal pI markers flanking AAV VPs, 1–9 E12 vp/mL AAV sample injection concentration, 6 M urea and 5% β-mercaptoethanol (BME), 6-min incubation at 75°C, and 2-min pre-focusing at 500 V, followed by 5-min focusing at 3,000 V.

Figure 2

Overlay plots of the CCD, illustrating optimum separation conditions (A–C) This panel shows the overlay plots of responses of the three most abundant charge species (peak 1, peak 2, and peak 3) with temperature fixed at 70°C (A), 75°C (B), and 80°C (C) and time fixed at three levels (4, 6, and 8 min). Urea and BME were evaluated across the range of 4–8 M and 4%–7%, respectively.

Figure 3

VP1, VP2, and VP3 of AAV9 isolation by HPLC with the RP method and mass characterization by MS (A) Chromatogram of AAV9 VP separation. The signal was measured using a fluorescence detector with excitation and emission wavelengths of 280 and 350 nm, respectively. (B) Each peak was identified with LC-MS deconvoluted spectra. The letter “d” after the capsid protein (i.e., VP3d) denotes deamidation.

Figure 4

icIEF peak identification by LC-MS-characterized RP HPLC fractionation This stacked figure shows the icIEF profile of an unfractionated AAV9 control (top trace) and five fractions collected from RP HPLC. Fractions 1, 2, 3, 4, and 5 were characterized by LC-MS in Figure 3 as VP1, VP2, VP3d, VP3, and VP3d∗, respectively. d2 denotes the presence of 2 deamidated species on the capsid protein. Although a peak was listed as VP2d for fraction 2, it migrates at the same position as VP3.

Figure 5

icIEF detection of acidic shift through mutant study AAV mutants were generated with point mutations of Asp (N) to Asp (D) to mimic deamidation. (A) WT AAV-a. (B) For a single-base mutant, N263D, where the Asn at position 263 was changed to Asp, a clear profile shift to more acidic values across the entire electropherogram is observed because this residue is conserved across all three VPs. (C) Similarly, for a single-base mutant, N57D, where the Asn at position 57 was changed to Asp, no impact on the VP2 and VP3 peaks in the electropherogram is observed because only the VP1 residue N57 contains the mutated residue. (D) icIEF profile overlay of WT (black trace) and mutant N57D (red trace). The peaks associated with VP1 show a subtle acidic shift.

Figure 6

Comparable results of deamidation estimated by icIEF, RP LC, and MAM for AAV9 (A and B) Overlays of day 0 (black line), day 3 (blue line), and day 7 (green line) forced degradation AAV9 samples with (A) icIEF and (B) RP HPLC. (C) Comparison of VP3 deamidation changes measured with the icIEF (gray bar), RP HPLC (blue bar), and MAM (orange bar) for day 3 day and day 7 samples compared with the day 0 results.

Figure 7

The icIEF assay is stability indicating and aligns with transduction efficiency observations for AAV-a (A) Thermal stability of AAV-a is monitored by icIEF, with increasing relative abundance of VP3d and decreasing abundance of VP3. (B) After 4 weeks at room temperature, deamidated species increased to approximately half that of the parent species. Transduction efficiency (i.e., potency) data by activity (blue dots) or expression (orange dots) are shown to correlate well with the ratio of VP3d to VP3 as determined by icIEF, indicating how icIEF can be used to monitor deamidation-driven loss of potency. (C) Potency data were first correlated with deamidation obtained using MAM. Here, a linear response is demonstrated between deamidation quantified by use of MAM and icIEF. (D) MAM data were used to correlate forced degradation-induced deamidation between N57 (VP1 only) and N263 (VP1-3). The correlation between N57 and N263 deamidation suggests usage of VP3 deamidation to monitor capsid quality.