Determination of AAV Capsid Protein Ratios by Sodium Dodecyl Sulfate Capillary Electrophoresis (CE‐SDS)—UV Absorption Method

ABSTRACT

Recombinant adeno‐associated viruses (AAVs) have become a leading platform for in vivo gene therapy. The AAV capsid is assembled from 60 individual protein subunits derived from three overlapping viral proteins (VPs): VP1, VP2, and VP3. The VP's ratio distribution affects the capsid trafficking and transgene expression. Quantitative analysis of the ratio is important in monitoring the potency of the AAVs. Sodium dodecyl sulfate capillary electrophoresis (CE‐SDS) offers automated and quantitative determination of the VP ratio distribution, and the method has been widely used in the application. The CE‐SDS method described was designed to ensure the consistent determination of AAV VP ratios across various sample concentrations. UV‐absorption detection was chosen for this purpose as it negated the need for any labeling in the process. To compensate for the lower sensitivity associated with UV‐absorption detection, methanol precipitation was employed during sample preparation, and a stacking injection technique was optimized for the samples in the study under CE conditions. These two techniques were found to be highly compatible. The method demonstrated robust linearity for AAV serotype 9 concentrations and exhibited good multiday precision. The VP ratios were successfully determined in AAV9 samples across a concentration range of 6.15 × 10¹² to 1.85 × 10¹³ GC/mL. The relative standard deviation (RSD%) for VP ratio determination within this concentration range was maintained below 7% for VP3, VP2, and VP1. Additionally, this method was further applied to other AAV serotypes, including AAV2 and AAV8, illustrating its versatility and broad applicability.

1. Introduction

Recombinant adeno‐associated viruses (AAVs) have become a leading platform for in vivo gene therapy [1]. The clinical translation of AAV‐based therapies has accelerated in recent years, marked by regulatory approvals of several therapies, with numerous candidates advancing through clinical trials for various genetic and acquired diseases [2]. As for all therapeutics, to ensure the therapeutic efficacy and safety of AAV vectors, rigorous analytical characterization is required. Requirements must adhere to the guidance of various regulatory agencies throughout product process development and for final product release [3].

The AAV particle itself is a non‐enveloped virion, approximately 25 nm in diameter, with a T = 1 icosahedral capsid structure [4, 5]. The capsid packages a single‐stranded DNA genome (up to ∼4.7 kb) [6, 7]. The capsid is assembled from 60 individual protein subunits derived from three overlapping viral proteins (VPs): VP1 (81.2 kDa), VP2 (66.2 kDa), and VP3 (59.7 kDa) [8, 9]. The typical stoichiometric ratio of the three proteins is approximately VP1:VP2:VP3 = 1:1:10. The VP's ratio distribution affects the capsid trafficking and transgene expression [10, 11]. Quantitative analysis of the ratio is important in monitoring the potency of AAVs.

Sodium dodecyl sulfate capillary electrophoresis (CE‐SDS) offers automated and quantitative determination of the VP ratio distribution [12, 13, 14, 15, 16, 17]. Two detection modes have been used for CE‐SDS determination of AAV VP ratio: UV absorption detection (UV absorption) and laser‐induced fluorescence (LIF) detection [15, 17]. UV absorption provides reliable quantitative data, as it is a direct method to quantify proteins without bias. However, its sensitivity may not be sufficient for some applications. In contrast, LIF detection has higher sensitivity but requires covalent dye labeling of proteins sample before separation [15]. The labeling may generate method‐related impurities and may be bias against certain components when different dyes are used [16]. Reviewing the literature, we found limited study of VP ratio determination within a certain sample concentration range by CE‐SDS. For a method used to determine the VP ratio of AAVs, the measured values in a given AAV concentration range need to be consistent. Capsid protein concentration linearity has been shown for total VP or individual VPs; however, to our knowledge, none demonstrate consistency of the measured VP ratios in this linear range.

In this study, the determination of the VP ratio via CE‐SDS was investigated throughout the sample concentration linear range. Both sensitivity and reproducibility were prioritized during the development of our analytical method. To ensure reproducible VP ratios across a defined range of sample concentrations, an optimal total AAV VP sample loading concentration was first established for VP ratio assessments. The signal linear range for total VPs of the method should be within ±50% of this predetermined loading concentration. Furthermore, the VP ratios calculated within this range were consistent.

To meet these criteria, UV absorption detection was implemented to negate the necessity for sample labeling, thereby eliminating variations in labeling across different sample concentrations. Additionally, the method's sensitivity was augmented by incorporating an organic solvent precipitation technique, which is widely utilized in proteomic research [18]. This precipitation strategy facilitated the application of sample stacking method in electrokinetic (EK) injection within our CE‐SDS protocol, further enhancing the method's sensitivity [19].

2. Materials and Methods

2.1. Materials and Instruments

AAV9 (5.8 × 10¹⁴ GC/mL), AAV2, and AAV8 products were obtained from their manufacturers. Methanol (HPLC plus grade) and 2‐mercaptoethanol (β‐ME, M3148) were purchased from Sigma (St. Louis, MO, USA). Reagents for CE‐SDS runs were from the Reagent Kit of Maurice CE‐SDS plus (PS‐MRK01‐S, ProteinSimple, San Jose, CA, USA). The reagents in the Reagent Kit included column condition solutions, running buffer for filling capillary column, and wash solution.

Maurice CE instrument and its CE‐SDS Plus cartridge (catalog # 090‐000 and PS‐MC02‐SP, ProteinSimple, San Jose, CA, USA) were used in the method.

2.2. Sample Preparation

A volume of 20 µL AAV sample solution was transferred to a 500‐µL centrifuge tube, whereas 180 µL ice‐cold methanol was added to the tube. The mixture was vortexed for 15 s. The mix was stored overnight for about 12 h at −80°C. The sample in the tube was centrifuged for 30 min at 13 200 rpm at −10°C. After the centrifuging, the supernatant was carefully removed from the tube. The precipitated pellet in the bottom of the tube was let air dry for 10 min. The pellet was dissolved with 10 µL 1× CE‐SDS Plus Buffer. The tube was vortexed for 10 s, then it was centrifuged for 5 s. The above two steps were repeated for three times. β‐ME (at 5% V/V in final 10 µL solution) and CE‐SDS internal standard (IS, 4% final) were added to the solution. The sample mixture was heated at 70°C for 10 min and was set cool to room temperature. The sample tube was centrifuged firefly for 10 s. Finally, 40 µL water was added into the tube, and it was mixed well.

2.3. CE‐SDS Method

In CE‐SDS methods run on Maurice CE instrument, all steps were predetermined by the software with minimum requirement for operator interference. Before sample injection, the capillary column was conditioned by Condition Solutions 1 and 2 with deionized water between the Condition Solution washes. The column was filled with running buffer by pressure. The samples were loaded onto the CE‐SDS Plus cartridge using EK injection for 20 s at 4600 kV. The separation of samples was conducted at 5750 V over 35 min, with detection accomplished through 220 nm absorbance, using Compass for ICE software for data acquisition.

3. Results and Discussion

3.1. Method Optimization Description

Three organic solvents, acetone, ethanol and methanol, were tested for precipitating VPs. For the AAV9 sample, methanol showed slightly higher recovery rate (data are not shown).

To keep a high portion of methanol in sample precipitation step, the ratio of 1 to 9 (sample volume to methanol volume) was used in the precipitation step. However, different sample volumes were tested: 5, 10, 20, 30, and 40 µL. These sample volumes gave the same recovery rate for different AAV9 sample concentrations. To balance the sample volume and precision, 20 µL sample:180 µL methanol ratio was used in the AAV precipitation.

3.2. Method Sensitivity

To circumvent the complications associated with sample labeling, such as impurities and introduced bias, UV absorption detection at 220 nm was opted for determining the VP ratio in this study. To address the lower sensitivity associated with UV detection, strategies like methanol precipitation of the AAV particles before denaturing the sample, followed by a streamlined sample stacking technique during injection, were applied.

The ratio of methanol volume to sample volume was fixed at 9:1. The precipitated solid pellet was then dissolved in 10 µL of sample buffer. This maintains that the final sample volume remains consistently at 10 µL in undiluted sample buffer, allowing the SDS and other compounds to stay at optimal concentrations for effective protein denaturation. The 10 µL of solution was subsequently heated for denaturing the AAV particles. This volume also matched well with the stacking injection method utilized under our CE conditions.

For the sample stacking in EK injection, after denaturation, to achieve the good stacking effect, the sample buffer in the sample needed a five‐fold dilution with a low conductivity solvent, such as water. After the precipitation step, the precipitated VP pellet was dissolved in 10 µL sample buffer (total 10 µL volume). This precipitation‐re‐dissolving step ensured that there was adequate room for the dilution by 40 µL water, resulting in the required final sample volume of 50 µL by the method for the injection into the instrument. The stacking effect during the EK injection was achieved by this diluted sample buffer in the sample. The method aligns closely with the process reported in reference [19]. In our application, although the separation capillary column was filled with undiluted sample buffer, the sample solution contained the same sample buffer but was diluted five times with water. The conductivity of the sample solution was much lower than that of the conductivity of the undiluted sample buffer that filled in the separation capillary. During the EK injection, a greater voltage drop happened in the low conductivity section (diluted sample solution in sample around the capillary column inlet) compared to that throughout the separation column, owing to the diluted solution's lower conductivity. This led to the accumulation of sample molecules at the capillary column inlet.

Figure 1 displays a comparison between two sample injections: one with stacking, A and the other without, B. Each sample was positioned at 50 µL volume and 2.0 × 10¹² GC/mL concentration under identical electrophoresis conditions. The primary difference was in sample buffer concentration; Sample A featured a five‐fold diluted sample buffer concluding at a 50 µL volume (sample stacking), whereas Sample B maintained an undiluted sample buffer also reaching a volume of 50 µL (for normal sample injection method). With the implementation of our stacking method, the peak area increased six‐fold (Figure 1) with stacking sample injection.

FIGURE 1.

(A) 2.0 × 10¹² GC/mL AAV9 concentration in 50 µL volume with 0.2× sample buffer concentration; (B) the same sample concentration as that in (A), 50 µL sample volume with 1× sample buffer concentration. All other electrophoresis conditions are the same for (A) and (B), as described in CE‐SDS condition.

Combined with methanol precipitation and stacking in sample injection, the limit of quantitation (LOQ) of the method, defined as 10× peak height/base line noise ratio, was 1.2 × 10¹¹ GC/mL for the VP3 peak.

3.3. Total VP Concentration Linear Range

The combined method of methanol precipitation and stacking injection had good linearity for total VPs, which expanded from 2.0 × 10¹² to 3.0 × 10¹³ GC/mL with R ² = 0.988, as shown in Figure 2.

FIGURE 2.

Linear range of total VP concentration. Each data point was the results of three injections (error bar in each point).

Despite the linearity across the relatively wide concentration range, the determined VP ratio was found to be inconsistent due to two primary factors. Initially, at the lower end of the total VP range, the peaks for VP1 and VP2 fell below the LOQ, resulting in a decreased representation in their peak areas. Conversely, at the upper end of the range, the detection response for the principal component, VP3, exhibited slight saturation, which consequently diminished the percentage of the VP3 peak area. Therefore, the sample concentration range yielding a consistent VP ratio should be more restrictive than the overall linear range.

3.4. Typical Sample Loading Concentration and Concentration Range for Consistent VP Ratio Determination

To find the concentration range that had the consistent determined VP ratio, we first determined the typical sample loading concentration by reviewing the data. The concentration was determined to be 1.23 × 10¹³ GC/mL for the AAV9 sample. The method precision at this concentration point, as shown in Table 1, was evaluated by running the concentration for 3 days, with three injections in each day. Figure 3 shows the representative electropherograms of the sample at the concentration obtained in 3 days. The method was reproducible.

TABLE 1.

Precision of the method at the typical loading concentration (1.23 × 10¹³ GC/mL).

	VP3	VP2	VP1
Day 1 ave	91.4	4.8	3.7
Day 2 ave	91.7	4.7	3.6
Day 3 ave	91.3	5.1	3.7
Ave	91.5	4.8	3.7
RSD%	0.2	4.2	1.4

Abbreviation: RSD%, relative standard deviation.

FIGURE 3.

Three‐day representative electropherograms of VPs at AAV concentration of 1.23 × 10¹³ GC/mL (B‐D) and the reagent blank (A).

After the typical loading concentration was determined, we expanded the concentration range for test surrounding this concentration point in 1.23 × 10¹³ ± 50% GC/mL (6.15 × 10¹² to 1.85 × 10¹³ GC/mL) of the AAV9 sample to see if the concentration range showed consistent VP ratio. This AAV9 concentration range ensured both the VP1 and VP2 lowest concentrations were above LOQ, and the VP3 highest concentration was in the linear range as shown in Figure 2. Figure 4 shows the electropherograms of the AAV9 samples in the concentration range of 1.23 × 10¹³ ± 50% GC/mL. The three VPs had good linearity in the concentration range with R ² > 0.998. In this AAV9 concentration range, the VP ratio was consistent as shown in Table 2. The relative standard deviation (RSD%) for three VP's percentage determination was within 7%, even for minor components VP2 and VP1.

FIGURE 4.

AAV9 in the concentration range 6.15 × 10¹² to 1.85 × 10¹³ GC/mL.

TABLE 2.

Viral protein (VP) ratio at different adeno‐associated viruses (AAV) concentration in the concentration range of Figure 4.

Concentration ×1012 GC/mL	VP3 (%)	VP2 (%)	VP1 (%)
6.2	92.2	4.6	3.2
6.2	92.3	4.2	3.5
6.2	92.1	4.5	3.5
9.3	91.4	4.9	3.7
9.3	91.6	4.8	3.7
12.3	91.1	5.1	3.7
12.3	91.4	5.0	3.6
15.4	91.0	5.2	3.9
15.4	91.2	5.1	3.7
15.4	90.8	5.3	3.9
18.5	90.9	5.3	3.9
18.5	91.1	5.1	3.8
18.5	91.2	5.1	3.8
Ave	91.4	4.9	3.7
RSD%	0.5	6.7	5.4

Abbreviation: RSD%, relative standard deviation.

3.5. Application to Other Serotypes

The method was applied to AAV2 and AAV8 samples. Figure 5 shows typical electropherograms of the two AAV samples. For the two samples, AAV2 and AAV8, a front peak was separated before VP3 peak as reported [15]. It was identified as VP3 clip by reversed‐phase liquid chromatography–mass spectrometry method [20]. The measured VP ratios of the two samples by our method are VP3:VP2:VP1 = 12.4:1.0:1.3 for AAV2 and 7.1:1.4:1.0 for AAV8.

FIGURE 5.

Typical electropherograms: (A) AAV2 (5.0 × 10¹² GC/mL), (B) AAV8 (1.0 × 10¹³ GC/mL), and (C) blank run only with internal standard (IS).

4. Conclusions

The CE‐SDS method developed in this study was designed to ensure the consistent determination of AAV VP ratios across various sample concentrations. UV‐absorption detection was chosen for this purpose as it negated the need for any labeling in the process. To compensate for the lower sensitivity associated with UV‐absorption detection, methanol precipitation was employed during sample preparation, and a stacking injection technique was adapted under CE conditions. These two techniques were found to be highly compatible.

The precipitation method effectively concentrated the samples into a consistent volume of 10 µL in the sample buffer, irrespective of the initial sample volume and concentration. This concentrated sample was then diluted to the standard analysis volume of 50 µL with water, creating a solution with low conductivity, which is crucial for effective stacking injections.

The method demonstrated robust linearity for AAV serotype 9 concentrations and exhibited good multiday precision. The VP ratios were successfully determined in AAV9 samples across a concentration range of 6.15 × 10¹² to 1.85 × 10¹³ GC/mL. The RSD% for VP ratio determination within this concentration range was maintained below 7% for VP3, VP2, and VP1. Additionally, this method was further applied to other AAV serotypes, including AAV2 and AAV8, illustrating its versatility and broad applicability.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.