Waters - An Anion-Exchange Chromatography Method for Monitoring Empty Capsid Content in Adeno-Associated Virus Serotype AAV8

During the recombinant adeno-associated virus (AAV) biomanufacturing process, capsids that do not contain ssDNA (empty capsids) are also produced. Empty capsids are unable to deliver ssDNA to the targeted cells and as a result are important to monitor during process development and product manufacturing. An anion-exchange chromatography (AEX) separation driven by differences in surface charge can monitor changes in empty capsid levels (1,2). AEX using fluorescent detection consumes small sample amounts and can provide reliable and reproducible results.

Experimental Conditions

Samples: AAV8-CMV-GFP (full capsid) and AAV8-null (empty capsid) 2 × 10¹² capsids/mL; other AAV-CMV-GFP serotypes: ranging from 1.3 × 10¹³ to 6.7 × 1013 GC/mL = genome copies (physical titer)/mL

LC system: ACQUITYTM UPLCTM H-Class Bio

Sample temp.: 10 ̊C

Column temp.: 30 ̊C

Injection volume: 0.2–6 μL

Column: Waters Protein-Pak Hi Res Q, 4.6 × 100 mm

Fluorescence detector: Excitation: 280 nm; Emission: 350 nm

Gradient: 70 mM bis-tris propane, pH 9.0, 100–300 mM tetramethylammonium chloride in 20 min at 0.4 mL/min

Results and Discussion

The AEX separation of empty and full AAV8 capsid was optimized for several parameters including pH and salt types (3). In addition, intrinsic protein fluorescence was monitored to provide greater sensitivity and selectivity, and less baseline drift (2). The separation achieved is shown in Figure 1. We observe that the ssDNA “empty” capsids are partially coelute with the ssDNA “full” capsids while showing no significant change in the UV absorbance ratio (280 nm/260 nm, data not shown), indicating the unlikelihood of monitoring AAV8 with partial ssDNA by AEX.

Figure 1: AAV8 ssDNA (CMV-GFP) empty and full capsids are separated on a Protein-Pak Hi Res Q Column using an optimized AEX method (see experimental conditions for method details).

A series of chromatograms for AAV8 samples with differing percentages of empty capsids is presented in Figure 2. The mixtures were generated from >99% pure “empty” and “full” samples. The sample concentrations were normalized based on FLR peak areas from a size-exclusion chromatography (SEC) separation with fluorescence detection and assumed a “full” to “empty” fluorescence response factor (RF_F/E) of 1.3 (2). For AEX, however, 1.9 was calculated for RF_F/E (Area_{100% Full}/Area_{100% Empty}). The discrepancy in RFF/E values between SEC and AEX may be the result of changes in quantum efficiency under different mobile-phase conditions. RFF/E is then divided into the full capsid peak area to calculate the percentage of empty capsid in a sample (Figure 2). The accuracy of this approach depends on the veracity of RFF/E, which requires reasonably pure “full” and “empty” samples, and accurate capsid concentrations. While RF_F/E was not rigorously defined in this work, the estimate of RF_F/E used here effectively demonstrates the approach to interpretation of the AEX data. Determining RFF/E is not essential, if only relative comparisons of empty capsid levels between samples are required.

Figure 2: Shown on the left is an overlay of AEX chromatograms for AAV8 samples with differing levels of empty capsid. On the right is the correlation between predicted and measured empty capsid content. Calculation of the percentage of empty capsid was based on the equation shown, which corrects for the difference in fluorescence response between empty and full capsids. A_Empty and A_Full are the AEX peak areas of the empty and full capsids, and RF_F/E is the fluorescence response factor.

Samples of several AAV-CMV-GFP serotypes were also evaluated (Figure 3). All tested serotypes were retained on the column, but their retention varied significantly.

Figure 3: Chromatograms of AAV1, AAV2, AAV5, AAV6, AAV8, and AAV9 serotypes (full capsids, CMV-GFP ssDNA). For comparison, the bottom chromatogram shows the optimized separation of AAV8 empty and full capsid. Separation conditions are as described in the text for all serotypes except for AAV9 (*) where the salt gradient was 0–200 mM KCl.

Conclusions

AEX will not likely provide a measurement of AAV8 capsid carrying partial ssDNA, therefore complementary methods such as analytical ultracentrifugation (AUC) would be needed for that analysis. However, due to the technical challenges and costs of these methods, AEX may still provide utility as an initial screening tool in support of manufacturing process development and as a more precise assessment of changes in empty capsid content in product quality testing.

Due to differences in capsid protein surfaces, method optimization may be needed when separating empty and full capsids of other AAV serotypes, and the AEX method described here can serve as a useful starting point.

References

(1) X. Fu, et al, Hum. Gene. Ther. Methods 30(4), 144 (2019).

(2) C. Wang, et al, Mol. Ther. Methods Clin. 19, 257 (2019).

(3) H. Yang, et al, Waters Application Note 720006825EN (2020).

Waters, The Science of What’s Possible, Protein-Pak, ACQUITY, and UPLC are trademarks of Waters Corporation. All other trademarks are the property of their respective owners.

Waters Corporation

34 Maple Street, Milford, MA 01757

tel. 508 478 2000

Website: www.waters.com