Optimizing Adeno-Associated Virus (AAV) Capsid Protein Analysis Using UPLC and UPLC-MS

This application note demonstrates the performance of the BioAccord System by developing LC-optical and LC-MS methods for the analysis of intact AAV capsid proteins and to apply these methods to the AAV8 serotype as a case study for the improved characterization of capsid proteins, including their identification, stoichiometry, and post-translational modifications. The developed methods were then applied to additional rAAV serotypes to demonstrate their general applicability for the intact protein analysis of AAV vectors.

Benefits

• An optimized LC-MS solution for the separation and intact mass analysis of viral proteins (VPs) from recombinant AAV vectors to support the characterization and development of gene therapy products
  
• A sensitive and quantitative LC-FLR method for stoichiometry measurement of AAV capsid proteins

Recombinant adeno-associated viruses (rAAVs) are the most widely used vectors in gene therapy development due to their low toxicity and long-term expression ability.¹ With different tissue tropism, a total of 13 common serotypes of AAVs were isolated, and many of them have been explored for the treatment towards multiple diseases.² As the protector of the viral genome and mediator of cellular internalization, the AAV capsid (Figure 1) consists of proteins that share high sequence homology across serotypes,³ requiring reliable and specific methods for vector identification during gene therapy development and commercialization. In addition, the AAV capsid composition is reported to be critical to viral infectivity and gene transduction.^4,5 To ensure the safety and quality of drug products, the structure and properties of the AAV capsid and its constituent proteins need to be well characterized and monitored throughout the gene therapy product development process. Conventional analytical techniques, such as ELISA, Western Blot, or SDS-PAGE, are frequently used to provide basic functional and compositional information, but these techniques are either laborious to deploy and validate or insensitive to AAV serotype.⁶ To this end, a robust, specific method that can provide reliable results in a timely fashion is highly desired for identification and characterization of AAV capsid proteins.

Mass spectrometry (MS) has been widely adopted for structural analysis of proteins for its sensitivity and specificity. However, extensive training has been traditionally required for users to operate MS instruments and develop methods, limiting the wider deployment of MS technology in the biopharmaceutical industry. Designed as a robust, easy-to-use, and small footprint LC-MS platform, the Waters BioAccord System was developed to deliver fit-for-purpose MS analysis of biotherapeutics accessible to organizations and operators not previously able to deploy LC-MS technologies. As demonstrated in previous publications,^7,8,9 the BioAccord System renders a highly reproducible chromatographic separation and accurate mass measurement in an automated manner through the integration of the ACQUITY UPLC I-Class PLUS System and the ACQUITY RDa Mass Detector. The system, operating under the compliant-ready waters_connect informatics platform is ideally suited for gene therapy product development and commercialization teams to provide structure, composition, and identity information for rAAV capsids.

The objectives of this application note are as follows: To demonstrate the performance of the BioAccord System by developing LC-optical and LC-MS methods for the analysis of intact AAV capsid proteins and to apply these methods to the AAV8 serotype as a case study for the improved characterization of capsid proteins, including their identification, stoichiometry, and posttranslational modifications. The developed methods were then applied to additional rAAV serotypes to demonstrate their general applicability for the intact protein analysis of AAV vectors.

Chemical and reagents

Multiple serotypes of AAV samples were donated by BioReliance (Rockville, MD, USA) or purchased from Vigene Bioscience (Rockville, MD, USA), including AAV1, AAV2, AAV5, AAV6, AAV8, and AAV9. Acetic acid was purchased from Sigma-Aldrich (St. Louis, MO, USA). LC-MS-grade water and acetonitrile were purchased from Honeywell (Charlotte, NC, USA) and used as received. Waters IonHance difluoroacetic acid (DFA) (p/n: 186009201) was used as the additive to prepare the mobile phase.

Sample preparation

AAV samples were diluted with Milli-Q water to a final physical titer of 1 × 1013 GC/mL or used as is if the received sample concentration is lower than 1 × 1013 GC/mL. As previously reported,⁶ AAV samples were treated with acetic acid at 10% (v/v) concentration for 15 min, then centrifuged at 12,000 rpm for 5 min. For LC-MS analysis, a 10-μL (~1 μg of proteins) AAV sample was used for each injection. For LC-fluorescence (FLR) analysis, a 1-μL (~0.1 μg of proteins) AAV sample was used, unless stated otherwise.

System Settings

RPLC-MS Method Optimization

AAV capsids are composed of three viral proteins (VP1, VP2, and VP3) at an approximately 1:1:10 ratio with the masses between 50 and 85 kDa. Due to limited sample availability and the difference in protein relative abundances, the characterization of AAV capsid proteins has been challenging through conventional LC-MS-based methods.6 When performing reversed-phase LC-MS analysis of AAV8 proteins using formic acid as mobile phase modifier, all VP proteins co-eluted as a single TIC peak (Figure 2A). The masses of VP1 and VP2 were not obtained upon the deconvolution of the summed MS spectra (Figure 2A inset), which we attribute to ion suppression from the more abundant VP3 ions during co-elution. To improve the separation of VP proteins, a new RPLC-MS method using difluoroacetic acid as the mobile phase additive was developed on an ACQUITY UPLC BEH C₈ Column. Compared to the conventional mobile phase modifiers, such as formic acid, chromatographic resolution among the capsid proteins was improved while maintaining sufficient MS sensitivity (Figure 2B).¹⁰ The previously co-eluting VP1 and VP2 species were now resolved from VP3, with a third additional peak observed at 12.2 min. This additional peak was assigned as a fragment of VP3 protein (labeled as “VP3 clip”) based on intact mass results. To further optimize the separation of the VP proteins, columns packed with BEH particles bearing alternative bonded phases (e.g., C₄ and C₁₈ ligands) were evaluated. As shown in Figure 2C, VP1 and VP2 were further resolved on the C₄ column, which can be attributed to the increased selectivity provided by the C₄ bonded phase. The pore size of the packed particles, in addition to selectivity of the surface chemistry, contributed to the improved separation. Compared to the use of the ACQUITY UPLC BEH C₈ Column packaged with particles of 130 Å pore size, the peak width was reduced using the wide pore (300 Å) C₄ column (Figure 2C), resulting in additional enhancement in MS response.

With the optimized mobile phase and column, the improved chromatographic resolution facilitated detailed MS analysis of individual VP proteins (Figure 3A). Loading 0.5 μg of AAV8, MS data were obtained for all three capsid proteins and their variants (Figure 3B–E). The deconvoluted masses for peaks at 6.74 min, 7.10 min, and 8.32 min were observed to be 81,668 Da, 66,518 Da, and 59,805 Da, respectively. As shown in Table 1, the observed masses of these proteins were consistent with theoretical masses of VP1, VP2, and VP3,6 confirming that the developed method is suited for measuring viral protein masses. In addition, the accurate mass measurement of variants allows assignments of potential post-translational modifications (PTMs) on the VPs, including acetylation and phosphorylation, while confirming removal N-terminal methionine on VP1 and VP3. The later eluting peak at 9.38 min was found to have a mass of 50,592 Da, matching the MW of a VP3 fragment due to labile Asp659-Pro660 bond hydrolysis.¹¹ Together, these results demonstrated the capability of the method for developmental characterization of AAV-related products.

While the theoretical ratio of VP1:VP2:VP3 is at 1:1:10, it has been reported that the production process can impact the relative abundances of VP proteins.¹² The stoichiometry of VP proteins can affect vector potency,¹³ making it necessary to develop a routine and reliable method to measure the ratio for the production of AAV-related drugs. The increased resolution achieved by the developed RPLC method allows the use of optical signals, such as UV or fluorescence (FLR), to directly measure the abundance of individual VP and calculate the VP1:VP2:VP3 ratio using the integrated peak areas.

Figure 4 shows the calculated relative abundance for each resolved chromatographic peak. The sensitivity of the method can be enhanced by selecting the protein intrinsic fluorescence response (λ_excitation: 280 nm, λ_emission: 350 nm) as the detection signal. With FLR detection (Figure 4B), the signal-to-noise ratio of VP3 was enhanced by approximately 50-fold compared to UV detection (Figure 4A). The use of fluorescence detection facilitates highly sensitive (50 ng) and reliable peak integration for AAV capsid proteins. To this end, the developed LC-FLR method can be used for rAAV samples with lower concentration to check the sample purity or serve as a serotype identity test for product release.

Applying the Developed Methods to other rAAV Serotypes

To evaluate the broader applicability of the developed methods for intact mass and stoichiometry of other rAAV serotypes, six serotypes of AAV (1, 2, 5, 6, 8, and 9) obtained from a different supplier were analyzed. Figure 5 shows the majority of these AAV serotypes (AAV1, AAV6, AAV8, and AAV9) have a comparable chromatographic profile. Furthermore, a front shoulder to the VP3 peak with the same mass as VP3 was observed (peak annotated with *) on these serotypes, which could be a structural isomer of VP3. Phosphorylation was observed for VP1 and VP2, while acetylation was also detected on VP1 and VP3 on all AAV serotypes. Table 2 lists the observed and theoretical masses of the VP proteins as well as the assignment based on the matches of these values. These data illustrate that the developed method has broader applicability for efficient separation and mass measurement of VPs across AAV serotypes.

While the developed method is effective in the separation of most AAV serotypes in the study, the chromatographic separation profiles differ for AAV2 and AAV5 samples. Co-elution of VP1 and VP2 of AAV2 was observed (Figure 5B), where a shallower gradient might be needed to fully resolve the peaks. VPs from serotype AAV 5 (Figure 5C) display a similar chromatographic profile as AAV2 although the deconvoluted MS spectra for the peak at 6.95 min did not show VP1. This errant behavior prompted us to explore additional experimental conditions to improve the analysis. An alternative column chemistry, an ACQUITY UPLC BEH C₁₈ Column, was then evaluated to see if more desirable results can be obtained. Figure 5H shows a total ion chromatogram from the analysis of the AAV5 sample using the ACQUITY UPLC BEH C₁₈ Column under the optimized gradient condition. The use of the ACQUITY UPLC BEH C₁₈ Column leads to an increased recovery of VP1 protein, which now elutes later than VP3 (Figure 5H). This change in the elution order suggests the higher hydrophobicity of VP1 in AAV5, which is also suggested by the more hydrophobic residues (such as proline and phenylalanine) in its amino acid composition. Collectively, these results demonstrated that the developed LC-FLR/MS-based method is effective for the measurement of rAAV capsid proteins from a range of AAV serotypes that are commonly seen in gene therapy development.

In this work, an optimized RPLC method was developed using DFA as a mobile phase modifier for improved separation of AAV capsid proteins, while retaining adequate mass spectrometry sensitivity. The combination of the mobile phases and judicious selection of column chemistry results in a chromatographic separation that facilitates intact mass measurement of individual VP isoforms and their variants. The developed method also renders a means to quantify VP stoichiometry through LC-optical response at sub-microgram levels. Overall, results acquired on the BioAccord System show that this platform can deliver the structural information necessary to improve the understanding of gene therapy products throughout product development and commercialization.

1. Finer, M.; Glorioso, J. A Brief Account of Viral Vectors and Their Promise for Gene Therapy. Gene Ther, 2017; 24: 1–2.
   
2. Lisowski, L.; Tay, S.S.; Alexander, I.E. Adeno- Associated Virus Serotypes for Gene Therapeutics. Curr Opin Pharmacol, 2015, 24: 59–67.
   
3. Van Vliet, K.; Mohiuddin, Y.; McClung, S. et al. Adeno-Associated Virus Capsid Serotype Identification: Analytical Methods Development and Application. J Virol Methods, 2009, 159: 167–177.
   
4. Girod, A.; Wobus, C.E.; Zadori, Z. et al. The VP1 Capsid Protein of Adeno-Associated Virus Type 2 is Carrying a Phospholipase A2 Domain Required for Virus Infectivity. J Gen Virol, 2002, 83: 973–978.
   
5. Popa-Wagner, R.; Porwal, M.; Kann, M. et al. Impact of VP1-Specific Protein Sequence Motifs on Adeno Associated Virus Type 2 Intracellular Trafficking and Nuclear Entry. J Virol, 2012, 86: 9163–9174.
   
6. Jin, X.; Liu, L.; Nass, S.; O'Riordan, C.; Pastor, E.; Zhang, K. Direct Liquid Chromatography/Mass Spectrometry Analysis for Complete Characterization of Recombinant Adeno-Associated Virus Capsid Proteins Human Gene Therapy Methods. Hum Gene Ther Methods, 2017, 255–267.
   
7. Shion, H. et al. Enabling Routine and Reproducible Intact Mass Analysis When Data Integrity Matters. Waters Application Note, 720006472EN, 2019.
   
8. Ranbaduge, N. et al. Routine Peptide Mapping Analysis Using the BioAccord System. Waters Application Note, 720006466EN, 2019.
   
9. Zhang, X. et al. Released N-linked Glycan Analysis Using the BioAccord System. Waters Application Note, 720006474EN, 2019.
   
10. Kellett, J. et al. Application of Difluoroacetic Acid to Improve Optical and MS Performance in Peptide LC-UV/MS. Waters Application Note, 720006482EN, 2019.
    
11. Vlasak, J.; Ionescu, R. Fragmentation of Monoclonal Antibodies. MAbs, 2011; 3(3): 253–263.
    
12. Snijder, J.; van de Waterbeemd, M.; Damoc, E. et al. Defining the Stoichiometry and Cargo Load of Viral and Bacterial Nanoparticles by Orbitrap Mass Spectrometry. J Am Chem Soc, 2014, 136: 7295–7299.
    
13. Bosma, B.; Plessis, F.; Ehlert, E.; Nijmeijer, B.; Haan, M.; Petry, H.; Lubelski, J. Optimization of Viral Protein Ratios for Production of rAAV Serotype 5 in the Baculovirus System. Gene Ther, 2018, 25: 415–424.