Molecular weight determination of adeno‐associate virus serotype 8 virus‐like particle either carrying or lacking genome via native nES gas‐phase electrophoretic molecular mobility analysis and nESI QRTOF mass spectrometry

Abstract Virus‐like particles (VLPs) are proteinaceous shells derived from viruses lacking any viral genomic material. Adeno‐associated virus (AAV) is a non‐enveloped icosahedral virus used as VLP delivery system in gene therapy (GT). Its success as vehicle for GT is due to its selective tropism, high level of transduction, and low immunogenicity. In this study, two preparations of AAV serotype 8 (AAV8) VLPs either carrying or lacking completely genomic cargo (i.e., non‐viral ssDNA) have been investigated by means of a native nano‐electrospray gas‐phase electrophoretic mobility molecular analyzer (GEMMA) (native nES GEMMA) and native nano‐electrospray ionization quadrupole reflectron time‐of‐flight mass spectrometry (MS) (native nESI QRTOF MS). nES GEMMA is based on electrophoretic mobility principles: single‐charge nanoparticles (NPs), that is, AAV8 particle, are separated in a laminar sheath flow of dry, particle‐free air and a tunable orthogonal electric field. Thus, the electrophoretic mobility diameter (EMD) of a bio‐NP (i.e., diameter of globular nano‐objects) is obtained at atmospheric pressure, which can be converted into its MW based on a correlation. First is the native nESI QRTOF. MS's goal is to keep the native biological conformation of an analyte during the passage into the vacuum. Subsequently, highly accurate MW values are obtained from multiple‐charged species after deconvolution. However, once applied to the analysis of megadalton species, native MS is challenging and requires customized instrumental modifications not readily available on standard devices. Hence, the analysis of AAV8 VLPs via native MS in our hands did not produce a defined charge state assignment, that is, charge deconvolution for exact MW determination was not possible. Nonetheless, the method we present is capable to estimate the MW of VLPs by combining the results from native nES GEMMA and native ESI QRTOF MS. In detail, our findings show a MW of 3.7 and 5.0 MDa for AAV8 VLPs either lacking or carrying an engineered genome, respectively.


| INTRODUCTION
Gene therapy (GT) aims to treat, or cure, a specific disease whose origin is linked to mutation(s) or incorrect expression of a gene. 1 The approach involves delivering an engineered genomic load to add, replace, or interfere with the genetic layout of a cell in question to modify and correct it. 2 The genomic cargo delivery relies on specific vehicles, which are generally grouped into viral and non-viral vectors based on their origin. Both groups have advantages and limitations; non-viral vectors are usually easier to synthesize and assemble than viral ones but have lower transduction efficiency. 3 Instead, viral vectors can efficiently transport their cargo to the target but are often hindered by higher immunogenicity. 4 In the vector-mediated gene therapy realm, adeno-associated virus (AAV) is the leading vehicle thanks to its high efficiency of transduction and low immunogenicity, as demonstrated by the growing number of clinical trials based on this delivery system. [5][6][7][8] AAV is a member of the family of Parvoviridae, genus Dependoparvovirus. It can accommodate up to 4.7 kb of singlestranded DNA (ssDNA) in a non-enveloped, proteinaceous capsid of approximately 26 nm in diameter. According to several sources, a molar ratio of 1:1:10 of the 60 viral proteins VP1, VP2, and VP3 arranged in a T = 1 icosahedral symmetry forms the protein shell. 4,5,[9][10][11] In nature, the AAV group is composed of 13 natural serotypes, each with a preferred tropism toward a specific tissue, thus making AAV a robust system for the transduction of specific cell types. 5,12 In this study, AAV serotype 8 (AAV8) has been used to produce virus-like particles (VLPs). VLPs are proteinaceous "empty" shells derived from viruses, which can be used as vaccine 13,14 or as viral vector for the delivery of genetic material or other therapeutics, [15][16][17] making them a highly adaptable platform. 18 They are non-infectious because the original viral genome is no longer present; instead, engineered genetic material can be encapsulated. In our study, two AAV8 preparations were available for analysis: (i) a so-called "empty" one composed of solely the proteinaceous capsid lacking any genomic cargo and (ii) a so-called "filled" preparation with an encapsulated engineered (non-viral) genome. These two types of preparations were analyzed via native nano-ES (electrospray) gas-phase electrophoretic mobility molecular analysis (nES GEMMA) and with a native nano-ESI (electrospray ionization) quadrupole reflectron time-of-flight mass spectrometry (nES QRTOF MS).
The nES GEMMA device, as first described by Kaufman et al., 19 is a suitable platform for analyzing proteins, viruses, VLPs, liposomes, and several nanoparticles and bionanoparticles, as demonstrated by various studies. [20][21][22][23][24][25] The system is also known under the name of differential mobility analyzer (nES DMA), macro ion mobility spectrometer (macroIMS), LiquidScan ES, or scanning mobility particle sizer (SMPS), all describing the same conceptthe size-separation of surface-dry, single-charged (bio-)nanoparticles in the gas-phase at atmospheric pressure.
The nES GEMMA device is composed of three distinct units: (i) The nES source electrosprays the analytes dissolved in a volatile electrolyte solution, while charge equilibration for the production of a polydisperse aerosol of single-charged ions is achieved through a bipolar atmosphere generated by a radioactive source (e.g., 210 Po α-particle emitter), 26 a soft X-Ray charger, 27,28 or an alternating bipolar corona discharge process. 29 (ii) A differential mobility analyzer unit, where a laminar sheath flow of particle-free, dried air at atmospheric pressure, and an orthogonal tunable electric field, are used to achieve nanoparticle separation (i.e., gas-phase electrophoresis). The generated monodisperse (monomobile) aerosol is introduced in (iii), a condensation particle counter, where its elements (i.e., the bionanoparticles) act as condensation nuclei for droplet formation due to the supersaturated atmosphere of either n-butanol or water. By means of a laser beam, the formed μm-sized droplets were detected as well as counted after size separation allowing particle-number concentrations to be obtained. It is important to note that particle size determination occurs in the gas-phase at atmospheric pressure.
Hence, nES GEMMA yields the surface-dry particle's size diameter (electrophoretic mobility diameter, EMD). 30,31 Therefore, for AAV8 VLPs, given the approximately spherical shape (i.e., icosahedral) and non-enveloped origin (proteinaceous-only capsid), the detected EMD can be directly correlated to the nanoparticles' diameter. Hence, the obtained EMD can be easily converted with good approximation in a molecular weight value thanks to an EMD/M W correlation based on VLPs MS-derived data. 22 in the applied instrument. 22 Native MS proved to be essential, and capable, for studying noncovalent protein-ligand 32 and protein-protein interactions, 33 protein complexes, 34 and supramolecular protein structures like viruses 23,35 and VLPs. 22,[34][35][36] The main challenge for this MS approach is to desorb/ionize and detect the multiple-charged analytes while preserving their labile non-covalent interactions and structure. Nonetheless, several VLPs have been successfully analyzed by employing commercially available MS instrumentation, such as nESI orbitrap [37][38][39][40] and nESI charge detection mass spectrometry (CDMS). 41,42 In our case, we employed a Synapt G1 (Waters Manchester, UK) modified by MS Vision (Almere, The Netherlands) to study AAV8 nanoparticles. The instrument is equipped with a nano-electrospray ionization source for the production of multi-charged bionanoparticles, and with several custom modifications to properly fine-tune the necessary settings (e.g., application of collision and cooling gas, vacuum levels, and voltage settings) for successful analysis. 43 Precise M W determination can be assessed via deconvolution of the charge state assignment of the detected bionanoparticles.
In this manuscript, our focus is to combine native nES GEMMA and native nESI QRTOF MS data as well as an EMD/M W correlation to expand, with great accuracy, the knowledge about the nanoparticles' size, sample quality, and molecular weight of AAV8 nanoparticles, either carrying or lacking an engineered genomic cargo.
For nES GEMMA as well as nESI QRTOF MS analysis, a buffer exchange step against 40-mM NH 4 OAc was carried out employing 10-kDa MWCO centrifugal filters (polyether sulfone membrane from VWR, Vienna, Austria). After three spin filtration repetitions (9.0 Â 10 3 g for 5 min each), the retentate was collected. Based on asymmetric flow field-flow fractionation (AF4 also known as AFFFF) analysis, the estimated final sample concentration for "empty" AAV8 VLPs was 22 μg/ml, while for "filled" AAV8 VLPs it valued 8.5 μg/ml.

| nES GEMMA
nES GEMMA analyses were carried out on a TSI Inc instrument (Shoreview, MN, USA), which consisted of a nano-electrospray unit with a charge reduction source (model 3480 including a 210 Po charge equilibration device), an electrostatic classifier equipped with a nanodifferential mass analyzer (nano-DMA; model 3080) and an n-butanol driven ultrafine condensation particle counter (CPC; model 3025A) for particle detection. For the spraying process, the nES unit is equipped with a 24 cm long, polyimide coated, fused-silica capillary with an inner diameter of 25 μm (Polymicro Technologies, a subsidiary of Molex; Phoenix, AZ, USA). The capillary is manually cut and tapered with a home-built grinding machine based on the work of Tycova et al. 44 Nanoparticle separation and detection were achieved by using the following settings: The filtered airflow on the nES generator was set to 1.6 Â 10 À5 m 3 /s (1 Lpm), the CO 2 gas flow to 1.6 Â 10 À6 m 3 /s

| nES QRTOF MS
A Synapt G1 (Waters, Manchester, UK) was modified by MS Vision (Almere, The Netherlands) in order to maximize ion transmission for native nESI MS in the kilodalton to megadalton range. This was achieved by (i) increasing the operating pressure of the first vacuum stage (source region) by a manually controlled throttle valve (i.e., 5 to 10 mbar); (ii) fine tuning of the second vacuum stage (transfer pressure region) by fitting a sleeve that restricts pumping of the gas entering from the source region; (iii) installation of a 32 kDa quadrupole mass filter; (iv) amenities to bleed cooling gas like Ar of Xe into the ion mobility stage of the instrument at optimal pressures for cooling and desolvation as well as for independent control of trap and transfer collision cell pressures; (v) customized data acquisition settings (profile binning) and pusher pulse interval (i.e., 128 μs) were adjusted to improve ion detection at ultrahigh mass range. Sample introduction was performed by a nESI source employing manually opened in-house pulled spray capillary. Sample concentration was chosen in order to achieve best results (i.e., avoid clogging of the tip and allow extremely long acquisition time). Spray capillary voltage was set to obtain ideal spraying condition (i.e., ranging between 1 to 2.5 kV). Gas pressures in the ion source region and in the ion mobility chamber (specifically the TriWave™ cell) before the orthogonal RTOF were finely tuned in order to increase ion transmission. Moreover, a relative high collision induced dissociation voltage (ranging up to 90 V) was applied to increase desolvation and optimize transmission efficiency. 45,46 The investigated mass range was between m/z 1000 and 40,000 in the positive ion mode. Mass spectra were analyzed using MassLynx (Waters, Manchester, UK) and OriginPro 9.1 (OriginLab, Northampton, MA, USA).

| Native nES GEMMA analysis of AAV8 VLPs
Gas-phase electrophoresis of several VLPs-based on bacteriophages, a norovirus serotype, hepatitis B virus, cowpea mosaic virus and a human rhinovirus-yielding surface dry particle EMDs has already been described. 22,23,42,47 In addition, AAV8 VLPs have likewise been measured via gas-phase electrophoresis as described in a previous work focusing on VLP aggregation (submitted manuscript). Focusing on the molecular weight of bionanoparticles in the current manuscript, Figure 1A depicts the nES GEMMA spectra of "empty" (blue profile) and "filled" (red profile) AAV8 VLPs in their native state. In order to better appreciate the fine difference between the two preparations, Figure 1B shows the magnification between 22-and 29-nm EMD of Figure 1A. The slight difference in the EMD size is enough to discriminate between the two sample preparations. To confirm this observation, a statistical evaluation over more than 5000 capsids per preparation (n = 3 independent nES GEMMA measurements, each) was made. Results show an average EMD of 25.10 ± 0.18 nm and 25.93 ± 0.07 nm for "empty" and "filled" AAV8 VLPs, respectively.
The difference in EMD is based on the stabilizing effect promoted by the genomic material inside the capsid of "filled" AAV8 VLPs. Lack of the genomic material as a scaffold in the working environment condition of nES GEMMA causes the partial shriveling of the capsid, hence reducing its EMD.

| nES GEMMA-based molecular weight determination
The correlation between EMD data, obtained from nES GEMMA measurements, and the M W of several VLPs or virus particles, either from literature or measured via MS instrumentation, has already been reported. 22,23 The application of the EMD/M W correlations provided in the studies mentioned above is presented in Figure 2. The data produced via nES GEMMA analysis for AAV8 generate M W of 3670 ± 69 kDa ( Figure 2A) and 4751 ± 47 kDa ( Figure 2B) for "empty" and "filled" capsids, respectively. A summary of M W values is presented in Table 1.
The M W resulting from the EMD/M W correlation for the "empty"

| Native nES QRTOF MS analysis of AAV8 VLPs
The analysis of VLPs in their native state is a delicate and laborious job. In this study, megadalton-range species were targeted, which further increased the analytical challenges. The biggest challenge for analyzing such massive species is the passage of desorption/ionization region and transfer into the vacuum part of the mass spectrometer.
Parameters like sample concentration, quality and shape of the capillary tip, and the mass spectrometer's pressure in the first two differentially pumped vacuum stages greatly influenced the outcome. The response to each of these settings was rather drastic, to the magnitude where analytes' detection was either successful or not.
In Figure 3A, the positive ion mass spectra of AAV8 VLPs, either "empty" (blue profile) or "filled" (red profile), are shown. The blue profile shows a single dominant peak with an apex center at 23,047 m/z. At the same time, the red profile shows two peaks, a dominant one at 23,205 m/z and a second at 31,092 m/z. Although charge resolution was not achieved and hence no molecular weight determination based on peak charge assignment was possible, it is highly plausible that the detected peaks belong to "empty" (label e) and "filled" (label f ) AAV8 VLPs. Further support comes from the presence of a shared peak between the two preparations (i.e., label e, Figure 3B) because the "filled" AA8 VLPs preparation contains at least 33% of AAV8 VLPs lacking genomic cargo.
Consequently, the peak labeled f, detected only in the "filled" AAV8 preparation ( Figure 3C) represents the portion of capsids carrying the genomic cargo. Moreover, although the concentration of "filled" capsids in the sample exceeds "empty" ones' concentration, this is not reflected in the mass spectra. This discrepancy can be explained by a lower transmission efficiency due to the increased F I G U R E 2 Electrophoretic mobility diameter (EMD)/M W correlations for (A) "empty" virus-like particles (VLPs) and (B) "filled" VLPs (i.e., intact virus). Readapted with permission from Weiss et al. 22 Therefore, based on (i) the assumption that the apex center of peak e (i.e., 23,047 m/z) in Figure 3A,B is generated only by "empty" mono- Because the genome encapsulated in the proteinaceous capsid is shielded from the external environment, we suppose that it does not affect the number of charges enveloping the capsid but only its molecular weight. To support this claim, the same range of positive charges assigned to peak e, have been applied to peak f (i.e. 31,092m/z, Figure 3C). Thus, a M W of 4959 ± 78 kDa is obtained. As a result, this calculation highly correlates with the molecular weight for 'filled' VLPs obtained from the EMD/M W correlation (i.e., 4751 ± 47 kDa, difference 4.4%) or from the expected theoretical M W mentioned before (i.e., 4988 kDa, difference 0.6%). Moreover, because the molecular weight of the encapsulated genome is known, its size can be used to narrow down the range of possible charges of the capsid by comparing the difference in weight between "filled" and "empty" VLPs. As a result, a total of 161 positive charges, for both "empty" and "filled" VLPs, is the value that produces the lowest difference to the genome's molecular weight (i.e., 0.4%).

| CONCLUDING REMARKS
Native positive ion mass spectra of "empty" (blue profile) and "filled" (red profile) AAV8 VLPs. The "empty" VLPs preparation shows a dominant peak e assigned to monomeric "empty" capsids. The "filled" VLPs preparation contains the shared peak e and a second peak f assigned to "filled" capsids. The figure comprises (A) the entire m/z range and the magnification of the range for (B) peak e and (C) peak f