Expression and Purification of Adeno-associated Virus Virus-like Particles in a Baculovirus System and AAVR Ectodomain Constructs in coli

[Abstract] Adeno-associated virus (AAV) is a promising gene therapy vector and the biophysical characterization of its interactions with host proteins is a critical foundation for engineering tissue targeting and immune escape. Presented here are protocols for the production of: (a) the outer protein shells (virus-like particles or VLPs) for serotype 2 (AAV-2) and (b) two fragments from the binding ectodomain of AAV’s cellular receptor, AAVR. His 6 PKD1-2 comprises the first two polycystic kidney disease (PKD) domains, the minimal required for efficient binding of AAV, expressed with an N-terminal histidine tag. MBP-PKD1-5 is a fusion of the maltose binding protein with all five of the PKD domains of the AAVR receptor. Presented are the expression and purification of milligram quantities, ample for in vitro analyses. For AAV-2, the protocol offers an alternative to the use of (infectious) wild-type virus or transducing vectors. One of the methods for producing transducing vector is in Sf9 cells, and the production of VLPs is based on this. For AAVR, the protocols enable biochemical and biophysical characterization of virus-binding. The minimal two-domain construct allows more saturated binding to symmetry-equivalent sites on the virus, while the larger construct might be better expected to reflect the native receptor.

When structural studies of AAV were first attempted in the 1990s, methods were not available to produce the milligram levels of AAV needed for crystallization. Structural studies became possible with the optimization of methods to produce wild-type (i.e., infectious) AAV-2 by transfection of cultured human cells with virus-encoding bacterial plasmid (Samulski et al., 1982;Laughlin et al., 1983) in the presence of the adenovirus helper needed for AAV replication (Xie et al., 2004). The elucidation of the type-species structure, AAV-2 (Xie et al., 2002), paved the way for studies of immunologically distinct serotypes. Structural studies of several serotypes (e.g., AAV-3, AAV-4, AAV-5, AAV-6) used sample prepared in analogous ways (Kaludov et al., 2003;Walters et al., 2004;Xie et al., 2008;Lerch et al., 2010). More recently, biophysical studies have taken advantage of other preparative methods.
At first, it was not practical to use recombinant AAV (rAAV) vectors produced in human cells by double or triple transfection (Grimm et al., 1998;Matsushita et al., 1998), because the quantities, while sufficient for transduction experiments, were insufficient for biophysics. The higher-yield production of rAAV-2 vector in insect cells using baculoviral vectors  inspired a different approach, that was used for the crystal structures of AAV8 and AAV9 (Nam et al., 2007, DiMattia et al., 2012. The same bacmids encoding capsid proteins are used, but bacmids encoding the AAV ITR vector genome and the replication protein are omitted. This generates empty capsid protein shells, or virus-like-particles (VLPs), instead of vectors (Lane et al., 2005). The same approach has been used for cryo-electron microscopy (EM), as exemplified by the engineered vector, AAV-DJ which has been solved at near atomic resolution (Lerch et al., 2012, Xie et al., 2017. This protocol provides details for purifying milligram quantities of type species AAV-2 VLPs based on the AAV-DJ strategy summarized earlier (Lerch et al., 2012). We combine AAV8-like insect cell expression of VLPs with high-capacity purification steps used for the biophysics of wild-type viruses (Xie et al., 2004;Lerch et al., 2012) that can be more stringent than in rAAV productions designed for in vivo use. Repeated CsCl density gradient ultracentrifugation and desalting via dialysis is employed in contrast to the more common use of iodixanol for rAAV vectors (Hermens et al., 1999) and VLPs (Drouin 3 www.bio-protocol.org/e3513 The shift from crystallography to cryo-EM means that AAV structure is now possible with the smaller quantities available by rAAV triple transfection production methods (Zhang et al., 2019). (In our hands, preparation of a single EM grid uses ~1 μg, but for the optimization of conditions one would like > 25 μg.) The AAV-2 structure has now been determined by crystallography for infectious wild-type (wt) particles (Xie et al., 2002), by cryo-EM for VLPs (Drouin et al., 2016) and cryo-EM for rAAV vectors (Zhang et al., 2019). For most of the capsid protein, the structures superimpose within the experimental error at 2.8 to 3.8 Å resolution but some differences have been noted near the surface spikes (Drouin et al., 2016) where crystal packing contacts could influence structure. The three sources of material (wt, rAAV & VLP) can all support studies of atomic structure, but the highest resolution structures are coming from VLPs. There could be other explanations, but one possible reason is the purity achievable with the higher yield VLP protocol presented here.
The capsids of natural AAV are comprised mostly of viral protein (VP) 3, but about 10% of VP3s in the 60-subunit capsid are replaced by variant VP1, and another ~10% by VP2 (Berns, 1996). The three proteins are mostly the same, but VP1 and 2 have N-terminal extensions derived from alternate start codons. VLP constructs are designed to express variants in the right proportions .
However, it is only the 60-fold symmetric parts of AAV that have been resolved in structures. So, even though the properties of wt, rAAV and VLP appear similar, one cannot verify that the structures of the N-terminal extensions are configured exactly the same. In spite of this caveat, VLPs are considered to be a good proxy for wild-type and rAAV vectors, and are often chosen, because of the ease of producing large quantities of pure sample without biosafety level II containment.
The receptor protein, AAVR, is an integral membrane glycoprotein, that consists of the following domains, starting from the extracellular N-terminus: a signal peptide (SP), a MANEC domain (motif at the N-terminus with eight cysteines), five immunoglobulin-like polycystic kidney disease (PKD) domains (Ibraghimov-Beskrovnaya et al., 2000), a transmembrane region and a small cytoplasmic domain. A soluble fusion construct consisting of maltose binding protein (MBP), N-terminally fused to PKD1-5 (Pillay et al., 2016), and lacking SP, MANEC, TM and the C-terminal tail was pivotal in implicating the PKD domains in AAV-binding via inhibition of rAAV cell transduction. This soluble construct, herein denoted MBP-PKD1-5, was used to measure competitive inhibition of AAV cell-transduction by solubilized receptor ectodomain, to measure AAV-binding strength using ELISA and surface plasmon resonance, and also for low resolution cryo-Electron Tomographic imaging of an AAV-2/AAVR complex (Pillay et al., 2016, Meyer et al., 2019. The expression and purification of the MBP-PKD1-5 AAVR construct is described in this protocol. Over-expression of "mini-AAVR" on the cell surface supports AAV transduction in an AAVR knock-out chamber. The bore has a capacity of 50 μl (left) or 60 μl (right). Between the bottom screwcap and the cylinder is sandwiched a dialysis membrane, and at the top end is a microscope cover slip, providing a window. After assembling the bottom part, sample is loaded from the top, before the cover slip is added and the top sealed. The button is hung membrane-side down from the side of a beaker in exchange buffer. Air that is trapped in the pocket under the bottom cap must be removed to ensure contact between membrane and buffer.  c. Incubate at 28 °C while shaking at 135 rpm, protected from light (as suggested by the manufacturer).
Note: A culture from frozen stock can take a couple of weeks to grow to a suitable density.

Amplification of baculovirus stock (generation of P2 stock)
This continues to follow the Bac-to-Bac TM , modified as follows: Quantifying the P2 virus titer by qPCR.

Note: This tends to take about 3-4 days post infection.
6. Sf9 infection for VLP production a. Scale up a Sf9 culture to 600-800 ml and a density of 2 x 10 6 -4 x 10 6 cells/ml, as described in Step A1, above.
b. Infect the culture with the P2 baculovirus stock. Add P2 to a MOI of between 5 and 10.
c. Incubate the culture as before and check cell density and viability daily. When viability starts to decrease (at 40-60 h), check culture more frequently and harvest when the viability is between 80-60% (typically between 55-80 h). performed with higher stringency, further repeating, or accepting only a narrower fraction near the peak of the band (Xie et al., 2004)]. The yield for a VLP preparation is typically 2-4 mg/ml in 100 µl.
Lastly, since the sample contains a high concentration of CsCl which would interfere with electron microscopy visualization, the sample is dialyzed using a custom dialysis device (dialysis button) designed for 40-60 µl volumes. Due to the relative stability of the VLP sample in 3.3 M CsCl and the instability in low-salt solution, the last dialysis step is performed on small volumes of purified sample that will be used within 1-2 days. (AAV-2 is more prone to aggregation at low salt than other serotypes, many of which are stable at 4 °C for up to 4 weeks.) e. Gently transport the tubes to your working area to preserve gradient separation. Wipe the outside of each tube with ethanol and visualize the AAV band using a bright halogen lamp (see Figure 2). As a guide, mark the side of the tube about 2 mm above and below the AAV-2 band (e.g., with a Sharpie pen). 11 www.bio-protocol.org/e3513  f. Use a standard disposable transfer pipette to remove the liquid down to the mark 2 mm above the AAV-2 band and discard this into a 10% bleach solution.
g. Use a fine-tipped disposable transfer pipette to gently collect the sample containing AAV-2 (take 2 mm above to 2 mm below the band), and transfer it to a new ultracentrifuge tube.
Collect the AAV-2 band from all six tubes similarly and pool into a single ultracentrifuge tube.
Check and record the refractive index (RI).
h. Balance and centrifuge the tube containing the pooled AAV-2 samples as before (Steps C2c-C2e, above) (second spin).
i. Repeat the AAV-2 band collection and ultracentrifugation step one more time (third spin). b. Set the system to "Load" and flush the loop with 10 ml of PBS using a Luer lock syringe.
c. Set the system to "Inject" and remove the syringe. d. Load a 1 ml syringe with your VLP sample and connect it to the loop. e. Set the system back to "Load" and inject your sample into the loop, being careful not to inject air.
f. Set the flow to 1 ml/min and introduce your sample to the column by setting the system to "Inject" and flushing the loop with 3 ml of 100% buffer A. g. Set back to "Load" and continue to run 100% buffer A at 1 ml/min for 5 min.
h. Run a linear gradient from 0-85% buffer B over 25 min, collecting 0.5 ml fractions.
i. Examine the UV chromatogram and set aside the eluate fractions containing the AAV-2 peak for evaluation by SDS-PAGE.
j. Run a sample from each AAV-2 peak-containing fraction on an SDS-PAGE gel, following standard protocols, to verify the presence of VLP.
k. Pool the AAV-2-containing fractions and run a final CsCl gradient ultracentrifugation as before (see Steps C2c-C2e above). This gradient serves to concentrate the VLPs.
l. Store the collected sample in a low protein-binding microcentrifuge tube. m. Clean the column by running 10 column volumes of elution buffer B, ultra-pure DI water, and 20% ethanol, sequentially. Store the column in 20% ethanol.

D. Desalting by dialysis
Our protocol continues to use a custom mini-dialysis apparatus that we designed for low volume and high recovery before somewhat similar devices became available commercially. It is likely that the Slide-A-Lyzer MINI dialysis device (see Materials) could be substituted, following the manufacturer's directions, rejoining the protocol at Step D3i below, "Measure the absorbance…".We continue the protocol based on the custom device: 1. Prepare the dialysis membrane, according to the manufacturer's instructions, then: Using a 12 mm circular glass cover side as a template, cut small circles out of dialysis membrane; store at 4 °C in HM buffer plus 0.02% sodium azide until use.

Prepare the dialysis button
Assemble the custom dialysis button according to Figure 1 3. Inoculate a single transformed BL21(DE3) colony from a freshly streaked LB-ampicillin agar plate into a volume of LB-ampicillin (100 µg/ml) media that is 5% of the final total desired culture volume (e.g., 50 ml overnight culture into 1,000 ml final volume). Grow the starter culture overnight at 37 °C with shaking at 210 rpm.  ii. Load cell slurry into a French press cell and pass through two complete rounds of lysis at 1,000 psi using a slow flow rate to collect the lysate into a 50 ml centrifuge tube.
While spinning, prepare the AKTA system. 2) Filter through a 0.45 µm filter using a 10 ml Luer lock syringe. It may take multiple filters if membranes become fouled.
3) With the valve setting set to "Load", inject 10 ml of lysate into the superloop. 4) Set flow rate to 0.5 ml/min, fraction size to 10 ml, and the valve to "Inject". 5) Flow through the superloop into the column until the superloop is almost empty. 6) At this point, pause the FPLC, switch the valve to "Load", inject another 10 ml into superloop, switch back to "Inject", and resume flow. Repeat this process until entire lysate volume is loaded onto the column (with a high capacity superloop, the entire sample could be loaded in a single step).
iii. Elution of MBP-PKD1-5 1) After loading, wash the column by switching the valve to "Load" and increasing the flow of running buffer to 1 ml/min.
2) Once baseline is stabilized-at least 10 column volumes-reduce the flow to 0.5 ml/min and the fraction size to 2 ml. 2) Concentrate the pooled sample to ~1 ml using a 30 kDa Amicon Ultra centrifugal filter unit. 16 www.bio-protocol.org/e3513     ii. Load the cell slurry into a French press cell and pass through 2 complete rounds of lysis at 1,000 psi using a flow rate slow enough to avoid foaming, collecting the lysate into a 50 ml centrifuge tube.
While spinning, prepare the AKTA system. iii. Elution of His6PKD1-2 1) Wash the column by switching the injection valve to "Load" and increasing the flow to 1 ml/min.
2) Once the baseline stabilizes-at least 10 column volumes-reduce the flow rate to 0.5 ml/min and the fraction size to 2 ml.
iv. Processing IMAC column fractions 19 www.bio-protocol.org/e3513 2) Concentrate pooled sample to ~1 ml using a 3 kDa MWCO Amicon Ultra centrifugal filter unit. This SEC step allows dimers to be selected over higher order oligomers shortly before use, and must be repeated, if this separation is desired, after the sample has been stored at 4 °C for an extended time.
i. Washing and equilibrating the Superdex 75 Wash and equilibrate the Superdex 75 column with HN buffer exactly as described for the Superdex 200 (Step F1c, above).
ii. Loading sample onto the Superdex 75 1) After equilibration, pause the AKTA and, as with the Superdex 200, inject a sample volume that is no more than 1% of the column volume (set valve to "Load").
2) Set flow rate to 1 ml/min and resume flow (valve setting = Inject).