Upstream
An expression vector encoding the gene for S_dF_2P or HexaPro along with a DHFR selection marker was transfected into CHO-DG44 cells by electroporation using the MaxCyte STX® scalable transfection system (MaxCyte, Gaithersburg, MD) as previously described28. Transfected cells were cultivated in an Multitron shaker (Infors HT, Switzerland) set to 37 °C, 5% CO2, and 80% relative humidity with a shaking speed of 130 rpm (orbital throw of 1 inch) in CDM4CHO medium with 6 mM L-glutamine. Forty-eight hours after transfection, methotrexate (MTX) was added to the culture to a final concentration of 100 nM. Viable cell density and viability for the culture was assessed every three to four days using the Cedex HiRes (Roche CustomBiotech, Indianapolis, IN). Once a week, the cells were centrifuged at 100 × g for 10 min and resuspended in fresh CDM4CHO medium with 6 mM L-glutamine and 100 nM MTX. When the viability of the pools recovered to ≥ 80%, the medium was replaced with ActiCHO P medium containing 6 mM L-glutamine and 100 nM MTX.
Clarification and concentration/buffer exchange
For harvest volumes less than 5 L, the harvest material was clarified of whole cells and cell debris by centrifugation at 3000 rpm for 30 min, followed by 0.8/0.2 µm sterile filtration (Sartorius Stedim, Germany). Alternatively, for larger volumes, the harvest was subjected to a depth filtration train consisting of Clarisolve 20MS followed by Millistak + F0HC filters (MilliporeSigma, Burlington, MA) with a subsequent 0.8/0.2 µm sterile filter. The depth filters were arranged in series and equilibrated with 1X PBS. The cell culture harvest was pumped through the filters at a 60 LMH feed flux based on the F0HC filter area and chased with 1X PBS. Clarified harvest was stored at 2–8 °C for further development activities.
UF/DF1
The clarified harvest was buffer exchanged using 100 kDa or 300 kDa Millipore Pellicon flat sheet membranes (MilliporeSigma, Burlington, MA) with a five-fold ultrafiltration and a fivefold diafiltration into various buffers as needed for capture chromatography. The feed flux was set to 330 LMH with a trans-membrane pressure of 10 psi. The 300 kDa flat sheet method was scaled up to a 1 m2 filter, with loading densities constant at around 10 L/m2.
Capture resin screen
Thirty-two anion exchange resins (Figure S1) were screened in duplicate at two pH conditions (pH 7.0 and pH 8.0) with a step gradient of NaCl elution conditions, in 50 mM NaCl increments ranging from 100 to 500 mM NaCl, followed by a 1 M NaCl strip. The resin screen was performed using the TECAN Evo system (TECAN group, Männedorf, Switzerland) in conjunction with robocolumns containing 0.1 mL of each resin (Repligen, Waltham, MA). Each column was loaded with concentrated, buffer exchanged harvest in 25 mM phosphate, 25 mM HEPES, 50 mM NaCl at either pH 7.0 or pH 8.0 to 222 mg/mL-r as measured by OD280 at a 2-min residence time. The loading density was set to 222 mg/mL-r to ensure enough product would be loaded for analysis, based on an expected product titer ~ 20 mg/L. The elution fractions were collected in UV-transparent 96-well microplates (Corning, NY) and transferred to the in-line plate reader. The total protein content of each fraction was measured by pathlength-corrected A280 and the S_dF_2P content was measured by binding to a monoclonal antibody targeting the N-terminal domain on the Octet binding platform.
Capture resin selection and optimization
Based on promising S_dF_2P binding and elution pattern data from the resin screen, resins were selected for further screening and development. Each resin was tested at pH 7.0 and pH 8.0 on an AKTA Avant (Cytiva, Picastaway, NJ), mimicking the process parameters from the resin screen (i.e., 2-min residence time) with the loading density decreased to 50 g/L-r. All elution fractions were analyzed by SDS-PAGE. Following the initial screen, ToyoPearl QAE-550C (Tosoh Biosciences, King of Prussia, PA), POROS 50 D (ThermoFisher, Waltham, MA), and GigacapQ 650 M (Tosoh Biosciences, King of Prussia, PA) were selected and further tested at pH 6.5 in an MES buffer system and pH 7.0 in a Sodium Phosphate buffer system to assess the impact of lower pH and buffer system on recovery (as measured by Octet titer) and purity (measured by HP-SEC) while including a head-to-head comparison to previous experiments performed in Sodium Phosphate pH 7.0. Subsequently, POROS 50 D was tested at pH 6.5 in both buffer systems listed above to investigate the impact each factor individually (i.e., buffer system and pH) (Figure S3).
Polish resin screen: CEX, HIC, and MM
Aliquots of POROS 50 D eluate were dialyzed using dialysis cassettes (ThermoFisher, Waltham, MA) into 50 mM Sodium Citrate, 50 mM NaCl pH 4.0 and pH 5.0. Thirty-one cation exchange resins (Figure S2), in duplicate, were loaded to 10 mg/mL-r by A280 measurement (1 OD = 1 mg/mL) for each pH condition with the same elution schema as the capture step resin screen. Due to low pH interference with the Octet titer assay, GXII was used to determine the purity of each fraction in addition to measuring total protein by pathlength-corrected A280. Fourteen hydrophobic interaction (HIC) and two mixed mode (MM) chromatography resins were evaluated at the robocolumn scale but did not yield promising separation based on SDS-PAGE (data not shown).
Polish step selection and optimization: CEX
Two CEX resins were selected for AKTA-scale confirmation runs: Toyopearl SP-650 M (Tosoh Biosciences, King of Prussia, PA) in flow through mode, and Nuvia HR-S (BioRad, Hercules, CA) in bind and elute mode based on high purity by GXII. AKTA-scale confirmation runs were analyzed via SDS-PAGE, purity by HP-SEC, and NS-EM. Toyopearl SP-650 M optimization experiments included analyzing recovery by Octet titer at pH 3.5 vs. 4.0. Nuvia HR-S elution optimization experiments from 180–250 mM NaCl pH 4.0 were conducted to maximize recovery and HCP clearance (Figure S5). Elution fractions were analyzed by SDS-PAGE for purity, Octet titer for recovery, and HCP ELISA for HCP clearance.
20 nm Filtration/Low pH treatment
Low pH treatment was evaluated for feasibility by holding process intermediate material at pH 3.5 for 30, 60, 90, and 120 min, followed by neutralization with 1 M Tris Base. The neutralized products were measured for binding on the Octet platform to assess any potential changes in antigenicity.
Nanofiltration performance was assessed by measuring flux and mass throughput on small scale, decoupled trains consisting of a Viresolve Shield or Shield H Prefilter and a 20-nm Viresolve Pro Filter (MilliporeSigma, Burlington, MA), run in constant pressure mode at 30 psi. A developmental lot of cation exchange-polished material was selected as the feed stream for 20 nm filtration, and either loaded directly at pH 4.0 or conditioned to pH 7.0 using 1 M Tris HCl, pH 8.0 prior to loading.
UF/DF II
Flat sheet membranes with 300 kDa and 100 kDa pore sizes (MilliporeSigma, Burlington, MA) were screened for final concentration, buffer exchange, and host cell protein (HCP) removal. Cation exchange chromatography elutions with high HCP (~ 250,000 ppm) were pooled from selection and optimization experiments and loaded onto 50 cm2 membranes. At a flux of 300 LMH and TMP of 7.3 psi, the material was concentrated two-fold and then diafiltered against 20 diavolumes of 10 mM Histidine, 150 mM NaCl, 5% Sucrose, pH 6.5. Samples of the retentate and permeate were taken at the end of ultrafiltration and at every five diavolumes. The filter was chased with one system volume of diafiltration buffer and the chase was pooled with the retentate for an additional sample point. Each fraction was analyzed by Octet titer for S_dF_2P-specific recovery, purity by HP-SEC, and residual HCP.
Proof of concept
The developed process described in Fig. 1 was applied to a CHO-DG44 stable pool harvest expressing the HexaPro stabilized spike construct. Cell culture harvest (6.5 L) was flowed through a depth filtration train consisting of one 0.11 m2 Clarisolve 20MS and one 0.11 m2 Millistak + F0HC filter (MilliporeSigma, Burlington, MA) at 60 LMH, followed by 0.8/0.2 um sterile filtration (Sartorius Stedim, Germany). The clarified harvest was concentrated five-fold and then buffer exchanged against five diavolumes of 20 mM MES, 25 mM NaCl pH 6.5 using a 0.5 m2 300 kDa flat sheet filter (MilliporeSigma, Burlington, MA). The buffer exchanged material was loaded onto POROS 50 D (ThermoFisher, Waltham, MA) at 20–25 mg/mL-r, and the elution, collected from 50–80 mAU, was subjected to a 60-min hold at pH 3.5. After low pH treatment, the material was diluted with 50 mM Sodium Citrate pH 4.0 to condition to the approximate equilibration conditions of the polish steps. The conditioned material was loaded onto Toyopearl SP-650 M (Tosoh Bioscience, King of Prussia, PA) at < 15 mg/mL-r and chased with 5 CV of equilibration buffer. The flow-through and chase were pooled and loaded onto Nuvia HR-S (BioRad, Hercules, CA) at ~ 30 mg/mL-r, then eluted at 50 mM Sodium Citrate, 180 mM NaCl pH 4.0. Fractions of the Nuvia HR-S product were used for viral filtration studies, then the Nuvia HR-S product was pooled with the small scale aliquots of the 20 nm filtrate to forward process. The product pool was concentrated two-fold and buffer exchanged against 20 diavolumes of 10 mM Histidine, 150 mM NaCl, 5% Sucrose pH 6.5 on a 100 kDa flat sheet filter (MilliporeSigma, Burlington, MA).
Analytical methods
Octet
The binding assay was performed by biolayer interferometry (BLI) using Octet Red384 Instrument (FortéBio, Menlo Park, CA). For quantitative binding analysis of S_dF_2P (referred to as Octet titer), all reagents, calibrator, and samples are prepared by dilution in 1X kinetics buffer (KB) (FortéBio, Menlo Park, CA). The monoclonal antibody S652-118 (referred to as mAb118) (Vaccine Production Program, VRC, NIAID, NIH, Gaithersburg, MD) was immobilized onto a protein G biosensor (FortéBio, Menlo Park, CA), and followed by binding of S_dF_2P sample in a range of dilutions. The binding response is compared to a calibration curve of S_dF_2P of known concentrations. Serial dilutions of calibrator were performed at top of curve of 100 µg/mL scheme down to 0.78 µg/mL. Positive controls were in the form of a spike sample prepared in 1X KB at 40 µg/mL and diluted to 2X, 4X, and 8X, also in 1X KB. Each sample was diluted into the linearity range of the assay. The mAb118 stock was diluted to a concentration of 10 µg/mL. Four steps of assay include: (1) regeneration: 5 s × 3 cycles with 500 mM phosphoric acid and 1X KB; (2) loading: 120 s with mAb118; (3) baseline: 30 s with 1X KB; (4) associate: 120 s with sample. The %CV for the calibration standard curve replicates was ≤ 20% for all points above 3.1 µg/mL. 4PL curve fit R2 was > 0.98. The recovery of spike was in a range of 80–120%.
For full curve binding analysis of S_dF_2P with mAb118, all reagents, calibrator, and samples are prepared by dilution in 1X PBS (Lonza). Serial dilutions of S_dF_2P sample and calibrator were performed at top of curve of 100 µg/mL scheme down to 0.78 µg/mL and a zero. The assay consisted of five steps: (1) regeneration: 5 s × 3 cycles with 500 mM phosphoric acid and 1X KB; (2) baseline: 60 s with 1X PBS; (3) loading 180 s with mAb 118; (4) baseline: 60 s with 1X PBS; (5) association: 180 s with sample diluted serially in 1X PBS. The resulting data were fit to a 1:1 binding model. The %CV of response values for all sample and calibrator replicates was ≤ 20% for all points above 0.78 µg/mL.
GXII
Four microliters of sample were mixed with 16 µL of reducing buffer (a mixture of SDS, LDS, and DTT) and denatured at 90 °C for 5 min. Samples were allowed to cool to room temperature prior to the addition of 4 µL of dye. The samples were covered in foil, vortexed, and left to incubate in the dark for 1 h. The dye reaction was quenched with 210 µL of stop solution and 105 µL of the labeled protein was loaded into a GXII plate. The plate was loaded into the instrument and run using the HT Pico Protein Express 200 Programming (PerkinElmer, Waltham, MA).
High performance size-exclusion chromatography (HP-SEC)
HP-SEC is a method where molecules are separated by size, specifically their hydrodynamic radius, and in this case detected through fluorescence (FLR). The S_dF_2P product purity is assessed using the SRT 500A SEC column (Sepax, Newark, DE) by FLR detection at excitation wavelength 280 nm and emission wavelength at 348 nm. The S_dF_2P purity is determined by the percent area of the main peak, while the S_dF_2P aggregation is determined by the percent area of the high molecular weight species and smaller proteins are eluted as the lower molecular weight species. The approximate molecular weight can also be determined with HP-SEC by comparing it with the gel filtration standard (GFS). The retention times of each peak that correspond to various molecular weights of the GFS can then be compared with the S_dF_2P main peak with an overlay of the chromatograms, which determined that the S_dF_2P main peak (S_dF_2P glycoprotein) is greater than 670 kDa.
Host cell protein (HCP)
The CHO HCP assay is a two-site immunoenzymetric kit assay obtained from Cygnus Technologies, (Oakton, VA). All samples were processed as per manufacturer’s instructions.
Host cell DNA (HCD)
The residual CHO HCD assay kit (ThermoFisher, Waltham, MA) employs both a DNA extraction procedure and a QPCR quantitation procedure. CHO DNA extraction is performed utilizing the semi-automated MagMAX extraction method with the PrepSEQ Residual DNA Sample Preparation system. QPCR quantitation of residual DNA is performed utilizing the resDNASEQ Human Residual DNA Quantitation System. The primers and Taqman probe of the assay are highly specific, detecting only a hamster-specific region of a multicopy genetic element, with no cross-reactivity with unrelated DNA. The broad linear range of the QPCR assay allows for the testing of samples with variable levels of Human DNA in the sample assay, with a lower limit of quantitation (LLOQ) of 6 pg/mL.
A280
Unless otherwise stated, concentration was determined by measuring absorbance at 260 nm, 280 nm, 340 nm, 900 nm, and 975 nm and using the pathlength correction displayed in Eq. 1 for high-throughput experiments. For lab-scale optimization experiments, absorbance at 280 nm was coupled with the empirically determined extinction coefficient of 1.00 for concentration measurement.
$${text{A }} = , 0.{173}*left( {{text{A28}}0 , {-}{text{ A34}}0} right)/left( {{text{A975}} – {text{A9}}00} right)$$
(1)
Negative-stain electron microscopy
For protein preparations at neutral pH, the sample was diluted to 0.02 mg/ml with 10 mM HEPES, pH 7.4, supplemented with 150 mM NaCl. For protein preparations at acidic pH, 10 mM sodium-acetate supplemented with 150 mM NaCl was used instead, with the pH of the dilution buffer matching that of the sample. A 4.7-µL drop of the diluted sample was placed on a glow-discharged carbon-coated copper grid (CF200-Cu, Electron Microscopy Sciences, Hatfield, PA) for 15 s. The drop was then removed with filter paper, and the grid was washed by applying consecutively three 4.7-µL drops of the buffer used for dilution in the same manner. Negative staining of protein molecules adsorbed to the carbon layer was performed by applying consecutively three 4.7-µL drops of 0.75% uranyl formate in the same manner, and the grid was air-dried. Datasets were collected using an FEI T20 transmission electron microscope (FEI Company, Hillsboro, Oregon) operated at 200 kV and equipped with an Eagle CCD camera. The nominal magnification was 100,000x, corresponding to a pixel size of 2.2 Å, and the defocus was set at − 1.0 µm. Data was collected automatically using SerialEM35. Particles were picked from the micrographs automatically using in-house written software (YT, unpublished). 2D classification was performed using Relion 1.436.
SDS-PAGE
SDS-PAGE were performed using ThermoFisher Scientific (Waltham, MA) materials, including Bolt™ 4–12% Bis–Tris Plus gels and a running buffer of 1X MOPS. All samples were subjected to NuPage reducing agent and diluted in Bolt 4X LDS sample buffer prior to loading. BenchMark Protein ladder was used as a molecular weight reference for each gel. Each gel was subjected to 150 V for 55 min, rinsed with DI water, and then stained with GelCode Blue Safe protein stain.
DLS and DSC
DLS and DSC methods were performed as previously reported37.
