J ISSN 1615-9306 · JSSCCJ 42 (16) 2587– 2742 (2019) · Vol. 42 · No. 16 · August 2019 · D 10609 JOURNAL OF S SEPARATION S SCIENCE 16 19 Methods www.jss-journal.com Chromatography · Electroseparation Applications Biomedicine · Foods · Environment Received: 6 May 2019 Revised: 4 June 2019 Accepted: 4 June 2019 DOI: 10.1002/jssc.201900441 RESEARCH ARTICLE At-line multi-angle light scattering detector for faster process development in enveloped virus-like particle purification Patricia Pereira Aguilar1 ∗ Irene González-Domínguez2 ∗ Tobias Amadeus Schneider3 Francesc Gòdia2 Laura Cervera2 Alois Jungbauer1,3 1 Department of Biotechnology, University At-line static light scattering and fluorescence monitoring allows direct in-process of Natural Resources and Life Sciences, Vienna, Austria tracking of fluorescent virus-like particles. We have demonstrated this by coupling 2 Department d’Enginyeria Química Biològica at-line multi-angle light scattering and fluorescence detectors to the downstream pro- i Ambiental, Universitat Autònoma de cessing of enveloped virus-like particles. Since light scattering intensity is directly Barcelona, Bellaterra, Barcelona, Spain 3 Austrian proportional to particle concentration, our strategy allowed a swift identification Centre of Industrial Biotechnology, Vienna, Austria of product containing fractions and rapid process development. Virus-like particles containing the Human Immunodeficiency Virus-1 Gag protein fused to the Green Correspondence Fluorescence protein were produced in Human Embryonic Kidney 293 cells by tran- Professor Dipl.-Ing. Dr.nat.techn. Alois Jungbauer, Department of Biotechnology, sient transfection. A single-column anion-exchange chromatography method was University of Natural Resources and Life used for direct capture and purification. The majority of host-cell protein impurities Sciences Vienna, Muthgasse 18, 1190 Vienna, passed through the column without binding. Virus-like particles bound to the column Austria. Email: alois.jungbauer@boku.ac.at were eluted by linear or step salt gradients. Particles recovered in the step gradient purification were characterized by nanoparticle tracking analysis, size exclusion chro- ∗ Both authors contributed equally to this paper. matography coupled to multi-angle light scattering and fluorescence detectors and transmission electron microscopy. A total recovery of 66% for the fluorescent par- ticles was obtained with a 50% yield in the main product peak. Virus-like parti- cles were concentrated 17-fold to final a concentration of 4.45 × 1010 particles/mL. Simple buffers and operation make this process suitable for large scale purposes. KEYWORDS enveloped bionanoparticles, fluorescent virus-like particles, monoliths, nanoparticle tracking analysis 1 I N T RO D U C T I O N Bradford assay and SDS-PAGE are used to gain insights into the total protein content. In addition, assays such as Western The majority of the analytical techniques used in virus-like blot and ELISA allow the specific detection and quantifica- particle (VLP) downstream processing (DSP) were imported tion of VLP specific proteins. However, none of these meth- from protein DSP [1]. Typically, colorimetric methods such as ods confirms the presence of correctly assembled VLPs, but Article Related Abbreviations: CV, column volume; DSP, downstream processing; FL, fluorescence; Gag, group specific antigen; GFP, green fluorescence protein; HEK 293, Human Embryonic Kidney 293 cells; HIV-1, Human Immunodeficiency Virus 1; LS, light scattering; MALS, multi-angle light scattering; NTA, nanoparticle tracking analysis; PEI, polyethilenimine; SEC-MALS-FL, SEC coupled to MALS and fluorescence detectors; VLP, virus-like particle. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2019 The Authors. Journal of Separation Science published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. 2640 www.jss-journal.com J Sep Sci 2019;42:2640–2649. PEREIRA AGUILAR ET AL. 2641 only the presence of specific viral proteins, which, when in Different DSP strategies have been developed for the purifi- their free state, represent one of the product related impurities. cation of bionanoparticles [5,11,20–22]. Monolithic columns Particle morphology is usually confirmed by high-resolution have been often used for the direct capture and purification of microscopy techniques such as TEM and multi-frequency enveloped virus and VLPs from cell culture supernatant [23– atomic force microscopy [2]. Particle number and size dis- 27]. Due to its convective pore structure, high binding capaci- tributions are frequently measured by nanoparticle tracking ties can be achieved in monolithic columns while maintaining analysis (NTA) and dynamic or static light scattering [3– high flow rates. This results in higher productivities when 5]. However, these methods are non-specific regarding the compared with traditional VLP purification methods such as particle’s composition and consider all structures with the density gradient centrifugation [28]. As a model purification same hydrodynamic or geometric radius as equals. As a result, strategy, we used anion exchange monolithic columns for a combination of biochemical, biophysical, and biological the direct capture and purification of HIV-1 Gag-GFP VLPs analytical methods is required to ensure an accurate detec- directly from cell culture supernatant. We show how at-line tion and quantification of VLPs. This results in very laborious MALS and fluorescence detectors simplified DSP process and time-consuming process analytics and hinders DSP pro- development and optimization. Furthermore, size exclusion cess development, which is strongly dependent on the ability chromatography coupled to multi-angle light scattering and to detect, quantify and characterize the product of interest as fluorescence (SEC-MALS-FL) is an effective method for well as on the capacity to discriminate the product from its particle quantification and characterization. related impurities. Multi-angle light scattering (MALS) is one of the most used techniques for quantification and characterization of dif- 2 M AT E R I A L S A N D M E T H O D S ferent nanoparticles [6]. Usually, monodisperse samples are required to allow for structural characterization based on light 2.1 Chemicals scattering. Typically, SEC or asymmetric flow field-flow frac- The chemicals used for all experiments were acquired from tionation are used to first separate the samples before the mea- Merck (Darmstadt, Germany) or Sigma Aldrich (St. Louis, surement [4,7]. Nevertheless, the intensity of scattered light MO, USA). by particles in solution is directly proportional to the parti- cle concentration [8,9]. Thus, an at-line MALS detector can 2.2 Production of virus-like particles be used for fast detection and semi-quantification of particles without the requirement of monodisperse samples. This rapid 2.2.1 Cell line, media, and culture conditions method could be used as an in-process control to speed up A serum-free suspension-adapted Human Embryonic Kid- process development and optimization. ney 293 cells (HEK 293) cell line (HEK293SF-3F6, National Outbreaks of Flu, Ebola, and Zika in the last decade Research Council, Montreal, Canada) kindly provided by Dr. reinforced the need for faster process development and Amine Kamen from McGill University (McGill, Montreal, flexible manufacturing platforms which enable the produc- Canada) was used. Cells were cultured in Freestyle 293® tion of millions of vaccine doses in a short time [10,11]. medium supplemented with 0.1% Pluronic® (both Invitrogen, Among different candidates, retroviral-based VLPs are Carlsbad, CA, USA), 1.6 mg/L of r-transferrin (Merck Mil- promising towards the development of vaccines and drug lipore, Kankakee, IL, USA), 19.8 mg/L of r-insulin (Novo delivery systems [12,13]. Enveloped virus-like particles Nordisk Pharmatek, Køge, Denmark), and 0.9X of an in- (eVLPs) are produced by recombinantly expressing one or house developed lipid mixture to maximize cell growth [8]. more viral proteins [10–12]. As a model, we used eVLPs Cells were routinely maintained in 20 mL of culture medium. produced in Human Embryonic Kidney (HEK) 293 cells Flasks were shaken at 130 rpm using an orbital shaker (Stu- by recombinantly expressing the Human Immunodeficiency art, Stone, UK) placed in an incubator maintained at 37˚C in Virus-1 (HIV-1) group specific antigen (Gag) protein which a humidified atmosphere of 5% CO2 in air. was fused to Green Fluorescence protein (GFP) [2,14–16]. Similar to the native HIV-1 production process, upon 2.2.2 Plasmids recombinant expression, Gag polyprotein self-assembles The pGag-eGFP plasmid used in this work codes for a underneath the cell membrane and VLPs bud to the extracel- Rev-independent HIV-1 Gag protein fused in frame to lular space as spherical particles [17]. The resulting particles the enhanced GFP [27]. The plasmid from the NIH AIDS are enveloped by a host-cell derived lipid bilayer and have Reagent Program (Cat 11468) was constructed by cloning the a diameter of 100–200 nm [4,18,19]. The integration of Gag sequence from pCMV55M1-10 [28] into the pEGFP-N1 GFP permits the use of fluorescence-based techniques as an plasmid (Clontech, Takara Bio, Mountain View, CA, USA). orthogonal method for detection and quantification of HIV-1 The plasmids were prepared and purified as previously Gag-GFP VLPs. described with Endofree Plasmid Mega kit (Qiagen, Hilden, 2642 PEREIRA AGUILAR ET AL. Germany) [29]. Snap Gene Viewer was used to analyse the collected in 96 deep-well plates and pooled according to the plasmid features (GSL Biotech, Chicago, IL, USA). chromatograms. 2.2.3 DNA/polyethilenimine complex 2.4 Nanoparticle tracking analysis formation and transient transfection of HEK 293 cells Particle concentration and particle size distribution were HEK 293 suspension cells were transiently transfected determined by NTA using a NanoSight NS300 (Malvern using 25 kDa linear polyethilenimine (PEI) (PolySciences, Instruments, Worcestershire, UK) equipped with a blue laser Warrington, FO, USA). Transfections were performed using module (488 nm), a neutral density filter and a 500 nm a final DNA concentration of 1 μg/mL. PEI/DNA com- fluorescence filter. To obtain a particle concentration of 20 to plexes were formed by adding PEI to plasmid DNA (1:2 w/w 80 particles per video frame in the measuring chamber, sam- DNA:PEI ratio) diluted in fresh culture medium (10% of the ples were serially diluted using particle-free water or 0.1 μm total culture volume to be transfected) [8]. filtered 50 mM HEPES pH 7.2 buffer. Each sample was mea- Cells were cultured for 72 h post transfection to maximize sured with both scattering (LS) and fluorescence (FL) modes VLP yields [30]. Cell culture supernatants were primary har- in three different dilutions in triplicates. In total, nine videos vested by centrifugation at 4000 g for 30 min at 4˚C. Recov- of 60 s were recorded per sample. The camera level varied ered supernatants were stored at 4˚C before purification. between 14 and 16 and it was manually adjusted prior to each measurement. Recorded videos were analysed using the NanoSight NTA software version 3.2 (Malvern Instruments, 2.3 Chromatographic purification Worcestershire, UK). Detection thresholds between 3 and 2.3.1 Chromatographic equipment and 5 were used. mobile phases Chromatographic experiments were performed using an Äkta 2.5 Total protein and double stranded DNA pure 25 M2 with a sample pump S9 and fraction collector F9- quantification C (GE Healthcare, Uppsala, Sweden). During the purification Total protein concentration was determined by Bradford assay runs, pH, conductivity and UV absorbance at 280 and 260 nm using Coomassie blue G-250-based protein dye reagent (Bio- wavelengths were monitored. Unicorn software versions 5.10 Rad Laboratories, Hercules, CA, USA). The calibration curve or 6.4.1 (GE Healthcare, Uppsala, Sweden) were used for was obtained using BSA standards (Thermo Fisher Scien- method programming, system control, and data acquisition. tific, Waltham, MA, USA) diluted in TE-Buffer to a con- Mobile phase A and B consisted in 50 mM HEPES, pH 7.2 centration range of 50–200 μg/mL. Double stranded DNA and 50 mM HEPES, 2 M NaCl, pH 7.2 respectively. Sanitiza- (dsDNA) quantification was performed using the Quant- tion buffer consisted in 1 M NaOH. iTTM PicoGreen® dsDNA kit (Life Technologies, Waltham, MA, USA). Protein and dsDNA assays were performed 2.3.2 Preparative scale purification according to the respective instructions from the manufac- Clarified cell culture supernatant containing HIV-1 Gag-GFP turer in a 96-well plate format. Since HIV-1 Gag-GFP VLPs VLPs was 0.8 μm filtered (Millex AA syringe filter, Millipore emit at the same range as the Quant-iTTM PicoGreen® Bedford, MA, USA) and 100 mL were loaded into a 1 mL reagent, the native fluorescence was measured prior to the radial flow monolith (CIMmultusTM QA, BIA Separations, reagent addition and later subtracted to the fluorescence after Ajdovščina, Slovenia). Before loading, the column was equili- the reaction. brated with 50 mM HEPES, 100 mM NaCl, pH 7.2 (5% buffer B). After the loading phase, the column was washed with equi- libration buffer (5% buffer B) for 15 column volumes (CV). 2.6 Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blot analysis In the linear gradient purification, a salt linear gradient from 100 to 1000 mM NaCl (5 to 50% buffer B) in 50 CV was used. Precast NuPAGE Bis/Tris gels 4–12% (Invitrogen, Carlsbad, For the step gradient purification, three steps of 300, 520 and CA, USA) were used in a MES-SDS buffer system. The pro- 1000 mM NaCl (15, 26 and 50% buffer B) with 15 CV each tocol was adapted from manufacturer’s instructions. Briefly: were used for elution. In both purification strategies, the col- 40 μL of sample were mixed with 20 μL of 4x LDS buffer umn was regenerated with 100% buffer B in a 10 CV step. and 2 M DTT to a final concentration of 1% v/v. Each sam- After regeneration, the column was sanitized using 10 CV of ple incubated at 95◦ C for 20 min. SeeBlue® Plus2 Pre-stained 1 M NaOH. All preparative purification runs were performed Protein Standard (Invitrogen, Carlsbad, CA, USA) was used using a flow rate of 1 mL/min. The sample was loaded into as protein marker. Gels were run at 200 V, 400 mA. Coomassie the column using the sample pump. Fractions of 1 mL were Brilliant Blue G-250 based EZBlueTM Gel Staining Reagent PEREIRA AGUILAR ET AL. 2643 (Sigma Aldrich, St. Louis, MO, USA) was used for protein 2.9 TEM staining. After SDS-PAGE, proteins were blotted using Trans- HIV-1 Gag-GFP VLP samples were prepared by air-dried Blot® turbo system (Bio-Rad Laboratories, Hercules, CA, negative staining method. Briefly, 8 μL of sample was placed USA) with 0.2 μm nitrocellulose membranes and blocked with on discharged carbon-coated copper or holly carbon 200 mesh 3% BSA in PBS with 0.1% w/v Tween-20 overnight. Detec- grids and incubated at room temperature for 1 min. Excess tion of HIV-1 Gag-GFP protein was performed by incuba- sample was drained carefully off the grid with filter paper. tion with primary mouse monoclonal antibody against HIV- Samples were stained negatively with 8 μL of uranyl acetate 1 p24 (Icosagen AS, Tartumaa, Estonia), diluted 1:1000 in (2%) by incubation for 1 min at room temperature. Excess PBS-T containing 1% BSA for 2 h. Anti-mouse IgG conju- stain was drained off as before, and grids were dried. Micro- gated with alkaline phosphatase (Sigma Aldrich, St. Louis, graphs were taken with a JEM-400 transmission electron MO, USA), diluted 1:1000 in PBS-T with 1% w/v BSA was microscope (JEOL USA, Pleasanton, CA, USA) equipped used as secondary antibody. Premixed BCIP®/NBT solution with an ES1000W Erlangshen charge-coupled device camera (Sigma Aldrich, St. Louis, MO, USA) was used as substrate (Model No. 785; Gatan, Pleasanton, CA, USA). solution. 2.7 At-line multi-angle light scattering and 3 RESULTS AND DISCUSSION fluorescence At-line MALS and fluorescence measurements were per- In this work, we aimed to streamline the process development formed using an Ultimate 3000 system (Thermo Fisher, for enveloped virus-like particle purification by including at- Waltham, MA, USA). A sample of all collected fractions dur- line multi-angle light scattering and fluorescence detectors for ing the purification runs was directly injected into the detec- high-throughput particle detection and semi-quantification. tors bypassing the column. The HPLC system was equipped Since it has been shown that strong anion-exchange monoliths with a LPG-3400SD quaternary pump, WPS-3000TSL ana- allow the simultaneous capture and purification of enveloped lytical autosampler, DAD 3000 UV-detector and FLD 3100 bionanoparticles such as eVLPs and exosomes [22,23], we fluorescence detector (Thermo Fisher, Waltham, MA, USA). used a QA monolith to capture and purify HIV-1 Gag-GFP Additionally, the system was connected to a multi-angle light VLPs directly from HEK 293 cell culture supernatant in a scattering detector DAWN HELEOS 18-angle and a differen- single step. As a starting point, a salt linear gradient was tial refractive index detector Optilab rEX (both Wyatt, Santa used for VLP elution. Later, the data obtained in the at-line Barbara, CA, USA). Chromeleon 7 (Thermo Fisher Scien- MALS and fluorescence measurements were used to develop tific, Waltham, MA, USA) and Astra 5.3.4 Wyatt, (Santa Bar- a salt step elution strategy, providing a base for potential bara, CA, USA) software were used for method program- scale-up. ming, system control and data acquisition. GFP fluorescence was monitored with an excitation wavelength of 480 nm and emission of 505 nm. Analysis time was 3 min/sample. 3.1 Linear gradient purification Light scattering intensity was accessed by calculating the HIV-1 Gag-GFP VLPs were produced by transient transfec- peak area of the light scattering signal obtained with the tion in HEK 293 cells. For VLP capture and purification in a 90◦ angle. single step, 100 mL of clarified and filtered cell culture super- natant were loaded into a 1 mL QA monolith. Elution was achieved by a 50 CV salt linear gradient from 100 to 1000 mM 2.8 Size exclusion chromatography coupled to NaCl. Fractions of 1 mL were collected in 96 deep-well plates multi-angle light scattering and fluorescence and directly injected into the at-line MALS and fluorescence SEC-MALS-FL measurements were performed using the detectors. For each elution fraction, the total light scattering same HPLC system, detectors and software as described in intensity and the total fluorescence were calculated by inte- Section 2.7. grating the signals measured by the MALS and fluorescence A TSKgel G5000PWXL 300.0 mm × 7.8 mm i.d. in com- detectors, respectively. Data were plotted together with the bination with a TSKgel PWXL guard column 40.0 mm × purification run chromatogram (Figure 1). Since light scat- 6.0 mm i.d. (both Tosoh Bioscience, Stuttgart, Germany) were tering intensity is directly proportional to particle concentra- used for size exclusion chromatography. The method was tion [8] and the main structural element of the VLPs (Gag- previously described by Steppert et al. [20]. Data analysis GFP protein) is fluorescent [30], this method allows a fast was done in Astra 6.1.2 using the number density procedure detection and semi-quantification of HIV-1 Gag-GFP VLPs and the sphere model fit with a particle refractive index of and subsequently a fast identification of the fractions contain- 1.46 [29]. ing the product of interest is possible. Considering the total 2644 PEREIRA AGUILAR ET AL. FIGURE 1 Chromatogram of the linear gradient purification of HIV-1 Gag-GFP VLP using a QA monolith. The loading material was 100 mL of clarified and 0.8 μm filtered HEK 293 cell culture supernatant. Bars represent the area under the curve of the light scattering intensity (grey) and fluorescence (green) at-line measurements. FT: flow-through; W: wash; P1-P5: polled fractions for peaks 1–5 TABLE 1 Total protein and dsDNA mass balance of the purification of HIV-1 Gag-GFP VLPs using a linear gradient elution (Figure 1). S: supernatant; L: loading material; FT: flow-through; W: wash; P1-P5: peaks 1-5 Volume Total protein Total dsDNA Sample [mL] [μg/mL] protein % [ng/mL] dsDNA % S 100 310.9 – 1387.5 – L 100 313.1 100.0 1289.4 100.0 FT 100 134.9 43.1 37.5 2.9 W 15