Received: 24 November 2022 | Revised: 5 April 2023 | Accepted: 10 April 2023 DOI: 10.1002/bit.28419 ARTICLE Downstream process design for Gag HIV‐1 based virus‐like particles Elianet Lorenzo1 | Laia Miranda1,2 | Francesc Gòdia1 | Laura Cervera1 1 Departament d'Enginyeria Química Biològica i Ambiental, Universitat Autònoma de Abstract Barcelona, Bellaterra, Spain Virus‐like particles‐based vaccines have been gaining interest in recent years. 2 University College London, London, UK The manufacturing of these particles includes their production by cell culture Correspondence followed by their purification to meet the requirements of its final use. The presence Elianet Lorenzo, Departament d'Enginyeria of host cell extracellular vesicles represents a challenge for better virus‐like particles Química Biològica i Ambiental, Universitat Autònoma de Barcelona, Bellaterra, Spain. purification, because both share similar characteristics which hinders their Email: elianet.lorenzo@uab.cat separation. The present study aims to compare some of the most used downstream processing technologies for capture and purification of virus‐like particles. Four steps of the purification process were studied, including a clarification step by depth filtration and filtration, an intermediate step by tangential flow filtration or multimodal chromatography, a capture step by ion exchange, heparin affinity and hydrophobic interaction chromatography and finally, a polishing step by size exclusion chromatography. In each step, the yields were evaluated by percentage of recovery of the particles of interest, purity, and elimination of main contaminants. Finally, a complete purification train was implemented using the best results obtained in each step. A final concentration of 1.40 × 1010 virus‐like particles (VLPs)/mL with a purity of 64% after the polishing step was achieved, with host cell DNA and protein levels complaining with regulatory standards, and an overall recovery of 38%. This work has resulted in the development of a purification process for HIV‐1 Gag‐eGFP virus‐like particles suitable for scale‐up. KEYWORDS chromatography, clarification, downstream processing, HIV‐1 Gag‐eGFP VLP, nanoparticle quantification 1 | INTRODUCTION The manufacturing of these particles includes their production by cell culture followed by their purification to meet the requirements of Vaccines stimulate the immune system allowing the individual itself to its final use (Mittal et al., 2022; Vicente et al., 2011). Downstream develop defense mechanisms against a specific agent. Among the processing (DSP) of nanoparticles is still a challenge. Although different types, virus‐like particles (VLPs)‐based vaccines have been filtration and centrifugation techniques are scalable to a certain gaining interest in recent years as they can trigger both humoral and extent and often used in industrial scale for vaccine and gene therapy cellular immune responses (Fuenmayor et al., 2017; Nooraei et al., 2021). products manufacturing, they are labor‐intensive and requiring This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. © 2023 The Authors. Biotechnology and Bioengineering published by Wiley Periodicals LLC. 2672 | wileyonlinelibrary.com/journal/bit Biotechnol Bioeng. 2023;120:2672–2684. 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License LORENZO ET AL. | 2673 expensive equipment (in the case of centrifugation) (Moleirinho Pereira Aguilar et al., 2020; Steppert et al., 2016) Moreover, a et al., 2019). Consequently, in the past few years, the focus has been combination of different binding mechanisms, for example, hydroxyl‐ moving to chromatography and membrane‐based separation tech- functionalized polymethacrylate monoliths is a type of chromatogra- niques (Kramberger et al., 2015). phy that has been used in the purification of HIV‐1 Gag VLPs Despite the efforts made in this field, it is known that even (Steppert et al., 2017). Affinity chromatography (AC) has recently purified VLP preparations are contaminated with some extracellular gained higher importance in biotherapeutics manufacturing field due vesicles (EVs) (González‐Domínguez et al., 2021). They share very to unique selectivity capacity (Bandeira et al., 2012; Segura similar properties that make difficult to obtain final VLP formulations et al., 2013; Zhao et al., 2019). It has been shown that viruses and with high purity (Lavado‐García et al., 2020). VLPs can be purified efficiently using heparin AC (Reiter et al., 2019; Here, we propose the study of a purification process, specifically Zhao et al., 2019). However, after this capture step, usually, it is for enveloped HIV‐1 Gag‐eGFP VLPs (referred here as Gag VLPs). necessary to perform a polishing step. Size‐exclusion chromatogra- Based on the process reported by González‐Domínguez et al. (2021), phy (SEC) and ultrafiltration/diafiltration are the most used alter- focussing on the evaluation of different operational units in each step natives at this stage (Moleirinho et al., 2018; Nweke et al., 2017; of the DSP strategy, with the aim of improving the results previously Vicente et al., 2011). obtained. This DSP proposal includes the study of the following steps: clarification, intermediate purification/concentration, capture chromatography, and polishing (Bandeira et al., 2012; González‐ 2 | M A T E R I A L S AN D M E T H O D S Domínguez et al., 2021; Moleirinho et al., 2018). The best alternatives were determined considering the yield obtained in terms All reagents were acquired from Sigma Aldrich, unless otherwise of recovery of Gag VLPs, the purity with respect to the total mentioned. nanoparticles, including EVs, and the reduction of contaminating proteins and dsDNA. After selecting the best options in the different steps of the proposed DSP, a complete run was performed and 2.1 | Cell culture conditions characterized to demonstrate the output of the overall process. As the target nanoparticles are produced extracellularly, the DSP HEK293SF‐3F6 cells, kindly provided by Dr. Amine Kamen from process is considered to start from supernatant with a clarification McGill University, were cultured in HyCell TransFx‐HTM (Hyclone) step (Mittal et al., 2022). It is necessary to remove cell debris, large medium supplemented with 2% GlutaMAXTM (Gibco, Thermo Fisher particles, and aggregates. Dead‐end disposable depth‐filters are used Scientific) and 0.1% PluronicTM (Gibco, Life Technologies, Thermo in this step and have shown a high yield of the product of interest Fisher Scientific). Cells were routinely maintained in disposable with a high impurity removal capacity (Besnard et al., 2016; polycarbonate Erlenmeyer flasks (Corning), and incubated at 37°C, González‐Domínguez et al., 2021). Long‐term storage at −80°C 5% of CO2, 85% relative humidity, and 130 rpm. Cell quantification ensuing clarification is frequently necessary for large‐scale processes, and viability were determined with the automated cell counter thus, a secondary clarification is sometimes required in DSP NucleoCounter® NC‐3000™ (Chemometec). strategies to remove the precipitates and/or aggregates resulted from thawed sample. Following clarification, an intermediate purifi- cation or concentration step is commonly included in the DSP train. 2.2 | Plasmid Ultrafiltration by tangential flow filtration (TFF) and multimodal chromatography (MC) are widely used (Lima et al., 2019; Moleirinho Pr55Gag polyprotein sequence from the pCMV55M1‐10 plasmid et al., 2018; Pereira Aguilar et al., 2020). Even when there is a bind‐ (Schwartz et al., 1992) was introduced into the pEGFP‐N1 plasmid elution step afterward, the concentration step is beneficial for (Clontech). The resultant plasmid (pGag‐eGFP), coding for a Rev‐ reducing loading times in chromatographic columns while increasing independent HIV‐1 Gag protein fused in frame to the enhanced GFP, the concentration of loaded material (Moleirinho et al., 2018). The can be found in the NIH AIDS Reagent Program (Cat #11468). advantages of including an MC as intermediate purification step is the elimination of impurities that could interact with the resins/ membranes, favoring a good capture of the molecule of interest in 2.3 | VLP production by transient transfection the following bind‐elution steps (Lima et al., 2019; Reiter et al., 2020; Zaveckas et al., 2018). HEK293SF‐3F6 cells were transfected at 1.8–2 × 106 cells/mL, and Different chromatographic modes can be applied for capture and viability greater than 90%. In essence, the DNA at a final purification of these particles. Ion exchange chromatography (IEC) is concentration of 1 μg/mL was added into fresh culture medium and commonly used to capture and concentrate VLPs. Anion exchange vortexed for 10 s. The transfection reagent PEIpro® (Polyplus‐ resins like CIMmultus QA Monolith, Mustang Q, and Capto Q transfection SA) at a DNA:PEI ratio of 1:2 w/w was then added. ImpRes have been the most used for the purification of these The mix was vortexed three times for three seconds and incubated nanoparticles with good yields and purity (Moleirinho et al., 2018; for 15 min at room temperature. Last, formed complexes were added 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2674 | LORENZO ET AL. into the cell culture. Medium was also supplemented with 3.36 mM 2.5 | Preparative chromatography valproic acid and 5.04 mM caffeine 4 h posttransfection to increase cell productivity upon transient gene expression (Cervera et al., 2015). 2.5.1 | Chromatographic system Culture was stopped at 72 hpt to maximize VLP yields. To measure the transfection efficiency, the percentage of GFP‐positive cells was Chromatographic runs were performed using an Äkta Pure 25M2 assessed using a BD FACS Canto II flow cytometer (BD Biosciences). system (Cytiva, GE Healthcare Live Sciences) with a S9 sample pump and a F9‐C fraction collector, equipped with a 1.4 mL mixer chamber. System control, data acquisition, and analyses were performed using 2.4 | Filtration equipment the Unicorn 6.4.1 software (Cytiva, GE Healthcare Live Sciences). UV absorbance (280, 260, and 488 nm wavelengths), pressure, pH, and ® Filtration experiments were performed using an automated Spectrum conductivity were continuously monitored. KrosFlo® Research 2i TFF System (Repligen). 2.5.2 | Chromatography media and mobile phases 2.4.1 | Clarification All preparative chromatographic experiments for Multimodal and Ion Cell culture was stopped at 72 hpt, and after 2 h of sedimentation, Exchange were performed using 50 mM HEPES, pH 7,2 as mobile bulk clarification of the supernatant was directly performed from the phase A and 50 mM HEPES, 2 M NaCl, pH 7,2 as mobile phase B. In shake flask (González‐Domínguez et al., 2021). Each filter was case of Hydrophobic interaction, it was also used 50 mM HEPES, pH preequilibrated with PBS (Hyclone) before filtration. For primary 7.2 as mobile phase A, but 50 mM HEPES, 1.5 M (NH4)2SO4, pH 7.2 clarification, MilliStak®+ D0HC μpod®+ depth filter (Merck), and as mobile phase B. The Capto heparin was performed with 0.1 M Tris, Supracap™ 50V100™ depth filter capsules (Pall Corporation) were 0.01 M Citric Acid, 0.23 M NaCl, pH 7.4 as mobile phase A and 0.1 M compared. The liquid supernatant after cell sedimentation (1L) was Tris, 0.01 M Citric Acid, 2M NaCl, pH 7.4 as B. For POROS heparin loaded to each filter tested. MasterFlex® 96410‐13 silicon tubes was used at 50 mM in Tris‐HCL, pH 7.5 as mobile phase A and 50 mM (Cole‐Parmer) were connected to the filter inlet and outlet; and a Tris‐HCL, 2 M NaCl, pH 7.5 as mobile phase B. Different concentra- pressure sensor (Cole‐Parmer) connected to the filter inlet. The tions of the modifier (NaCL or (NH4)2SO4) were obtained by mixing −1 ® filtration flux was set to 5 mL·min for the MilliStak + D0HC mobile phases A and B using the chromatography system. For SEC μpod®+ depth filter and 4 mL·min−1 for the Supracap™ 50V100 experiments a formulation solution (FS) composed of 20 mM depth filter capsules. After selection of the primary clarification filter, NaH₂PO₄, 50 mM NaCl, 2 mM MgCl₂, 2% Sucrose, pH 7.5 was used 500 mL clarified bulk was used as secondary clarification feed. For as mobile phase. If not further stated, cleaning in place was secondary clarification, the used filters were Sartopore® SartoScale performed using 1 M NaOH solution in MC, IEC and HIC, 0.1 M 25 PP3® (Sartorius AG) and Supor® EAV—Mini Kleenpak™ 20® filter NaOH in AC and 0.5 M in SEC. The chromatography media tested are capsules (Pall Corporation). Their filtration flux was set to 5 mL·min−1 summarized in Table 1. All materials were used for a single cycle. and the same experimental setup was used. After filtration, the syringe filters were emptied with air to recover all the product. Culture supernatant was either stored at 4°C or frozen at −80°C for 2.5.3 | Intermediate purification, capture, long‐time storage. The turbidity of the clarification samples was and polishing of GagVLPs measured using a portable turbidimeter (Thermo Fisher Scientific). For the intermediate purification and capture of Gag VLPs, clarified HEK293 cell culture supernatant from different produced batches 2.4.2 | TFF was directly loaded onto the prepacked columns/membranes HiScreen Capto Core 700, CIMmultus Monolith QA, Mustang QXT The ultrafiltration step was performed using a Centramate Cassette Acrodisc, HiScreen Capto Q ImpRes, HiTrap Capto Heparin, and with a 300 kDa molecular weight cut‐off (MWCO) membrane (Pall manually packed XK16/20 POROS Heparin. The sample loaded into Corporation). The omega PES membrane had an effective filtration CIMmultus Monolith OH− was previously diluted 1:2 with 50 mM area of 0.02 m2. Before starting, the membrane was sanitized with HEPES, 3 M (NH4)2SO4, pH 7,2. Polishing after capture step was 1 M sodium hydroxide (NaOH), neutralized with ultrapure water, and performed with two desalting columns. A prepacked HiTrap preconditioned with PBS 1X. A factor of concentration of 10 times Sephadex G‐25 and a manual packed XK 16/40 Sepharose 4 fast starting with 2L of sample (clarified cell culture supernatant) and a flow (4FF). In the cases of manual packed columns, the packing −1 flux of 35 mL·min were selected. Transmembrane pressure was protocol was performed following the instructions suggested by the maintained at 0.6 bar approximately. After filtration, the membrane manufacturer. In all cases, the flow rates used were those was sanitized with 0.5 M NaOH overnight and neutralized with recommended by the manufacturers. In all chromatographic experi- ultrapure water. ments, equilibration of the stationary phase was performed before 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License LORENZO ET AL. | 2675 TABLE 1 Chromatography media is used for preparative chromatography. Column Type of chromatography Type of column and name Manufacturer volume (mL) Multimodal Agarose‐based core beads HiScreen Capto Core 700 Cytiva 4.7 Ion exchange Polymethacrylate‐based monolithic CIMmultus Monolith QA Sartorius AG 1 column Polypropylene‐based membrane Mustang QXT Acrodisc Pall Corporation 0.86 High‐flow agarose‐based beads HiScreen Capto Q ImpRes Cytiva 4.7 Hydrophobic interaction Polymethacrylate‐based monolithic CIMmultus Monolith OH− Sartorius AG 1 column Affinity (heparin) Agarose‐based beads HiTrap Capto Heparin Cytiva 1 Cross‐linked poly(styrene‐ XK 16/20 POROS Heparin Thermo Fisher Scientific 5.6 divinylbenzene) Size exclusion Cross‐linked dextran‐based beads HiTrap Sephadex G‐25 Cytiva 5 Highly cross‐linked 4% agarose‐based XK 16/40 Sepharose 4 Cytiva 48 beads Fast Flow loading using the corresponding equilibration buffer. After loading, membranes were incubated with NBT‐BCIP solution for 2–3 min. columns were washed with equilibration buffer to ensure the removal The enzymatic reaction was stopped with ultrapure water. of unbound material from the column. In MC, elution was achieved by a 100% buffer B. In capture experiments, elution was achieved by salt step gradients. In SEC step, Gag VLPs were eluted with the FS in 2.7 | Nanoparticle tracking analysis (NTA) the void volume. Details of flow rates, loading volumes, and elution gradients are summarized in Table 2. After the elution phase, columns NTA was used to determine particle size distribution and particle were regenerated using 100% B buffer. Fractions were collected and concentration. NanoSight® NS300 device (Malvern Panalytical) pooled according to the chromatograms, considering the UV equipped with a blue laser module (488 nm) to quantify HIV‐1 Gag‐ absorbance signals of 280, 260, and 488 nm. eGFP VLPs, and a neutral density filter for total particle quantification by light scattering. Samples were serially diluted in particle‐free water to achieve a concentration of 20–80 particles per video frame (the 2.6 | Western blot and SDS‐PAGE instrument's linear range). For the injected samples, three replicates of each dilution were measured at 22°C, with a viscosity of 0.9 cP, Forty microliters of sample were mixed with 20 μL of 4x LDS Sample and three videos of 60 s length were acquired. Capture settings were Buffer and 7 μL of 2M DTT, followed by 20 min incubation at 96°C. recorded with an sCMOS camera, setting the screen gain to 1, Stored at 4°C until gelled, 20 μL of each sample were loaded onto manually adjusting the camera level before each measurement (eight precast NuPAGE Bis/Tris gels 4%–12% (Invitrogen). Five microliters for samples containing Gag‐eGFP VLPs and 11 for controls), and ® of SeeBlue Plus2 Prestained Protein Standard (Invitrogen) was used setting a detection threshold of 4. The remaining analysis parameters as low molecular weight control. Gels were run at 200 V, 400 mA, were automatically selected by the software and kept constants for 45 min, in MES‐SDS running buffer. For SDS‐PAGE gels, proteins all samples. Data were acquired and processed with the NanoSight were stained with Coomassie Brilliant Blue G‐250 based EZBlueTM NTA software version 3.2. Gel Staining Reagent (Sigma Aldrich). For Western blot analysis, proteins were transferred onto 0.2 μm nitrocellulose membranes using the Trans‐Blot® turbo system (Bio‐Rad Laboratories). Mem- 2.8 | Fluorescence intensity measurements branes were blocked with PBS 5% (w/v) nonfat dry milk for 30 min, washed with PBS 0.1% (w/v) Tween‐20, and then incubated The fluorescence intensity was measured in a Cary Eclipse Fluores- overnight at 4°C with the primary monoclonal antibody against cence Spectrophotometer (Agilent Technologies) to determine the HIV‐1 p24 (dilution 1:2000 in PBS) (Icosagen AS). After washing, concentration of Gag‐eGFP VLPs. The settings were λexcitation = product immunodetection was performed with an antimouse IgG 488 nm (slit = 5 nm) and λemission = 510 nm (slit = 10 nm). The rela- antibody conjugated with an alkaline phosphatase (dilution 1:5000 in tive fluorescence unit (RFU) values were calculated by subtracting PBS 1X), incubated 2 h at room temperature, and washed with PBS the obtained fluorescence intensity (FU) values from the non‐ 0.1% (w/v) Tween‐20. For protein bands visualization, the transfected negative control samples. An in‐house developed and | 2676 TABLE 2 Parameters used in the chromatography experiments. Flow rate Column Residence Loaded Step gradient Type of chromatography (mL/min) volume (mL) time (min) Loaded materiala volume (mL) Equilibration and wash elution MC Capto Core 700 (Lagoutte et al., 2016; 1.2 4.7 3.92 8.53E + 11 70,5 6% B in 5CV 100% B in 10CV Reiter et al., 2020) IEC Monolith QA (Pereira Aguilar et al., 2020; 1 1 1.00 4.44E + 12 120 5% B in 5CV 15%, 35%, 45%, and Steppert et al., 2016) 65% B in 20CV IEC Mustang Q (Carvalho, Silva, et al., 2019; 0.8 0.86 1.08 3.6E + 12 240 5% B in 5CV 15%, 35%, 45%, and González‐Domínguez et al., 2021) 65% B in 20CV IEC Capto Q ImpRes (Moleirinho et al., 2019) 1.2 4.7 3.92 7.0E + 11 70,5 6% B in 5CV 15%, 35%, 45%, and 65% B in 20CV HIC Monolith OH‐ (Steppert et al., 2017) 1 1 1.00 5.7E + 11 77 100% B in 5CV 100%, 65%, 0% B in 20CV AC Capto Heparin (Pereira Aguilar et al., 2020; 1 1 1.00 8.1E + 11 45 5% B in 8CV 12%, 25%, 50% B Reiter et al., 2019) in 10CV AC POROS Heparin (Zhao et al., 2019) 2 5.6 2.80 1.4E + 12 100 5% B in 5CV 15%, 35%, 45%, and 65% B in 20CV SEC Sephadex G‐ (Cervera et al., 2017) 1 5 5.00 9.20E + 10 0,5 FS in 5CV 100% FS in 2CV SEC Sepharose 4 FF (Carvalho, Silva, et al., 2019; 1 48 48.00 1.78E + 12 5 FS in 5CV 100% FS in 2CV González‐Domínguez et al., 2021) Abbreviations: AC, affinity chromatography; B, buffer corresponding to mobile phase B; CV, column volume; FS, formulation buffer; HIC, hydrophobic interaction chromatography; IEC, ion exchange chromatography; MC, multimodal chromatography; SEC, size exclusion chromatography. a Load material is referred to total nanoparticles (including Gag VLPs and EVs) quantify by NTA. LORENZO ET AL. 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License LORENZO ET AL. | 2677 validated quantification assay was used to convert RFU values to of 10 s were performed for each independent measurement. The Gag‐eGFP concentration values (Cervera et al., 2013). hydrodynamic diameter (MHD), particle size distribution in volume, derived count rate, and polydispersity index (PDI) average results were automatically obtained. 2.9 | Total protein and dsDNA quantification Host cell protein concentration was determined using the Micro BCA 3 | RESULTS AND DISCUSS I ON protein assay kit, according to the manufacturer's instructions. Briefly, serial PBS‐buffer dilutions (between 2‐ and 256‐fold) of 3.1 | Gag VLPs production and clarification step standard and samples were dispensed into the wells. After adding 150 μL of the Micro BCA working reagent into each well, plates were To produce enough particles to evaluate the different alternatives in incubated for 1 h at 37°C. The bovine serum standard curve ranged the DSP strategy, different HEK 293 cell culture batches were from 1.5 to 200 μg/mL, and the absorbance was measured at 562 nm transfected. At 72 hpt (Cervera et al., 2017), there was a transfection in the Victor 3 reader (Perkin Elmer). Protein concentration of the efficiency in each batch around ~80% samples was determined using the standard curve. After removal of cells by sedimentation, the first step in the DSP Host cell DNA concentration was determined using the Quant‐ strategy is critical to improve the subsequent purification steps and itTM Picogreen dsDNA assay kit, according to the manufacturer's consists in the clarification. For this step, two depth‐filters were instructions. Briefly, serial TE 1X‐ buffer dilutions (between 2 and evaluated regarding Gag VLPs recovery and enrichment in respect to 256‐fold) of standard and the samples were dispensed into 96‐well the total number of nanoparticles, and the reduction of main proteins microplates. After adding 100 μL of the Quant‐iTTM PicoGreen® and dsDNA contaminants. These parameters were referred to the reagent (dilution 1:1000) into each well, plates were incubated for loaded sample in each case. Supporting Information: Table S2 shows 5 min at room temperature. The dsDNA standard curve ranged from the results regarding two the parameters evaluated. 1.5 to 500 μg/mL, and the absorbance was measured on the Victor 3 The DLS profile followed the same trend in both cases, although (Perkin Elmer), prior and later to the Quant‐iTTM PicoGreen® reagent the peak was slightly higher for the Supracap™ 50V100 depth filter addition since HIV‐1 Gag‐eGFP VLPs emit at the same range. The since there was a lower loss of Gag VLPs (Figure 1a). The smaller peak settings were λexcitation = 480 nm and λemission = 520 nm. DNA of the supernatant seen on the right may be due to cell debris or concentration of the samples was determined using the standard aggregates, which was diminished by the clarification. Electrophor- curve and subtracting the native fluorescence. esis confirmed the presence of a band greater than 75 kDa (the molecular weight of Gag‐eGFP is approximately 88 kDa), confirmed to be Gag‐eGFP, either free monomer or forming part of VLPs, by 2.10 | Flow virometry analysis Western blot (Figure 1b). Higher recovery and purity were achieved with the Supracap™ Concentration of Gag‐eGFP fluorescent events was quantified using 50V100 depth filter. One possible explanation is the difference in a CytoFLEX LX (Beckman Coulter), equipped with a 405 nm filter. The nominal retention rating. The Supracap™ 50V100 depth filter has a threshold of the trigger signal (VSSC) was manually adjusted to 1500, nominal filter retention range between 2 and 4 μm; and the and gains were set as 9, 95, and 115 for VSSC, FSC, and B525‐FITC MilliStak®+ D0HC μpod depth filter has a nominal filter retention lasers, respectively. Samples were diluted with PBS 1X until an abort between 0.5 and 10 μm approximately (Carvalho, Silva, Moreira, rate value below 2% was achieved. A total of 300,000 events were et al., 2019). This implies that more heterogenous particles in size can analyzed at a flow rate of 10 μL/min per sample. VSSC‐H versus pass through the MilliStak®+ D0HC μpod depth filter and they can B525‐FITC density plots were used to gate the different particle also retain VLPs. It can be observed in the DLS data that the sample populations. Gating was adjusted manually for each channel. Events from MilliStak®+ D0HC μpod depth filter has a wider size distribution after 50 s were taken for analysis. The results were analyzed with than the sample from Supracap. Higher recoveries of Gag VLPs were CytExpert software. achieved in previous studies using the MilliStak®+ D0HC μpod depth filter (Boix‐Besora et al., 2022; González‐Domínguez et al., 2021). A possible cause could be that in both reported cases a lower amount 2.11 | Dynamic light scattering of Gag VLPs was loaded compared to the amount loaded here, however, it has been previously demonstrated that clarification based The dynamic light scattering technique was used to determine the on nominal filtration processes performs fairly well with recovery size of HIV‐1 Gag‐GFP VLPs. The measurements were carried out at yields of around 80%, which is consistent with our results (Carvalho, 25°C in a Zetasizer Nano ZS instrument (Malvern Instruments), Silva, Moleirinho, et al., 2019). equipped with a He/Ne 633 nm laser at 173°C. One hundred It should also be considered that a secondary clarification in DSP microliters of the sample were placed in disposable plastic cuvettes. strategies is sometimes required. An additional clarification step is Three consecutive measurements of each sample with 10–15 scans added in case that a previously frozen sample was to be used and 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2678 | LORENZO ET AL. precipitates and/or aggregates appear during its thawing. The preparation with Gag VLPs in respect to the total nanoparticles, since turbidity of the sample is an indication of how necessary it is to the large particles have already been removed by depth filtration. clarify again before continuing with the subsequent chromatography Moreover, although both filters reduced the amount of dsDNA steps. For that reason, two different filters were evaluated respect to the load, it appears that the Supor® EAV—Mini Kleenpak™ ® ® (Sartopore SartoScale 25 PP3 and Supor EAV—Mini Kleenpak™ 20 filter had some selectivity for dsDNA and certain proteins since 20 filters) in case this step needs to be included in the purification the observed reduction was higher compared to Sartopore® process. The same parameters analyzed in the case of primary SartoScale 25 PP3 filter. Although the GagVLPs did not increase clarification were also considered in these experiments and the with respect to total particles, it was possible to remove part of the results were summarized in Supporting Information: Table S3. contaminants, obtaining an optimal sample characteristics that allows Figure 2a shows the DLS results of both clarified materials, where a better performance of the column for the next chromatographic a similar profile is observed. The presence of Gag‐eGFP was step (Besnard et al., 2016). confirmed by SDS‐PAGE and Western blot at the expected molecular The best results in primary clarification were obtained with the size (Figure 2b). Supracap™ 50V100 depth filter. Regarding secondary clarification, The null enrichment of Gag VLPs in the secondary clarification the Supor® EAV—Mini Kleenpak™ 20 filter was selected as the best could be because conventional filtration does not serve to enrich the option, if required. The selected method offers a robust approach for F I G U R E 1 Primary clarification of Gag virus‐like particles (VLPs) supernatant. (a) Dynamic light scattering (DLS) analysis of the MilliStak®+ D0HC μpod depth filter load (polydispersity index [PDI] = 0.439, hydrodynamic diameter [MHD] = 223.8), MilliStak®+ D0HC μpod depth filter clarified PDI = 0.387, MHD = 195.6), Supracap™ 50V100 depth filter load (PDI = 0.457, MHD = 242.3), Supracap™ 50V100 depth filter clarified (PDI = 0.239, MHD = 161.2). (b) SDS‐PAGE and p24 Western blot of the depth filtration runs. M, molecular weight standard; L1 and L2, Load samples using for test MilliStak®+ D0HC μpod depth filter and Supracap™ 50V100 depth filter respectively; CL1 and CL2, clarified samples corresponding to MilliStak®+ D0HC μpod depth filter and Supracap™ 50V100 depth filter, respectively. F I G U R E 2 Secondary clarification of Gag virus‐like particles (VLPs). (a) Dynamic light scattering (DLS) analysis of the Sartopore® SartoScale 25 PP3 filter clarified (PDI = 0.329, MHD = 137.8), Supor® EAV—Mini Kleenpak™ 20 filter clarified (PDI = 0.321, MHD = 143.3). (b) SDS‐PAGE and p24 Western blot of the filtration runs, using a Sartopore® SartoScale 25 PP3 and a Supor® EAV—Mini Kleenpak™ 20 filters. PDI, polydispersity index; MHD, mean hydrodynamic diameter; M, molecular weight standard, CL2, Load sample using for test both filters; CL2.1 and CL2.2, clarified samples corresponding to Sartopore® SartoScale 25 PP3 filter and a Supor® EAV—Mini Kleenpak™ 20 filters, respectively. 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License LORENZO ET AL. | 2679 the clarification of GagVLPs. Similar results were reported Another alternative to follow in this intermediate purification (Carvalho, Silva, Moreira, et al., 2019) for the clarification of influenza step is the use of a MC. In this work the HiScreen Capto Core 700 VLPs produced in insect cells. column was evaluated considering the parameters described (Sup- porting Information: Table S5). The mass balance results suggest that most of the contaminating proteins in the evaluated sample have a 3.2 | Intermediate step high molecular weight, and therefore, 88% remain in the FT together with the Gag VLPs (Reiter et al., 2019). A narrower profile was Following clarification, a concentration step is often included in a observed by DLS (Figure 3a), with a lower PDI and MHD, which purification process. It is very useful when the volume of the starting indicated the presence of less aggregates in the FT fraction, which material is high, and it is also a step that allows the removal of was then later confirmed to be Gag‐eGFP by Western blot contaminants (Moleirinho et al., 2018; Vicente et al., 2011). For that (Figure 3b). reason, in this work we evaluated a TFF 300 kDa PES membrane. In In previous reports, a yield of 90.9% has been reported in the this case, the operation was characterized considering all the purification of Zika VLPs (Lima et al., 2019), 89.7% for yellow parameters evaluated in the previous steps (Supporting Information: fever VLPs (Lima et al., 2019) and 89% for influenza VLPs Table S4). (Lagoutte et al., 2016). However, it must be considered that for All Gag VLPs have been recovered and concentrated 10‐fold, HIV‐1 Gag VLPs a maximum of 73.1% recovery has been reported with presence of some aggregates and a shift in the particle size as (Reiter et al., 2019), close to the results obtained here. In respect shown by the DLS profile (Figure 3a). However, when this sample to other studies, although similar residence times and loading was diluted and analyzed by NTA, although the presence of volumes were evaluated, the total number of nanoparticles aggregates was also observed, the particle size obtained for the loaded to the column was lower (Lagoutte et al., 2016; Lima Gag VLPs was 146 nm, which corresponds to the expected results for et al., 2019; Reiter et al., 2019). Possibly this could be the cause these nanoparticles. In the electrophoresis (Figure 4f), a band greater of the low recovery percentages and reduction of contaminating than 75 kDa could be seen in the retentate, which did not appear in proteins observed. the permeate (data not shown). It was confirmed that this band corresponds to Gag‐eGFP by Western blot (Figure 3b). T‐Series cassettes with omega polyethersulfone (PES) mem- 3.3 | Capture step brane, and molecular weight cut‐off of 1000 and 300 kDa, have been reported for VLPs concentration enabling a removal of >80% of The evaluation of the capture step included six different proteins while recovering approximately 60%–80% of VLPs (Lin chromatography methodologies: three based on ion exchange, et al., 2015; Venereo‐Sanchez et al., 2016). In this case, ultra- one based on hydrophobic interactions and two based on heparin filtration/diafiltration using TFF mode to concentrate Gag‐eGFP VLPs affinity. The characteristics of the samples loaded in the six is proposed as an intermediate step in DSP train with better results chromatographies are summarized in Supporting Information: than previously reported (Nooraei et al., 2021; Venereo‐Sanchez Table S1. All runs were analyzed considering the parameters et al., 2016). described above for the other steps. F I G U R E 3 Tangential flow filtration (TFF) and multimodal chromatography experiments for concentration and intermediate purification of Gag virus‐like particles (VLPs). (a) Dynamic light scattering (DLS) analysis of the TFF retentate PDI = 0.501, MHD = 241.8), FT Capto Core 700 column PDI = 0.166, MHD = 165.2); (b) SDS‐PAGE and p24 Western blot of the filtration run with a 300 kDa PES membrane by TFF, and a chromatographic run with a Capto Core 700 column. PDI, polydispersity index; MHD, mean hydrodynamic diameter; M, molecular weight standard; L load sample used in TFF and MC; R, retentate sample resulted from TFF; FT flow through obtain in MC. 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2680 | LORENZO ET AL. Among several chromatographic techniques, IEC has been findings this method can separate the main impurities including host already used to purify VLPs. Here, three IEC columns were tested: cell protein and DNA. The recovery of Gag VLPs, in the main elution Monolith QA, Mustang Q, and Capto Q ImpRes. In all the cases, fraction is less than 50%, however these results are comparable to the fractions obtained during the runs were analyzed from the data previously reported (Pereira Aguilar et al., 2020; Steppert corresponding chromatogram. The second step in the elution phase et al., 2016). fraction (named E2) was considered the main product fraction due to For the Mustang Q, the mass balance results are show in the higher GagVLPs concentration and simultaneous lower total Supporting Information: Table S7. Here, an improvement in yield with protein and dsDNA content (Supporting Information: Figure S2). The the Mustang Q compared to previously reported results was obtained E2 fraction, corresponds to conductivity values of 45–55 mS/cm. (Boix‐Besora et al., 2022; González‐Domínguez et al., 2021). A Similar elution conditions for VLPs have been reported (Pereira possible explanation for this could be that, in this case, the loaded Aguilar et al., 2019; Steppert et al., 2016). Therefore, the subsequent sample had been previously clarified using both filters, Supracap™ comparative analyses were performed focusing on the E2 fraction. 50V100 and Supor® EAV—Mini Kleenpak, which allows a better In the Monolith QA chromatographic run the results achieved are performance of the membrane reducing as many precipitates as summarized in Supporting Information: Table S6. According to our possible in the sample and avoiding system overpressure. In addition, F I G U R E 4 Purification of Gag virus‐like particles (VLPs) by an ion exchange, hydrophobic interaction, and affinity chromatography. (a) Dynamic light scattering (DLS) analysis of the E2 fraction of the Monolith QA column (PDI = 0.190, MHD = 137.9), Mustang Q column (PDI = 0.176, MHD = 135.5) and Capto Q ImpRes column (PDI = 0.151, MHD = 152.0), E3 fraction of the Monolith OH column (PDI = 0,18, MHD = 132), W fraction of the Capto Heparin column (PDI = 0,13, MHD = 140) and E1 of the POROS Heparin column (PDI = 0,1, MHD = 162,8). (b) SDS‐PAGE and p24 Western blot of the production, load material and the GagVLPs main fraction of each chromatographic run. E1, E2, and E3 elution peaks from respective step gradient W, fraction corresponding to wash step; PDI, polydispersity index; MHD, mean hydrodynamic diameter; M, molecular weight standard, P, Supernatant production sample; L, load sample MQA, MQ, CQ, MOH, CH, PH, main fraction corresponding to Monolith QA, Mustang Q, Capto Q ImpRes Monolith OH, Capto Heparin and POROS heparin columns, respectively. F I G U R E 5 Purification of Gag virus‐like particles (VLPs) by size exclusion chromatography. (a) Dynamic light scattering (DLS) analysis of the VV fraction of the Sephadex G‐25 column (PDI = 0.174, MHD = 139.9), and Sepharose 4FF column (PDI = 0.217, MHD = 172.8). (b) SDS‐PAGE and p24 Western blot of the chromatographic runs using a Sephadex G‐25 column and a Sepharose 4FF column. MHD, mean hydrodynamic diameter; PDI, polydispersity index; VV, void volume; M, molecular weight standard; SG25, VV from Sephadex G‐25 column; S4FF, VV from Sepharose 4FF column¸ L, loaded sample in SG25 and S4FF. 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License LORENZO ET AL. | 2681 Note: Total particle, total protein, and dsDNA concentrations were measured by nanoparticle tracking analysis (NTA), BCA, and Picogreen assays, respectively. VV, void volume fraction; E2, step gradient elution the concentration of loaded material in this case was lower than in Gag‐eGFP/total the previously reports, which has also allowed better results. protein (μg/mL) protein (%) Highest recovery was achieved using Capto Q ImpRes which 3 allowed not only the separation of GagVLPs from host cell protein 2 32 70 and dsDNA but also the enrichment respect to total nanoparticles (Supporting Information: Table S8). The yield obtained was similar when using the same resin for adenoviruses (~80%), according to Gag‐eGFP previous reports (Moleirinho et al., 2018). These data support the use of strong quaternary amine ion exchangers to concentrate and purify 15.1 10.2 175 21 HIV‐1 VLPs. The use of hydroxyl‐functionalized polymethacrylate monolith as reductiona (%) one more possibility in the capture step was also tested. The fraction corresponding to elution step 3 showed the highest content of Gag dsDNA VLPs (22%). In this case the elution buffers used were different than ‐ 100 65 96 96 in IEC experiments and the major Gag VLP concentration fraction presents conductivity values approximately of 20 mS/cm (Supporting Information: Figure S3c and Table S9). Similar results were reported (ng/mL) dsDNA 488.4 regarding the recovery of Gag VLPs in two main elution peaks using 613 217 5.8 this column (Steppert et al., 2017). Mass balance of Gag virus‐like particles (VLPs) obtained from final downstream processing (DSP) validation run. Regarding the enrichment of VLPs, no increase was observed, reductiona (%) however, an almost total elimination of contaminating host cell Total protein proteins and DNA was observed (Supporting Information: Table S9). The overall advantages for macroporous polymethacrylate monoliths 2 98 83 100 are well established (Jungbauer & Hahn, 2004). From the two types ‐ of functionalized monoliths tested, the best results were obtained Total protein with the Monolith QA, which suggests that the ion exchange This percentage was calculated taking into account that the sample loaded in SEC was diluted four times. (μg/mL) principle provided by the QA ligand allows a better recovery of Final recovery/reduction of the entire process with respect to the initial sample (crude supernatant). 560.9 30.0 550 542 the Gag VLPs, compared to the hydrophobic exchanges provided by the OH ligand. Gag‐eGFP VLPs/ total particle (%) Several studies have shown the potential of heparin AC for the purification of VLPs allowing its separation from other nanoparticles populations (Reiter et al., 2019). Different chromatography media have been used to couple the heparin ligand (Zhao et al., 2019). Here, 38 43 60 64 Capto heparin and POROS heparin columns were tested. In both cases it was observed a similar chromatographic profile where the recoverya (%) fraction with higher amount of Gag VLPs elute at 20‐25 mS/cm values of conductivity with high reduction of protein and dsDNA VLPs's 55b 90 76 38 100 contaminants (Supporting Information: Figure S4). Despite these Refers to the recovered/reduced from the previous step. similarities, POROS heparin was superior in its ability to capture the Gag VLPs (Supporting Information: Tables S10 and S11). The results obtained show the effectiveness of this resin to separate the Gag eGFP VLPs 11 Total gag‐ 8.8 × 1011 11 8.4 × 1010 6.04 × 10 7.9 × 10 VLPs from other nanoparticles present in the sample. The PDI and MHD analyzed by DLS were very similar in all peak 2; S4FF, Sepharose 4 fast flow. capture chromatography evaluated (Figure 4a). Also, it was observed in all the cases a band with the expected size by SDS‐PAGE, Volume confirmed to be Gag‐eGFP by Western blot (Figure 4b). (mL) 2 6 100 100 Considering that the initial samples used for the tests were different, the main criteria of evaluation were based on the recovery E2 fraction CaptoQ VV fraction S4FF of Gag VLPs, considering the total Gag VLPs/mL of column/ membrane loaded and eluted in each case. According to the obtained Production ImpRes TABLE 3 results, the Capto Q ImpRes column offers a better recovery of the Clarified c Sample Overall product of interest, enriching its presence with respect to the rest of the nanoparticles, and at the same time maintaining low amounts of b a c 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2682 | LORENZO ET AL. contaminants. These parameters, together with the molecular techniques showing a better performance at each step. First, characterization of the Gag‐VLPs, make this column the optimal clarification was performed using the Supracap™ 50V100 depth option for the capture and purification of the Gag VLPs. filter. Secondary clarification was not necessary, since the turbidity values were acceptable after the first filtration, which ruled out the presence of large aggregates and precipitates. The next step was 3.4 | Polishing step capture with the HiScreen Capto Q ImpRes column. The TFF step was not included since in this case it was not necessary to concentrate the Polishing is a critical step to achieve a final clinical grade material. sample before chromatography. In general terms, it is suggested to After capture step we evaluated size exclusion chromatography (SEC) consider the TFF step in the DSP train when it is necessary to as a polishing step with two different columns, HiTrap® Sephadex G‐ concentrate the sample. After IEC, desalting was performed with XK 25 column, and a XK 16/40 Sepharose 4FF column. The same 16/40 Sepharose 4FF column. The mass balance of this complete parameters as in the previous steps were used for characterization. In validation run is summarized in Table 3. SEC, Gag VLPs elute in the void volume (VV) (Supporting Information: Furthermore, it can be noted that the VLPs obtained after the Figure S5). The mass balance results of both runs are summarized in polishing step were in a formulation solution, which is an Supporting Information: Tables S12 and S13. The PDI and MHD are advantage for their subsequent lyophilization. The results obtained very similar. Also, a band greater than 75 kDa was observed by from the global yield of the process and the purity of the product electrophoresis and Western blot (Figure 5b). of interest improved those previously reported (Boix‐Besora The Sepharose 4 FF gave better results compared to the et al., 2022; González‐Domínguez et al., 2021; Steppert Sephadex G‐25 column. A possible explanation could be the use of et al., 2016, 2017). different resins, the difference in the column dimensions, or both On the other hand, Figure 6 shows a molecular characterization (González‐Domínguez et al., 2021). Even so, in this step, it was still of Gag VLPs obtained in each step. Final concentrated and purified possible to eliminate more dsDNA and total protein for both cases. Gag VLPs are clearly observed in cryo‐TEM micrographs (Figure 6a), This polishing step also has the advantage of enabling the exchange compared to other nanoparticles present in the sample. Gag VLPs are of the process buffer for a formulation or storage buffer (Peixoto detected as spherical electrodense nanoparticles surrounded by a et al., 2007) (Moleirinho et al., 2018). lipid membrane. On the other hand, as observed in the SDS‐PAGE, as the process progresses, the sample corresponding to Gag‐GFP is concentrated and other contaminating proteins are eliminated. 3.5 | Evaluation of the optimized purification Furthermore, in all steps the presence of Gag‐eGFP was confirmed process by Western blot (Figure 6b). The DLS also allowed to characterize the sample in each step of the process. It is observed how the peak To finally validate the process, a 100 mL sample from a production narrows with respect to the production supernatant as the sample is batch was treated through the complete purification train, using the increasingly purified (Figure 6c). F I G U R E 6 Molecular and biophysical characterization of Gag virus‐like particles (VLPs) obtained in all steps of the final downstream processing (DSP) run (a) Cryo‐TEM micrographs of purified material; (b) SDS‐PAGE, p24 Western blot, (c) Dynamic light scattering (DLS) M, molecular weight standard; S, supernatant; C, clarified; IEC, elution peak 2 from Capto Q ImpRes and SEC, void volume from Sepharose 4FF. Red, blue, and black narrows indicate Gag VLPs, others nanoparticles and Gag‐eGFP protein, respectively. 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License LORENZO ET AL. | 2683 4 | C ONC LUS I ON S Besnard, L., Fabre, V., Fettig, M., Gousseinov, E., Kawakami, Y., Laroudie, N., Scanlan, C., & Pattnaik, P. (2016). Clarification of vaccines: An overview of filter based technology trends and best The design of HIV‐1 Gag‐eGFP purification process, comparing practices. Biotechnology Advances, 34(1), 1–13. https://doi.org/10. different unit operations in each step has been achieved. In the 1016/j.biotechadv.2015.11.005 primary clarification, it is possible to separate most of the cell Boix‐Besora, A., Lorenzo, E., Lavado‐García, J., Gòdia, F., & Cervera, L. (2022). aggregates and debris from the HIV‐1 Gag‐eGFP VLPs of the Optimization, production, purification and characterization of HIV‐1 GAG‐based virus‐like particles functionalized with SARS‐CoV‐2. supernatant by means of depth filtration, without involving a Vaccines, 10(2), 250. https://doi.org/10.3390/vaccines10020250 significant loss of VLPs. This is refined by secondary clarification by Carvalho, S. B., Silva, R. J. S., Moleirinho, M. G., Cunha, B., Moreira, A. S., filtration, and the dsDNA content is reduced. TFF enables further Xenopoulos, A., Alves, P. M., Carrondo, M. J. T., & Peixoto, C. (2019). concentration, and more than 85% of dsDNA and total protein Membrane‐based approach for the downstream processing of content removal. Capto Q ImpRes IEC shows the best results among influenza virus‐like particles. Biotechnology Journal, 14, 8. https:// doi.org/10.1002/biot.201800570 all chromatography tested in capturing and purifying VLPs. Last, it is Carvalho, S. B., Silva, R. J. S., Moreira, A. S., Cunha, B., Clemente, J. J., possible to achieve a higher purity, as well as a desalting and buffer Alves, P. M., Carrondo, M. J. T., Xenopoulos, A., & Peixoto, C. (2019). exchange with the S4FF SEC. The process has shown a total recovery Efficient filtration strategies for the clarification of influenza virus‐like of 38% of Gag‐eGFP VLPs with a total protein, and dsDNA removal particles derived from insect cells. Separation and Purification Technology, 218, 81–88. https://doi.org/10.1016/j.seppur.2019.02.040 of almost 100% from the crude supernatant. Along with a robust Cervera, L., Fuenmayor, J., González‐Domínguez, I., Gutiérrez‐Granados, purification platform, efficient process monitoring tools are critical S., Segura, M. M., & Gòdia, F. (2015). Selection and optimization of for process development and robust characterization to meet transfection enhancer additives for increased virus‐like particle regulatory demands. Analytical methods such as DLS, NTA, flow production in HEK293 suspension cell cultures. Applied Microbiology and Biotechnology, 99(23), 9935–9949. https://doi.org/10.1007/ virometry, electrophoresis, Western blot, and others performed in s00253-015-6842-4 this study, have been widely applied to monitor multiple quality Cervera, L., Gonz Alez‐Dom Inguez, I., Ia, M., Segura, M., & Godia, F. attributes of VLPs (González‐Domínguez et al., 2021; Pereira Aguilar (2017). Intracellular characterization of Gag VLP production by et al., 2020; Reiter et al., 2019). Overall, the DSP unit operations and transient transfection of HEK 293. Biotechnol Bioeng, 114(11), 2507–2517. https://doi.org/10.1002/bit.26367 the analytical tools studied here are suitable for large‐scale Cervera, L., Gutiérrez‐Granados, S., Martínez, M., Blanco, J., Gòdia, F., & manufacturing, supporting the use of the developed DSP bioprocess Segura, M. M. (2013). Generation of HIV‐1 Gag VLPs by transient as platform for HIV‐1 Gag VLPs purification. transfection of HEK 293 suspension cell cultures using an optimized animal‐derived component free medium. Journal of Biotechnology, A C KN O W L E D G M E N T S 166(4), 152–165. https://doi.org/10.1016/j.jbiotec.2013.05.001 Fuenmayor, J., Gòdia, F., & Cervera, L. (2017). Production of virus‐like The authors would like to acknowledge the help of Marti de Cabo particles for vaccines. New Biotechnology, 39, 174–180. https://doi. (Servei de Microscopia, UAB, Barcelona, Spain) for the assistance org/10.1016/j.nbt.2017.07.010 with cryo‐TEM experiments. DLS and NTA measurements were González‐Domínguez, I., Lorenzo, E., Bernier, A., Cervera, L., Gòdia, F., & performed at NanoQAM service located at the Service of Preparation Kamen, A. (2021). A four‐step purification process for gag vlps: from culture supernatant to high‐purity lyophilized particles. Vaccines, and Characterization of Soft Materials located at Institut de Ciència 9(10), 1154. https://doi.org/10.3390/vaccines9101154 de Materials de Barcelona (ICMAB‐CSIC, Campus UAB), where the Jungbauer, A., & Hahn, R. (2004). Monoliths for fast bioseparation and help of Jose Amable Bernabé is appreciated. We would also like to bioconversion and their applications in biotechnology. Journal of thank Manuela Costa Garcia from the Servei de Cultius Cel·lulars, Separation Science, 27(10–11), 767–778. https://doi.org/10.1002/ Producció d'Anticossos i Citometria (UAB, Barcelona, Spain) for its jssc.200401812 Kramberger, P., Urbas, L., & Štrancar, A. (2015). Downstream processing precious help with Flow virometry experiments. The help of Paula and chromatography based analytical methods for production of Grau García and Irene González‐Domínguez is kindly appreciated. vaccines, gene therapy vectors, and bacteriophages. Human Vaccines & Immunotherapeutics, 11(4), 1010–1021. https://doi.org/10.1080/ D A TA A V A I L A B I L I T Y S T A T E M E N T 21645515.2015.1009817 Lagoutte, P., Mignon, C., Donnat, S., Stadthagen, G., Mast, J., Sodoyer, R., The data that support the findings of this study are available from the Lugari, A., & Werle, B. (2016). Scalable chromatography‐based corresponding author upon reasonable request. purification of virus‐like particle carrier for epitope based influenza A vaccine produced in Escherichia coli. Journal of Virological Methods, ORCID 232, 8–11. https://doi.org/10.1016/j.jviromet.2016.02.011 Elianet Lorenzo http://orcid.org/0000-0002-5442-0831 Lavado‐García, J., Jorge, I., Cervera, L., Vázquez, J., & Gòdia, F. (2020). Multiplexed quantitative proteomic analysis of HEK293 provides insights into molecular changes associated with the cell density REFERENCES effect, transient transfection, and virus‐like particle production. Bandeira, V., Peixoto, C., Rodrigues, A. F., Cruz, P. E., Alves, P. M., Journal of Proteome Research, 19(3), 1085–1099. https://doi.org/10. Coroadinha, A. S., & Carrondo, M. J. T. (2012). Downstream 1021/acs.jproteome.9b00601 processing of lentiviral vectors: Releasing bottlenecks. Human Gene Lima, T. M., Souza, M. O., & Castilho, L. R. (2019). Purification of flavivirus Therapy Methods, 23(4), 255–263. https://doi.org/10.1089/hgtb. VLPs by a two‐step chomatographic process. Vaccine, 37(47), 2012.059 7061–7069. https://doi.org/10.1016/j.vaccine.2019.05.066 10970290, 2023, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bit.28419 by Readcube (Labtiva Inc.), Wiley Online Library on [03/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2684 | LORENZO ET AL. Lin, S. Y., Chiu, H. Y., Chiang, B. L., & Hu, Y. C. (2015). Development of sequence in human immunodeficiency virus type 1 results in EV71 virus‐like particle purification processes. Vaccine, 33(44), Rev‐independent gag expression. Journal of Virology, 66(12), 5966–5973. https://doi.org/10.1016/j.vaccine.2015.04.077 7176–7182. https://doi.org/10.1128/jvi.66.12.7176-7182.1992 Mittal, M., Banerjee, M., Lua, L. H., & Rathore, A. S. (2022). Current status Segura, M. M., Mangion, M., Gaillet, B., & Garnier, A. (2013). New and future challenges in transitioning to continuous bioprocessing of developments in lentiviral vector design, production and purification. virus‐like particles. Journal of Chemical Technology & Biotechnology, Expert Opinion on Biological Therapy, 13(7), 987–1011. https://doi. 97(9), 2376–2385. https://doi.org/10.1002/jctb.6821 org/10.1517/14712598.2013.779249 Moleirinho, M. G., Rosa, S., Carrondo, M. J. T., Silva, R. J. S., Hagner‐ Steppert, P., Burgstaller, D., Klausberger, M., Berger, E., Aguilar, P. P., McWhirter, Å., Ahlén, G., Lundgren, M., Alves, P. M., & Peixoto, Schneider, T. A., Kramberger, P., Tover, A., Nöbauer, K., Razzazi‐ C. (2018). Clinical‐Grade oncolytic adenovirus purification Fazeli, E., & Jungbauer, A. (2016). Purification of HIV‐1 gag virus‐like using polysorbate 20 as an alternative for cell lysis. Current Gene particles and separation of other extracellular particles. Journal Therapy, 18(6), 366–374. https://doi.org/10.2174/15665232186 of Chromatography A, 1455, 93–101. https://doi.org/10.1016/j. 66181109141257 chroma.2016.05.053 Moleirinho, M. G., Silva, R. J. S., Alves, P. M., Carrondo, M. J. T., & Steppert, P., Burgstaller, D., Klausberger, M., Kramberger, P., Tover, A., Peixoto, C. (2019). Current challenges in biotherapeutic particles Berger, E., Nöbauer, K., Razzazi‐Fazeli, E., & Jungbauer, A. (2017). manufacturing. Expert Opinion on Biological Therapy, 20(5), 451–465. Separation of HIV‐1 gag virus‐like particles from vesicular particles https://doi.org/10.1080/14712598.2020.1693541 impurities by hydroxyl‐functionalized monoliths. Journal of Nooraei, S., Bahrulolum, H., Hoseini, Z. S., Katalani, C., Hajizade, A., Separation Science, 40(4), 979–990. https://doi.org/10.1002/jssc. Easton, A. J., & Ahmadian, G. (2021). Virus‐like particles: Prepara- 201600765 tion, immunogenicity and their roles as nanovaccines and drug Venereo‐Sanchez, A., Gilbert, R., Simoneau, M., Caron, A., Chahal, P., nanocarriers. Journal of Nanobiotechnology, 19(1), 59. https://doi. Chen, W., Ansorge, S., Li, X., Henry, O., & Kamen, A. (2016). org/10.1186/s12951-021-00806-7 Hemagglutinin and neuraminidase containing virus‐like particles Nweke, M. C., McCartney, R. G., & Bracewell, D. G. (2017). Mechanical produced in HEK‐293 suspension culture: An effective influenza characterisation of agarose‐based chromatography resins for bio- vaccine candidate. Vaccine, 34(29), 3371–3380. https://doi.org/10. pharmaceutical manufacture. Journal of Chromatography A, 1530, 1016/j.vaccine.2016.04.089 129–137. https://doi.org/10.1016/j.chroma.2017.11.038 Vicente, T., Roldão, A., Peixoto, C., Carrondo, M. J. T., & Alves, P. M. Peixoto, C., Sousa, M. F. Q., Silva, A. C., Carrondo, M. J. T., & Alves, P. M. (2011). Large‐scale production and purification of VLP‐based (2007). Downstream processing of triple layered rotavirus like vaccines. Journal of Invertebrate Pathology, 107, S42–S48. https:// particles. Journal of Biotechnology, 127(3), 452–461. https://doi. doi.org/10.1016/j.jip.2011.05.004 org/10.1016/j.jbiotec.2006.08.002 Zaveckas, M., Goda, K., Ziogiene, D., & Gedvilaite, A. (2018). Purification Pereira Aguilar, P., Reiter, K., Wetter, V., Steppert, P., Maresch, D., of recombinant trichodysplasia spinulosa–associated polyomavirus Ling, W. L., Satzer, P., & Jungbauer, A. (2020). Capture and VP1‐derived virus‐like particles using chromatographic techniques. purification of human immunodeficiency virus‐1 virus‐like particles: Journal of Chromatography B, 1090, 7–13. https://doi.org/10.1016/j. Convective media vs porous beads. Journal of Chromatography A, jchromb.2018.05.007 1627, 461378. https://doi.org/10.1016/j.chroma.2020.461378 Zhao, M., Vandersluis, M., Stout, J., Haupts, U., Sanders, M., & Pereira Aguilar, P., Schneider, T. A., Wetter, V., Maresch, D., Ling, W. L., Jacquemart, R. (2019). Affinity chromatography for vaccines Tover, A., Steppert, P., & Jungbauer, A. (2019). Polymer‐grafted manufacturing: Finally ready for prime time? Vaccine, 37(36), chromatography media for the purification of enveloped virus‐like 5491–5503. https://doi.org/10.1016/j.vaccine.2018.02.090 particles, exemplified with HIV‐1 gag VLP. Vaccine, 37(47), 7070–7080. https://doi.org/10.1016/j.vaccine.2019.07.001 SUPP ORTING INFO RM ATION Reiter, K., Aguilar, P. P., Wetter, V., Steppert, P., Tover, A., & Jungbauer, A. (2019). Separation of virus‐like particles and extracellular vesicles by Additional supporting information can be found online in the flow‐through and heparin affinity chromatography. Journal of Supporting Information section at the end of this article. Chromatography A, 1588, 77–84. https://doi.org/10.1016/j. chroma.2018.12.035 Reiter, K., Pereira Aguilar, P., Grammelhofer, D., Joseph, J., Steppert, P., & Jungbauer, A. (2020). Separation of influenza virus‐like particles How to cite this article: Lorenzo, E., Miranda, L., Gòdia, F., & from baculovirus by polymer‐grafted anion exchanger. Journal of Cervera, L. (2023). Downstream process design for Gag HIV‐1 Separation Science, 43(12), 2270–2278. https://doi.org/10.1002/ based virus‐like particles. Biotechnology and Bioengineering, jssc.201901215 Schwartz, S., Campbell, M., Nasioulas, G., Harrison, J., Felber, B. K., & 120, 2672–2684. https://doi.org/10.1002/bit.28419 Pavlakis, G. N. (1992). Mutational inactivation of an inhibitory