Scholarly article on topic 'Purification of HIV-1 gag virus-like particles and separation of other extracellular particles'

Purification of HIV-1 gag virus-like particles and separation of other extracellular particles Academic research paper on "Chemical sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Journal of Chromatography A
OECD Field of science
Keywords
{Vaccines / Chromatography / "Extracellular particles" / "Host cell DNA" / "Host cell protein" / "Density gradient centrifugation"}

Abstract of research paper on Chemical sciences, author of scientific article — Petra Steppert, Daniel Burgstaller, Miriam Klausberger, Eva Berger, Patricia Pereira Aguilar, et al.

Abstract Enveloped virus-like particles (VLPs) are increasingly used as vaccines and immunotherapeutics. Frequently, very time consuming density gradient centrifugation techniques are used for purification of VLPs. However, the progress towards optimized large-scale VLP production increased the demand for fast, cost efficient and scale able purification processes. We developed a chromatographic procedure for purification of HIV-1 gag VLPs produced in CHO cells. The clarified and filtered cell culture supernatant was directly processed on an anion-exchange monolith. The majority of host cell impurities passed through the column, whereas the VLPs were eluted by a linear or step salt gradient; the major fraction of DNA was eluted prior to VLPs and particles in the range of 100⿿200nm in diameter could be separated into two fractions. The earlier eluted fraction was enriched with extracellular particles associated to exosomes or microvesicles, whereas the late eluting fractions contained the majority of most pure HIV-1 gag VLPs. DNA content in the exosome-containing fraction could not be reduced by Benzonase treatment which indicated that the DNA was encapsulated. Many exosome markers were identified by proteomic analysis in this fraction. We present a laboratory method that could serve as a basis for rapid downstream processing of enveloped VLPs. Up to 2000 doses, each containing 1ÿ109 particles, could be processed with a 1mL monolith within 47min. The method compared to density gradient centrifugation has a 220-fold improvement in productivity.

Academic research paper on topic "Purification of HIV-1 gag virus-like particles and separation of other extracellular particles"

Accepted Manuscript

Title: Purification of HIV-1 gag virus-like particles and separation of other extra cellular particles

Author: Petra Steppert Daniel Burgstaller Miriam Klausberger Eva Berger Patricia Pereira Aguilar Tobias Schneider Petra Kramberger Andres Tover Katharina Nöbauer Ebrahim Razzazi-Fazeli Alois Jungbauer

PII: DOI:

Reference:

S0021-9673(16)30644-6

http://dx.doi.Org/doi:10.1016/j.chroma.2016.05.053 CHROMA 357580

To appear in:

Journal of Chromatography A

Received date: Revised date: Accepted date:

2-12-2015

4-5-2016

13-5-2016

Please cite this article as: Petra Steppert, Daniel Burgstaller, Miriam Klausberger, Eva Berger, Patricia Pereira Aguilar, Tobias Schneider, Petra Kramberger, Andres Tover, Katharina Nobauer, Ebrahim Razzazi-Fazeli, Alois Jungbauer, Purification of HIV-1 gag virus-like particles and separation of other extra cellular particles, Journal of Chromatography A http://dx.doi.org/10.1016/j.chroma.2016.05.053

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Purification of HIV-1 gag virus-like particles and separation of other

extra cellular particles

Petra Steppert1, Daniel Burgstaller1, Miriam Klausberger1, Eva Berger2, Patricia Pereira Aguilar1, Tobias Schneider2, Petra Kramberger3, Andres Tover4, Katharina Nobauer5, Ebrahim Razzazi-Fazeli5 and Alois Jungbauer1,2*alois.jungbauer@boku.ac.at

1 Department of Biotechnology, University of Natural Resources and Life Sciences Vienna, Vienna, Austria

2 ACIB GmbH, Muthgasse 11, Vienna, Austria

3 BIA Separations d.o.o., Ajdovscina, Slovenia

4 Icosagen AS, Tartumaa, Estonia

5 VetCore Facility for Research, University of Veterinary Medicine Vienna, Vienna Austria *Corresponding author: Muthgasse 18, 1190 Vienna, Austria. Fax: +431476546675

35 Highlights

36 • A method to separate VLP from extra cellular vesicles

37 • Extra cellular particles have similar size range as VLPs

38 • After separation detected by proteomics tools

39 • Method in principle suited for manufacturing of pandemic vaccine

62 Abstract

63 Enveloped virus-like particles (VLPs) are increasingly used as vaccines and

64 immunotherapeutics. Frequently, very time consuming density gradient centrifugation

65 techniques are used for purification of VLPs. However, the progress towards optimized large-

66 scale VLP production increased the demand for fast, cost efficient and scale able purification

67 processes. We developed a chromatographic procedure for purification of HIV-1 gag VLPs

68 produced in CHO cells. The clarified and filtered cell culture supernatant was directly

69 processed on an anion-exchange monolith. The majority of host cell impurities passed

70 through the column, whereas the VLPs were eluted by a linear or step salt gradient; the major

71 fraction of DNA was eluted prior to VLPs and particles in the range of 100-200 nm in

72 diameter could be separated into two fractions. The earlier eluted fraction was enriched with

73 extracellular particles associated to exosomes or microvesicles, whereas the late eluting

74 fractions contained the majority of most pure HIV-1 gag VLPs. DNA content in the

75 exosome-containing fraction could not be reduced by Benzonase treatment which indicated

76 that the DNA was encapsulated. Many exosome markers were identified by proteomic

77 analysis in this fraction. We present a laboratory method that could serve as a basis for rapid

78 downstream processing of enveloped VLPs. Up to 2000 doses, each containing 1 x 109

79 particles, could be processed with a 1 mL monolith within 47 min. The method compared to

80 density gradient centrifugation has a 220-fold improvement in productivity.

89 Keywords: vaccines; chromatography; extracellular particles; host cell DNA; host cell

90 protein; density gradient centrifugation

1. Introduction

Enveloped virus-like particles (VLPs) are promising candidates for vaccination, gene therapy, and cancer immunotherapy [1-7]. Purification of VLPs is in its infancy and used methods often have been adopted from virus purification procedures, mostly density gradient centrifugation and combinations of filtration and flow through chromatography [8-11]. It is often overlooked that enveloped VLPs and viruses are secreted together with extracellular vesicles such as microvesicles and exosomes [12, 13]. Such mixtures are difficult to separate and often the presence of microvesicles in VLP and virus preparations are ignored and their biological relevance is not understood [14, 15]. Monoliths are well suited for purification of large bio-particles, such as plasmid DNA, viruses and VLPs [16-22]. For plasmids, large scale processes have already been established [23]. Pretreatment of the feedstock before purification by such methods as multiple depth and membrane filtration steps and polishing of viruses and VLPs often accounts for not only reduction in yield but also contributes to the costs of the process [8, 24]. Thus, it is desirable to directly load the culture supernatant onto the chromatography column with minimum efforts of pretreatment.

In this study we focus on HIV-1 gag VLPs as a model system. These VLPs, using the structural protein gag of HIV-1, are produced in CHO cells. Such an expression system secretes the VLPs into the supernatant and high productivity can be obtained [25-28]. Thus such a system would be suited for production of pandemic vaccines or requirements for treatment of large patient population [3]. However, the studied system must be strictly considered as a model system.

The HIV-1 gag VLPs are composed of a lipid bilayer, which originate from the host cell, in our case the CHO cells. The gag polyprotein drives the self-assembly of spherical particles, buttressing the lipid bilayer, and once assembled, bud from the cell membrane [29]. The biophysical properties which are important for purification development are not exactly known. The physical size is in the range of 100-150 nm in diameter [29], with density around 1.18 g/cm3 [10]. It has been assumed that enveloped viruses have multiple positive and negative charges distributed on the surface [30]. This is the most important property to design an ion-exchange chromatography step either in flow through or binding and eluting mode. The processes for such particles must be developed in an empirical manner, because their properties are not fully known.

A further complication of purification of VLPs is the potential contamination of the feedstock with extracellular vesicles such as microvesicles and exosomes [12, 13]. These

125 particles usually range from 40 to 1000 nm in diameter although a lot of these particles have

126 a size between 40 -100 nm which is very close to enveloped VLPs [13]. The large vesicles

127 are easily separated during the clarification steps by centrifugation or filtration while the

128 vesicles of the same size as VLPs have to be separated during a chromatographic purification

129 step based on the particles structure. Studies with HIV viruses have shown that these vesicles

130 carry similar membrane proteins [15]. Budding processes for the virus and the extracellular

131 vesicles are similar, or at least have been shown to be for HIV [12, 15]. It is not clear to what

132 extent CHO budding of VLPs and extracellular vesicles follow the same route. It is well

133 know that CHO cells secrete such vesicles, because they have been considered for use as a

134 measure of the quality of the cell [31]. A mixture of VLPs or viruses and extracellular

135 vehicles is extremely difficult to separate [14]. In addition, the analysis and discrimination of

136 VLPs from extracellular vesicles is very challenging since they are similar in size and are in a

137 large part composed of the same structural proteins. Usually, analytical methods for

138 characterization of VLPs are based on identification of structural proteins and evaluation of

139 the particle number and particle size [32]. Evaluation of the particle number and size can be

140 done by nanoparticle tracking analysis (NTA). However, this method cannot distinguish

141 between different particle structures and is associated with a variability of measurements up

142 to 15-20%, which is still in the acceptable error range for accuracy of bioanalytical methods

143 recommended by the FDA [33, 34]. TEM is used as an orthogonal method for NTA, to

144 corroborate the particle size and as a visual proof for particle formation but does not provide

145 any information about the particle composition. Biochemical methods such as Western

146 blotting and mass spectrometry (MS) are used for detection of structural proteins and

147 peptides and suggests information about the particle composition, but also detect unstructured

148 protein impurities that are not assembled into particles. Impurities, such as dsDNA,

149 frequently packaged in extracellular vesicles [35] can be detected by fluorescent nucleic acid

150 stains in combination with endonuclease treatment procedures. Only the combination of

151 multiple complementary analytical methods can gather conclusions about the particle

152 structure [36].

153 VLP separation can be managed by a combination of flow through chromatography,

154 membrane chromatography, micro-/ultrafiltration steps, and size exclusion chromatography

155 [8, 37, 38]. We have focused on monoliths with channels of 2 ^m in diameter. In previous

156 work, we were able to purify baculovirus by monoliths. These rod-shape infective viruses

157 have a completely different physical shape compared to the enveloped VLPs. For VLPs, we

158 assumed a homogenous distribution of charged membrane proteins on the surface whereas

baculoviruses have a head region that accumulates virus protein gp64 and is distinct from the tail [17].

We aimed to directly load the cell free culture supernatant on a monolith column and the exclusive pretreatment, after cell and cell debris removal by centrifugation, should consist only of 0.8 ^m membrane filtration. Optional endonuclease pretreatment was tested to reduce the dsDNA content before chromatography. Binding and elution conditions must be found which allow removal of host cell impurities and concentrate VLPs to a particle number which would be equivalent to a vaccine. The pretreated culture supernatant was loaded and respective gradients were applied for elution. A combination of multiple complementary methods (NTA, TEM, Western blotting) were applied to detect VLPs. For discrimination between VLPs and other extracellular particles proteomic analysis was performed and potentially encapsulated dsDNA was detected after endonuclease treatment. We also wanted to benchmark our process with the most common method, density ultracentrifugation.

2. Material and methods

2.1 Chemicals

All chemicals were purchased from Merck (Darmstadt, Germany) or Sigma Aldrich (St. Louis, MO, USA).

2.2 Expression of HIV-1 gag VLPs

For production of VLPs based on HIV-1-gag protein, a pQMCF expression vector expressing HIV-1 gag protein under the control of the CMV promoter was constructed. For production CHOEBNALT85 cell line (Icosagen, Tartumaa, Estonia) grown at 37 °C was used. Cells (6x106) were transfected by electroporation with 1 ^g of VLP expression vector pQMCF-HIVgag. Forty eight h after transfection 700 ^g/mL of G418 Geneticin® (Gibco/Thermo Fischer, Waltham, MA) was added to the selected plasmid-containing cell population. After selection, on day 10, the temperature of the cell culture was shifted to 30 °C and fed with CHO CD Efficient FeedB (Thermo Fisher Scientific, Waltham, MA USA) for 9 days. Cell culture media was a mixture of CDCHO and 293SFM (Thermo Fisher Scientific, Waltham, MA USA) including HT Supplement and Glutamax (Thermo Fisher Scientific, Waltham, MA USA). Production of HIV-1 gag VLPs was confirmed by Western blot analysis detecting HIV-1 p24. After production of the HIV-1 gag VLPs, cells were removed by centrifugation (1000 x g, 30 min) and 0.01% NaN3, was added.

2.3 Density gradient centrifugation

The HIV-1 gag VLPs were pelleted from the VLP containing cell culture supernatant through a 20% (w/v) sucrose cushion at 77100 □ g for 2.5 hours at 4 °C by a Beckmann L8-80M ultracentrifuge using a SW41Ti rotor (Brea, CA, USA). The VLP-containing pellet was resuspended in PBS buffer and loaded onto a 20% to 60% (w/v) sucrose gradient and centrifuged at 93500 □ g for 17.5 hours at 4 °C. The gradient was fractionated from the top in 300 ^L aliquots at 22 °C. Density was determined gravimetrically (Sartorius, Gottingen, Germany) and absorbance at 280 nm was measured by NanoDropND-1000 (Thermo Fisher Scientific, Waltham, MA USA) at 22 °C.

2.4 Chromatographic experiments

2.4.1 Chromatographic equipment

Preliminary chromatographic experiments were conducted on an Agilent Series 1100 System (Agilent, Waldbronn, Germany) consisting of a well plate automatic liquid sampler (WP ALS) for injection, a degasser, a quaternary pump, and a diode array detector (DAD). The ChemStation for LC 3D systems (Rev. B. 04.03) software was used for data acquisition and control. UV absorbances were monitored at 280 and 260 nm simultaneously. Elution fractions were collected manually and pooled according to the chromatograms.

Chromatographic experiments on preparative scale were performed on an AKTA explorer 100 equipped with a P-960 sample-pump and fraction collector (Frac-950) (GE Healthcare, Uppsala, Sweden). For control and data acquisition Unicorn software 10.1 was used. Conductivity, pH, and absorbance at 280 and 260 nm were monitored simultaneously. Elution fractions of 1 mL were collected by fraction collector and pooled according to chromatogram.

2.4.2 Preliminary chromatographic experiments

Preliminary chromatographic experiments were performed on an analytical scale using quaternary amine (QA), diethylamine (DEAE) and sulfate (SO3) CIMacTM analytical monoliths (V= 0.1mL) (BIA Separations, Ajdovscina, Slovenia). For anion-exchange chromatography, 20 mM Tris and 50 mM HEPES buffer systems were tested over the pH range from 7.2 to 8.5 and for cation-exchange chromatography 20 mM phosphate buffer was used over the pH range from 6.0 to 8.0. Optional 150 mM NaCl was added to equilibration buffers (mobile phase A) for anion-exchange chromatography and all elution buffers (mobile phase B) contained 2 M NaCl. Analytical monoliths were equilibrated for 15 bed volumes

with appropriate equilibration buffer. Aliquots (100 ^L to 500 ^L) of HIV-1 gag VLP standard material obtained by density gradient centrifugation or clarified and 0.8 ^m filtered (Millex AA filter, Millipore Bedford, MA, USA) CHO supernatant were loaded onto the monolith and eluted by a linear gradient 0-50% B in 50 bed volumes followed by a regeneration step to 100% B for 30 bed volumes and sanitization with 1 M NaOH for 30 bed volumes. Flow rate during chromatography was 1 mL/min.

2.4.3 Preparative chromatographic purification of HIV gag VLPs

Preparative purifications of HIV-1 gag VLPs from clarified and 0.8 ^m filtered (Millex AA filter, Millipore Bedford, MA, USA) culture supernatant were conducted by anion-exchange with the 1 mL radial flow monoliths CIMmultus QA or CIMmultus DEAE (BIA Separations, Ajdovscina, Slovenia). Equilibration buffer (mobile phase A) was 50 mM HEPES, 100 mM NaCl pH 7.2 for linear gradient elution and 50 mM HEPES, 350 mM NaCl pH 7.2 for stepwise elution. A wash step of 20 bed volumes was introduced after loading and before linear gradient elution. For step gradient elution the wash step length was reduced to 15 bed volumes. Elution and regeneration was performed with 50 mM HEPES, 2 M NaCl pH 7.2 (mobile phase B). Linear gradients were conducted from 0-50% B in 50 bed volumes and stepwise elution was achieved by 0-25-45% B steps with a hold of 15 bed volumes each. After regeneration, sanitization was performed with 1 M NaOH. Equilibration was performed for 15 bed volumes, regeneration for 15 bed volumes and sanitization for 60 bed volumes. Flow rate during development of the purification process was 1 mL/min. After optimization of purification procedure the flow rate was adjusted to 5 mL/min.

2.5 Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and

Western blot analysis

NuPAGE Bis/Tris gels 4-12% (Invitrogen, Carlsbad, CA, USA) and MES-SDS running conditions were used to perform electrophoresis under reducing conditions in accordance to manufacturer's instructions. If required, samples were diluted in deionized water to obtain similar protein concentrations. SeeBlue® Plus2 Pre-stained Protein Standard (Invitrogen, Carlsbad, CA, USA) was used as the protein molecular weight ladder. Protein bands were stained by Coomassie Brilliant Blue G-250 based EZBlue™ Gel Staining Reagent (Sigma Aldrich, St. Louis, MO, USA). After SDS-PAGE, proteins were electroblotted onto a nitrocellulose membrane (Whatman, Dassel, Germany). The membrane was blocked with 3% BSA in PBS-T (0.1% w/v Tween-20 in PBS) for 2 h. Detection was performed by incubation

255 of primary mouse monoclonal antibody against HIV-1 p24 (Icosagen AS, Tartumaa,

256 Estonia), diluted 1:1000 in PBS-T containing 1% BSA for 2 h, followed by secondary

257 antibody incubation with anti-mouse IgG conjugated with alkaline phosphatase (Sigma

258 Aldrich, St. Louis, MO, USA), diluted 1:1000 in PBS-T with 1% BSA for 1 h. Visualization

259 was carried out by Lumi PhosTM (Thermo Fisher Scientific, Waltham, MA USA) on Lumi

260 Imager (Boehringer Ingelheim, Ingelheim, Germany).

261 2.6 Protein concentration and dsDNA content

262 Protein concentration was determined by Bradford assay using Coomassie blue G-250-

263 based protein dye reagent (Bio-Rad Laboratories, Hercules, CA, USA). The calibration curve

264 was obtained by bovine serum albumin (BSA) standards diluted in TE-Buffer. DNA content

265 was determined by Quant-iTTM PicoGreen® dsDNA kit (Life technologies, Waltham, MA,

266 USA). Protein and DNA assays were performed according to the particular manufacturer's

267 instructions in 96-well plate format. Signals were measured by Genius Pro plate reader

268 (Tecan, Mannedorf, Switzerland).

269 2.7 Nanoparticle tracking analysis (NTA)

270 For determination of particle concentration NTA measurements were performed by a

271 NanoSight LM-10 (Malvern Instruments Ltd., Worcestershire, UK) equipped with a blue

272 laser (405 nm). Samples were serially diluted in particle-free water to reach a suitable particle

273 concentration (60 to 100 particles per video frame) for analysis. Videos (60 s) of three

274 dilution steps for each sample were captured at room temperature. The videos were analyzed

275 and evaluated by NTA 2.0 software. Camera level was adjusted manually and optimized

276 analysis parameters were kept constant during all measurements. Particle number was

277 evaluated for particles with diameters between 100-200 nm.

278 2.8 Mass spectrometric analysis of proteins

279 Bands of interest were excised manually from Coomassie blue stained 1D-gels. After

280 washing and destaining [39] spots were reduced with dithiothreitol and alkylated with

281 iodoacetamide [40]. In-gel digestion was performed with trypsin (Trypsin Gold) with a final

282 trypsin concentration of 20 ng/^l in 50 mM aqueous ammonium bicarbonate and 5 mM

283 CaCl2. Digest proceeded for 8 hours at 37 °C [41]. Afterwards, peptides were extracted with

284 three changes of 30 ^L of 5% TFA in 50% aqueous CH3CN supported by ultrasonication for

285 10 min per change. Extracted peptides were dried down in a vacuum concentrator (Eppendorf

286 AG, Hamburg, Germany). De-salted peptides were dissolved in 10 (il 0.1% TFA and 1 ^l

was injected into the nano-HPLC Ultimate 3000 RSLC system (Thermo Fisher Scientific, Waltham, MA USA). Sample pre-concentration and desalting was accomplished with a 5 mm Acclaim PepMap ^-Precolumn (300 ^m inner diameter, 5 ^m particle size, and 100 A pore size) (Thermo Fisher Scientific, Waltham, MA USA). Separation was performed on a 25 cm Acclaim PepMap C18 column (75 ^m inner diameter, 3 ^m particle size, and 100 A pore size) with a flow rate of 300 nl/min. The gradient started with 4% B (80% ACN with 0.1% formic acid) and increased to 35% B in 90 min. It was followed by a washing step with 90% B. Mobile Phase A consisted of mQ H2O with 0.1% formic acid. For mass spectrometric analysis a Triple TOF 6600 instrument (Sciex, Framingham, MA, USA) was used. MS1 spectra were collected in the range 400-1500 m/z. The 20 most intense precursors with charge state 2-4 which exceeded 100 counts per second were selected for fragmentation, and MS2 spectra were collected in the range 100-1800 m/z for 150 ms. The precursor ions were dynamically excluded from reselection for 10 s. The nano-HPLC system was regulated by Chromeleon 8.8 (Thermo Fisher Scientific, Waltham, MA USA) and the MS by Analyst Software 1.7. Processed spectra were searched via the software Protein Pilot (Sciex, Framingham, MA, USA) in a UniProt database containing all proteins from Cricetulus griseus (identifier: 10029) as well as all proteins from Human immunodeficiency virus 1 (identifier: 11676) using the following search parameters: Global modification: Cysteine alkylation with iodoacetamide, Species: Mus musculus, Search effort: rapid, FDR analysis: Yes.

Proteins with more than 2 matching peptides at 95% confidence were selected.

2.9 Endonuclease treatment

Clarified and filtered cell culture supernatant was treated for 1 h with Benzonase purity grade II (Merck KgA, Darmstadt Germany) at a final concentration of 150 U/mL at 22 °C before 0.8 ^m filtration and chromatographic purification. For all analytical purposes, samples were diluted in Benzonase buffer (1 M Tris-HCl, 30 mM MgCl2, pH 8.0) before being treated with Benzonase at a final concentration of 150 U/mL at 37 °C for 1 h. For analytical purposes digestion was stopped by addition of 50 mM EDTA.

2.10 Transmission electron microscopy (TEM)

The samples were incubated for 1 min on 400-mesh copper grids, coated with Pioloform film and shaded with carbon. After fixation with 2.5% glutaraldehyde solution for 15 min and three wash steps with water samples were stained with 1% uranyl acetate solution

319 for 30 seconds followed by air drying step [42]. The negatively stained specimens were

320 analysed in a Tecnai G2 200 kV transmission electron microscope (FEI, Eindhoven, The

321 Netherlands), operating at 80 keV.

323 3 Results and discussion

324 VLPs were first purified by density gradient ultracentrifugation in order to produce a

325 reference standard and to obtain material for development of a chromatographic purification

326 method. This experiment also served to allow comparison of the new process with the well-

327 established density gradient ultracentrifugation method.

329 3.1 Purification of HIV-1 gag VLPs by density gradient centrifugation

330 Two high speed ultracentrifugation steps were performed to enrich HIV-1 gag VLPs

331 at densities between 1.16 and 1.18 g/cm3 (Figure 1 A) which are similar to values reported in

332 literature for HIV-1 VLPs [10]. Density gradient fractions were analyzed by Western blot

333 detecting HIV-1 gag specific p24, SDS-PAGE (Figure 1 B, Figure 1 C) and the presence of

334 VLP-like structures was confirmed by TEM (

Figure 2 A). Quantification of particles was made by NTA and showed that 6.6 x 1010 to 9.1 x 1010 part/mL were present in the pooled VLP fractions, resulting in yields between 22.8 to 48.0%.

3.2 Purification process

Screening experiments for development of a chromatographic purification process for HIV-1 gag VLPs were performed on an analytical scale. A strong (QA) and a weak (DEAE) anion-exchange as well as a strong cation-exchange (SO3) monoliths were tested in order to bind and elute HIV-1 gag VLPs. Purified standard material or clarified and filtered cell culture supernatant containing HIV-1 gag VLPs were injected. Different buffer systems with pH ranges from 6.0 to 8.5 were tested to elute HIV-1 gag VLPs by a linear gradient from 0 to 1000 mM NaCl. HIV-1 gag VLPs did not bind to the SO3 monolith, whereas QA and DEAE monoliths resulted in similar performances and a 50 mM HEPES buffer system at pH 7.2 was identified to provide optimal conditions for elution of VLPs between 560 and 770 mM NaCl (45-90 mS/cm). The complex surface structure of enveloped viruses and VLPs has not been well defined. However, a lot of enveloped viruses are negatively charged with isoelectric points between 1.9 and 8.4 and viruses with very basic isoelectric points have not been reported so far [30]. The phospholipid bilayer originated from the host cell membrane during the VLP budding processes and the associated membrane proteins contribute to the binding. The polar head groups of the most common phospholipids provide a negative net charge at physiological pH and are oriented outside, towards the membrane surface [43]. Thus, it is reasonable that the HIV-1 gag VLPs bind to anion-exchangers.

Purification experiments were scaled up to 1 mL radial flow QA and DEAE monoliths and clarified and filtered cell culture supernatant (50 mL) was directly loaded onto the monoliths and eluted by a linear gradient from 100 mM to 1000 mM NaCl at flow rates of 1 mL/min. A representative chromatogram obtained by QA monolith is presented in Figure 3 A. Both monoliths were tested and compared in terms of process recovery, yield, host cell (hc) protein, and dsDNA depletion. Results obtained from the strong anion-exchanger (Figure 3, Table 1 and Table 2) and weak anion-exchanger showed comparable purity (1.2 ^g total protein/109 particles and 15.0 ng dsDNA/109 particles purified by DEAE) and yield (20.8% for DEAE). Further process development was done with the strong anion-exchanger, because a higher operational stability can be expected, because pH-shifts are smaller in strong ion-exchangers [44]. Analysis of flow through fraction (FT) by Western blot detecting p24,

368 (Figure 3 C) showed that nearly all HIV-1 gag VLPs bound to the monolith which was

369 confirmed when measuring particle concentration by NTA (Table 1) and by representative

370 pictures generated by TEM (data not shown). About 50% of hc proteins did not bind and

371 consequently were present in the flow through fractions. VLPs were eluted over a broad

372 range of salt concentrations between 100 and 1000 mM (Figure 3, Table 1) where about 15%

373 of particles were eluted at the beginning of the gradient at low salt concentrations with

374 conductivities between 12 and 32 mS/cm. Majority of dsDNA (65.6%) was co-eluted with

375 about one third of particles at an intermediate salt concentration equivalent to 32 to 48

376 mS/cm but the majority of particles (43.5%) were eluted at the end of the gradient between

377 48 and 89 mS/cm.

378 All three particle-containing fractions were examined by TEM and the presence of

379 particles was confirmed but no remarkable visible differences between particles could be

380 observed (

Figure 2). However, different particle size distributions were measured by NTA (Figure 4), indicating that particles eluting in fraction P1 were slightly smaller than particles eluted in P2 and P3. Particles eluting in P3 were characterized by a broader particle size distribution compared to particles eluted in P2. Additionally, a different protein band pattern in SDS-PAGE (Figure 3 B) for these fractions was observed. With focus on the main band at 55 kDa representing the gag protein and a semi-quantitative evaluation of the band thickness, intermediate and late eluting fractions P2 and P3 were more dominant compared to P1. Furthermore, when the SDS-PAGE band profile of standard material obtained from density gradient centrifugation (Figure 1 B) was compared to those of chromatographic eluting fractions P1 to P3 (Figure 3 B) nearly no similarities were noted between P1 and the standard material, but several analogous components between P2, P3 and the standard material were identified. Particles eluting in P1 were composed of a variety of contaminant proteins, but gag protein (band at 55 kDa in Figure 3 B) was not the main structural component or, conceivably, a great number of free hc proteins were also aggregated with these particles. Analysis of significant protein bands (marked bands 1 to 15 in Figure 3 B) by MS (results are presented in the supplementary information) indicated that particles eluting in P3 consisted mainly of hc membrane-associated proteins (bands 12, 13 and 15 in Figure 3 B) and HIV-1 gag (band 14 in Figure 3 B). The hc proteins were part of the VLPs and were incorporated into the particle during the budding process. Proteome analysis of particles eluting in P2 showed that a large number of histones (bands 8 to 11 in Figure 3 B) and an increased number of exosome marker proteins (results are presented in the supplementary information) were present in this fraction. Histones are usually present in the cell nucleus to package DNA and should not be incorporated into correctly assembled VLPs. Therefore, we assume that we are eluting a portion of particles enriched with extracellular vesicles which partially carry encapsulated DNA. DNA digestion by Benzonase, done after the chromatographic method has been converted to stepwise elution, showed that dsDNA content could be minimized by enzymatic treatment in the VLP-containing fraction to 0.9 ng dsDNA/109 particles but not in the fraction containing the DNA (32.4 ng dsDNA/109 particles). This observation supports the assumption that DNA must be encapsulated in these particles. It is known that CHO cells produce a larger number of exosomes [31] which bud simultaneously together with VLPs and carry encapsulated DNA [35].

An endonuclease treatment step with 150 U/mL of Benzonase for 1 h at room temperature was introduced before chromatography to optimize and improve DNA removal

and to avoid potential competitive binding of DNA on the AIEX surface. The dsDNA content of the cell culture supernatant was reduced 22.8 times (95.6%) before 0.8 ^m filtration and loading onto the anion-exchange monoliths. The elution profile of endonuclease treated supernatant changed (compare Figure 3 with Figure 5). According to our expectations, signals of intermediate eluting fraction were reduced because the majority of DNA had been already digested into oligonucleotide fragments before loading. Elution order of particles, particle yield and particle concentration of elution fractions were not significantly affected by enzymatic treatment of starting material (Table 1, Table 2). Changes are explained by different fraction sizes because of differently pooled fractions and were within the methodical error of measurement using NTA. The majority of HIV-1 gag VLPs (49.1%) were eluted at high salt concentrations but residual dsDNA content of main HIV-1 gag VLP fraction was 3.2 times reduced (Table 1, Table 2). The enzymatic pretreatment of the feedstock resulted in a HIV-1 gag VLP fraction with an increased purity compared to purification of non-treated supernatant (Table 2). However, an additional process step is required subsequently to remove the Benzonase. Furthermore, it is commonly known that Benzonase is a huge cost factor and usage of high concentrations during early process steps might increase the total process costs. We did not focus on the optimization of the endonuclease treatment step and arrived at a reasonable concentration of Benzoase for 1 hour incubation time. This short incubation time makes the process more robust and the whole process sequence can be performed within one working day. Benzonase seems expensive at first glance, but also labour cost and plant utilization are an important cost factor. When the process time is reduced the plant can be more efficiently utilized. Only a professional cost analysis using modelling tools such as BioSolve (Biopharm Services, Chesham, UK) or SuperProDesigner (Intelligen Inc., Scotch Plains, NJ, USA) could give information about the cost effectiveness of the process operated with or without endonuclease treatment.

Often stepwise elution facilitates purification processes, especially for use on a large scale. The equilibration buffer was adapted to contain 350 mM NaCl, the length of the wash step was reduced to 15 bed volumes, and elution was performed by 0-25-45% B steps with a hold volume of 15 bed volumes each with elution buffer containing 2 M NaCl. In Figure 6, step elution profile and corresponding SDS-PAGE and Western blot results are shown and mass balance is presented in Table 3. Robustness and reproducibility of the purification performance were demonstrated by three independent purification cycles with two different supernatant batches. The total process recovery of particles in all fractions were between 70.5 and >99.9%. Consistently, 50.3 ± 1.3% of hc proteins did not bind to the monolith and were

449 present in the flow through fractions and 33.6 ± 3.3% of hc proteins and about 14 ± 8.1% of

450 particles could be removed during the wash step. Particles eluting during the wash step,

451 initially eluted in P1 during the linear gradient elution (compare Figure 3 B, P1 with Figure 6

452 B, W), were either aggregated to free hc proteins or were not mainly assembled from the gag

453 protein (Figure 6 B, C). In peak one (P1, Figure 6) generated by 25% B an average of 80.8 ±

454 7.3% of dsDNA co-eluted with about one third of particles (32.7 ± 9.4%) mainly consisting

455 of extracellular vesicles. However, co-elution of a noteworthy portion of correctly assembled

456 HIV-1 gag VLPs cannot be excluded. Whereas, majority and most pure HIV-1 gag VLPs

457 eluted during the 45% B step resulting in a process yield of 41.9 ± 9.6% (P2, Figure 6 and

458 Table 2). In summary, an average over-all hc protein depletion of 90.9% and an average

459 dsDNA depletion of 98.2% was achieved.

460 3.3 Dynamic binding capacity (DBC)

461 DBC was determined by direct loading of the filtered culture supernatant (Figure 7).

462 After about 100 mL loading, the particles start to breakthrough and after 200 mL (equivalent

463 to a total particle load of 8.6 x 1012 part/mL QA) the breakthrough of particles reached the

464 starting concentration (4.3 x 1010 part/mL). This result could be observed by the increase of

465 UV signals and was confirmed by Western blot analysis (Figure 7 B). The early breakthrough

466 contained only a minimal fraction of the total load (0.1%). Slightly shifted to the

467 breakthrough of HIV-1 gag VLPs, the breakthrough of dsDNA was detected by Picogreen

468 assay. To detect if there was potential competitive binding between the VLPs and DNA,

469 which might reduce the capacity for VLPs, the DBC was again determined by loading

470 endonuclease pretreated and filtered culture supernatant. A comparable binding capacity of

471 1.6 x 1012 part/mL QA was determined, indicating that there is no competitive binding of

472 VLPs and DNA. This result supports the application of the purification process cope without

473 endonuclease pretreatment.

474 3.4Evaluation of the optimized purification process

475 Using the optimized method that we defined from DBC and elution conditions, we

476 performed confirmation runs where 160 mL filtered culture supernatant (equivalent to a total

477 particle number of 1.0 x 1013 part/mL QA) was loaded at flow rates of 1 mL/min and 5

478 mL/min. In both cases yield was about 20% (Table 2) and the main HIV-1 gag VLP fraction

479 contained a total number of 2.1 x 1012 and 2.2 x 1012 particles, respectively. The residual

480 dsDNA levels were about 20 to 40% above the limits required for a licensed vaccine (10 ng

dsDNA/dose), when we assume that one vaccination dose consist of 1.0 x 109 particles (Table 2). When a flow rate of 5 mL/min was applied 2075 doses were purified within 47 minutes. In comparison the process performed at 1 mL/min, yielded in 2219 doses within 4 hours (Table 2). With slight improvements or for example by implementation of an endonuclease treatment after chromatography it would be possible to reduce the DNA level to the values claimed by the authorities.

Conclusion

Monoliths serve as a method to tackle extremely challenging separation problems in the field of bionanoparticles. Enveloped VLPs are efficiently overproduced in CHO cells, but they are contaminated by hc proteins, hc DNA, and potentially by extracellular particles. This method is able to separate these main impurities including particles with different characteristic than HIV-1 gag VLPs and is in principle suited for purification of pandemic vaccines. The process at the current stage is a laboratory process but all process parameters have been developed, which are needed to scale up such a process. With a 1 mL monolith, at least 160 mL supernatant, equivalent to a total load of 1.0 x 1013 particles, were processed within less than one hour. In comparison, a maximum of 20 mL supernatant, equivalent to a total load of 9.3 x 1011 particles, could be purified by the existing density gradient centrifugation processes within 20 hours resulting in about one-tenth of doses (230) compared to monoliths. Currently 8 L monoliths are commercially available and a 40 L monolith has been presented as an industrial prototype. With an 8 L monolith, about 107 doses could be purified and with the 40 L prototype about 5 x 107 doses. Our work provides a direction for how a pandemic vaccine could be efficiently purified.

Acknowledgements

The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 312004. This work has been also partly supported by the Federal Ministry of Science, Research and Economy (BMWFW), the Federal Ministry of Traffic, Innovation and Technology (bmvit), the Styrian Business Promotion Agency SFG, the Standortagentur Tirol and ZIT - Technology Agency

Ía3 dij 7i fii Ig

513 of the City of Vienna through the COMET-Funding Program managed by the Austrian

514 Research Promotion Agency FFG. We thank Gerhard Sekot for assisting with TEM pictures.

517 References

[1] A.F. Rodrigues, H.R. Soares, MR. Guerreiro, P.M. Alves, A.S. Coroadinha,;1; Viral vaccines and their manufacturing cell substrates: New trends and designs in modern vaccinology, Biotechnology Journal, 10 (2015) 1329-1344.

[2] R.S. Felberbaum,;1; The baculovirus expression vector system: A commercial manufacturing platform for viral vaccines and gene therapy vectors, Biotechnology Journal, 10 (2015) 702-714.

[3] F. Krammer,;1; Emerging influenza viruses and the prospect of a universal influenza virus vaccine, Biotechnology Journal, 10 (2015) 690-701.

[4] F. Krammer, S. Nakowitsch, P. Messner, D. Palmberger, B. Ferko, R. Grabherr,;1; Swine-origin pandemic H1N1 influenza virus-like particles produced in insect cells induce hemagglutination inhibiting antibodies in BALB/c mice, Biotechnology Journal, 5 (2010) 1723.

[5] G.P. Pijlman,;1; Enveloped virus-like particles as vaccines against pathogenic arboviruses, Biotechnology Journal, 10 (2015) 659-670.

[6] T. Frensing,;1; Defective interfering viruses and their impact on vaccines and viral vectors, Biotechnology Journal, 10 (2015) 681-689.

[7] S.C. Lin, W.C. Liu, Y.F. Lin, Y.H. Huang, J.H. Liu, S.C. Wu,;1; Heterologous prime-boost immunization regimens using adenovirus vector and virus-like particles induce broadly neutralizing antibodies against H5N1 avian influenza viruses, Biotechnology Journal, 8 (2013)1315-1322.

[8] C.L. Effio, J. Hubbuch,;1; Next generation vaccines and vectors: Designing downstream processes for recombinant protein-based virus-like particles, Biotechnology Journal, 10 (2015) 715-727.

[9] P. Nestola, C. Peixoto, R.R.J.S. Silva, P.M. Alves, J.P.B. Mota, M.J.T. Carrondo,;1; Improved virus purification processes for vaccines and gene therapy, Biotechnology and Bioengineering, 112 (2015) 843-857.

[10] P.E. Cruz, C.C. Peixoto, K. Devos, J.L. Moreira, E. Saman, M.J.T. Carrondo,;1; Characterization and downstream processing of HIV-1 core and virus-like-particles produced in serum free medium, Enzyme and Microbial Technology, 26 (2000) 61-70.

BTl 7i fii Ig

600 601 602

[11] P. Pushko, P. Pumpens, E. Grens,;1; Development of Virus-Like Particle Technology from Small Highly Symmetric to Large Complex Virus-Like Particle Structures, Intervirology, 56 (2013) 141-165.

[12] S. Welsch, B. Müller, H.G. Kräusslich,;1; More than one door - Budding of enveloped viruses through cellular membranes, FEBS Letters, 581 (2007) 2089-2097.

[13] G. Raposo, W. Stoorvogel,;1; Extracellular vesicles: Exosomes, microvesicles, and friends, Journal of Cell Biology, 200 (2013) 373-383.

[14] J.W. Bess Jr, R.J. Gorelick, W.J. Bosche, L.E. Henderson, L.O. Arthur,;1; Microvesicles Are a Source of Contaminating Cellular Proteins Found in Purified HIV-1 Preparations, Virology, 230 (1997) 134-144.

[15] M. Hammarstedt, H. Garoff,;1; Passive and active inclusion of host proteins in human immunodeficiency virus type 1 Gag particles during budding at the plasma membrane, Journal of Virology, 78 (2004) 5686-5697.

[16] A. Podgornik, S. Yamamoto, M. Peterka, N.L. Krajnc,;1; Fast separation of large biomolecules using short monolithic columns, Journal of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences, 927 (2013) 80-89.

[17] P. Gerster, E.M. Kopecky, N. Hammerschmidt, M. Klausberger, F. Krammer, R. Grabherr, C. Mersich, L. Urbas, P. Kramberger, T. Paril, M. Schreiner, K. Nöbauer, E. Razzazi-Fazeli, A. Jungbauer;1;, Purification of infective baculoviruses by monoliths, Journal of Chromatography A, 1290 (2013) 36-45.

[18] A. Jungbauer, R. Hahn;1;, Polymethacrylate monoliths for preparative and industrial separation of biomolecular assemblies, Journal of Chromatography A, 1184 (2008) 62-79.

[19] C. Tarmann, A. Jungbauer,;1; Adsorption of plasmid DNA on anion exchange chromatography media, Journal of Separation Science, 31 (2008) 2605-2618.

[20] C.S. Burden, J. Jin, A. Podgornik, D.G. Bracewell,;1; A monolith purification process for virus-like particles from yeast homogenate, Journal of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences, 880 (2012) 82-89.

[21] L. Urbas, B.L. Jarc, M. Barut, M. Zochowska, J. Chroboczek, B. Pihlar, E. Szolajska,;1; Purification of recombinant adenovirus type 3 dodecahedric virus-like particles for biomedical applications using short monolithic columns, Journal of Chromatography A, 1218 (2011) 2451-2459.

[22] M. Zaveckas, S. Snipaitis, H. Pesliakas, J. Nainys, A. Gedvilaite,;1; Purification of recombinant virus-like particles of porcine circovirus type 2 capsid protein using ionexchange monolith chromatography, Journal of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences, 991 (2015) 21-28.

[23] J. Urthaler, R. Schlegl, A. Podgornik, A. Strancar, A. Jungbauer, R. Necina;1;, Application of monoliths for plasmid DNA purification: Development and transfer to production, Journal of Chromatography A, 1065 (2005) 93-106.

Ía3 dij 7i fii Ig

610 611 612

620 621 622

[24] M.W. Wolff, U. Reichl,;1; Downstream processing of cell culture-derived virus particles, Expert Review of Vaccines, 10 (2011) 1451-1475.

[25] L. Cervera, S. Gutiérrez-Granados, M. Martínez, J. Blanco, F. Godia, M.M. Segura,;1; Generation of HIV-1 Gag VLPs by transient transfection of HEK 293 suspension cell cultures using an optimized animal-derived component free medium, Journal of Biotechnology, 166 (2013) 152-165.

[26] C M. Thompson, E. Petiot, A. Mullick, M.G. Aucoin, O. Henry, A.A. Kamen,;1; Critical assessment of influenza VLP production in Sf9 and HEK293 expression systems, BMC Biotechnology, 15 (2015).

[27] W. Akahata, Z.Y. Yang, H. Andersen, S. Sun, H.A. Holdaway, W.P. Kong, M.G. Lewis, S. Higgs, M.G. Rossmann, S. Rao, G.J. Nabel,;1; A virus-like particle vaccine for epidemic Chikungunya virus protects nonhuman primates against infection, Nature Medicine, 16 (2010) 334-338.

[28] M. Li, P.T. Schmitt, Z. Li, T.S. McCrory, B. He, A.P. Schmitt,;1; Mumps virus matrix, fusion, and nucleocapsid proteins cooperate for efficient production of virus-like particles, Journal of Virology, 83 (2009) 7261-7272.

[29] L. Deml, C. Speth, M P. Dierich, H. Wolf, R. Wagner,;1; Recombinant HIV-1 Pr55 gag virus-like particles: Potent stimulators of innate and acquired immune responses, Molecular Immunology, 42 (2005) 259-277.

[30] B. Michen, T. Graule,;1; Isoelectric points of viruses, Journal of Applied Microbiology, 109 (2010) 388-397.

[31] A. Bedina Zavec, V. Kralj-Iglic, M. Brinc, T. Ficko Trcek,;1; Microvesicles as a new parameter for industrial bioprocess monitoring, Biotechnology Journal, 10 (2015).

[32] A. Zeltins,;1; Construction and characterization of virus-like particles: A review, Molecular Biotechnology, 53 (2013) 92-107.

[33] ;1;U.S. Department of Health and Human Services Food and Drug Administration Guidance for Industry, Bioanalytical Method Validation, in, Center for Drug Evaluation and Research (CDER) Center for Veterinary Medicine (CVM) 2001.

[34] P. Kramberger, M. Ciringer, A. Trancar, M. Peterka,;1; Evaluation of nanoparticle tracking analysis for total virus particle determination, Virology Journal, 9 (2012).

[35] B.K. Thakur, H. Zhang, A. Becker, I. Matei, Y. Huang, B. Costa-Silva, Y. Zheng, A. Hoshino, H. Brazier, J. Xiang, C. Williams, R. Rodriguez-Barrueco, J.M. Silva, W. Zhang, S. Hearn, O. Elemento, N. Paknejad, K. Manova-Todorova, K. Welte, J. Bromberg, H. Peinado, D. Lyden,;1; Double-stranded DNA in exosomes: A novel biomarker in cancer detection, Cell Research, 24 (2014) 766-769.

[36] L H. Lua, N.K. Connors, F. Sainsbury, Y.P. Chuan, N. Wibowo, A.P. Middelberg,;1; Bioengineering virus-like particles as vaccines, Biotechnol Bioeng, 111 (2014) 425-440.

BTl 7i fii nltf

1 ■ f

660 661 662

680 681

[37] T. Vicente, A. Roldâo, C. Peixoto, M.J.T. Carrondo, P.M. Alves,;1; Large-scale production and purification of VLP-based vaccines, Journal of Invertebrate Pathology, 107, Supplement (2011) S42-S48.

[38] T. Vicente, M.F.Q. Sousa, C. Peixoto, J.P.B. Mota, P.M. Alves, M.J.T. Carrondo,;1; Anion-exchange membrane chromatography for purification of rotavirus-like particles, Journal of Membrane Science, 311 (2008) 270-283.

[39] F. Gharahdaghi, C.R. Weinberg, DA. Meagher, B.S. Imai, S.M. Mische,;1; Mass spectrometric identification of proteins from silver-stained polyacrylamide gel: A method for the removal of silver ions to enhance sensitivity, ELECTROPHORESIS, 20 (1999) 601-605.

[40] C.R. Jiménez, L. Huang, Y. Qiu, A.L. Burlingame,;1; In-Gel Digestion of Proteins for MALDI-MS Fingerprint Mapping, in: Current Protocols in Protein Science, John Wiley & Sons, Inc., 2001.

[41] A. Shevchenko, M. Wilm, O. Vorm, M. Mann,;1; Mass spectrometric sequencing of proteins from silver-stained polyacrylamide gels, Analytical Chemistry, 68 (1996) 850-858.

[42] M.A. Hayat, S.E. Miller,;1; Negative Staining, McGraw-Hill, New York, 1990.

[43] L.K. Buehler,;1; Cell membranes, Graland Science, Taylor and Francis Group, 2016.

[44] T.M. Pabst, G. Carta;1;, pH transitions in cation exchange chromatographic columns containing weak acid groups, Journal of Chromatography A, 1142 (2007) 19-31.

686 Tables

Table 1 Mass balance of HIV-1 gag VLP purification with a 1 mL radial flow QA monolith by linear gradient elution. Loading material was 0.8 ^m filtered and optional endonuclease treated CHO cell culture supernatant. (Buffer A: 50 mM HEPES, 100 mM NaCl, pH 7.2; Buffer B: 50 mM HEPES, 2 M NaCl, pH 7.2; 0-50% B in 50 bed volumes).

Linear gradient elution

Volume (mL) Particles D 100 - 200 nm (part/mL) % Total protein (Hg/mL) dsDNA (ng/mL)

Load 50.0 5.7E+10 100.0 448 12988

FT 50.0 4.5E+08 0.8 232 43

Wash 20.0 2.9E+08 0.2 38 nd

P1 11.0 3.9E+10 15.1 229 46

P2 14.0 6.3E+10 30.7 241 30450

P3 20.0 6.2E+10 43.5 117 1281

Recovery 90.3

nd: not detectable

Benzonase treatment before linear gradient elution

Volume (mL) Particles D 100 - 200 nm (part/mL) % Total protein (Hg/mL) dsDNA (ng/mL)

Load 50.0 9.5E+10 100.0 379 717

FT 50.0 3.0E+07 0.0 190 46

Wash 20.0 4.9E+07 0.0 45 nd

P1 13.0 7.7E+10 20.9 353 201

P2 6.0 1.8E+10 2.2 164 2136

P3 13.0 2.1E+10 5.8 59 1961

P4 20.0 1.2E+11 49.1 131 755

Recovery 78.2

nd: not detectable

Table 2 Comparison of purification strategies and characterisation of purified HIV-1 gag VLPs.

Loading volume Flow rate Yield Total protein /109 part. dsDNA /109 part. Productivity

(mL) (mL/min) (%) (^g/1x109 part.) (ng/1x109 part.) (1X109 part./mL/min)

Linear gradient elution 50 1 43.5 1.9 20.6 8.3

Benzonase treatment before linear gradient elution 50 1 49.1 1.1 6.4 11.2

Step gradient elution (n=3) 50 1 41.9 ± 9.6 1.8 ± 0.3 10.8 ± 1.9 7.9

Step gradient elution 160 1 21.8 2.4 14.0 9.4

Step gradient elution 160 5 20.4 2.4 12.6 44.1

Density gradient centrifugation 10.2 - 48.1 1.6 2.6 0.2

698 Table 3 Mass balance of HIV-1 gag VLP purification with a 1 mL radial flow QA monolith

699 step gradient elution. (Buffer A: 50 mM HEPES, 350 mM NaCl, pH 7.2; Buffer B: 50 mM

700 HEPES, 2 M NaCl, pH 7.2; 0-25-45% B with a hold volume of 15 bed volumes).

Volume (mL) Particles D 100 - 200 nm (part/mL) % Total protein (MQ/mL) dsDNA (ng/mL)

Load 50.0 2.2E+10 100.0 349 10008

FT 50.0 2.3E+07 0.1 171 40

Wash 15.0 9.2E+09 12.5 371 65

P1 15.0 3.0E+10 40.2 175 24652

P2 15.0 4.8E+10 64.7 98 602

Recovery 117.5

Figure legends

Figure 1. Absorbance and density profile of sucrose gradient centrifugation of HIV-1 gag VLPs (A) and corresponding Coomassie stained SDS-PAGE (B) and Western blot analysis (C). 5 ^L of each sample were loaded. Western blot detection was performed using a primary mouse monoclonal antibody against HIV-1 p24 (1:1000) and anti-mouse IgG conjugated with alkaline phosphatase (1:1000) as secondary antibody. Visualization of bands was carried out by a chemiluminescent substrate detecting alkaline phosphatase. Lane numbers indicate fraction number in (A), M: molecular mass marker.

Figure 2. Transmission electron microscopy from HIV-1 gag VLPs purified by (A) sucrose density gradient centrifugation and anion-exchange chromatography eluted at low (B) intermediate (C) or high salt concentration (D). The scale bar corresponds to 500 nm in (A) and to 100 nm in (B) to (D).

Figure 3. (A) Chromatogram of HIV-1 gag VLP purification from 50 mL 0.8 ^m filtered CHO cell culture supernatant applied to a 1 mL QA radial flow monolith. Equilibration buffer was 50 mM HEPES, pH 7.2 and linear gradient elution was performed from 100 to 1000 mM NaCl in 50 bed volumes at a flowrate of 1 mL/min. (B) SDS-PAGE and (C) Western blot analysis of collected fractions. The amount of loaded protein per lane was 785 ± 23 ng. Western blot detection was performed using a primary mouse monoclonal antibody against HIV-1 p24 (1:1000) and anti-mouse IgG conjugated with alkaline phosphatase (1:1000) as secondary antibody. Visualization of bands was carried out by a chemiluminescent substrate detecting alkaline phosphatase. M: molecular mass marker, S: cell culture supernatant; L: load, filtered cell culture supernatant; FT: flow through; W: wash; P1-P3: fractions of eluting peaks; R: regenerate. Marked bands in (B) were analysed by MS.

Figure 4. Particle size distribution measured by NTA. Fractions, corresponding to Figure 3, were eluted form 1 mL QA radial flow monolith by a linear gradient from 100 to 1000 mM NaCl in 50 bed volumes.

Figure 5. (A) Chromatogram of HIV-1 gag VLP purification from 50 mL endonuclease treated and 0.8 ^m filtered CHO cell culture supernatant by 1 mL QA radial flow monolith.

735 Equilibration buffer was 50 mM HEPES, pH 7.2 and linear gradient elution was performed

736 from 100 to 1000 mM NaCl in 50 bed volumes at a flowrate of 1 mL/min. (B) SDS-PAGE

737 and (C) Western blot analysis of collected fractions. The amount of loaded protein per lane

738 was 643 ± 33 ng. Western blot detection was performed using a primary mouse monoclonal

739 antibody against HIV-1 p24 (1:1000) and anti-mouse IgG conjugated with alkaline

740 phosphatase (1:1000) as secondary antibody. Visualization of bands was carried out by a

741 chemiluminescent substrate detecting alkaline phosphatase. M: molecular mass marker; S:

742 cell culture supernatant; L1: endonuclease treated cell culture supernatant; L2: endonuclease

743 treated and filtered cell culture supernatant; P1-P4: fractions of eluting peaks; R: regenerate.

745 Figure 6. (A) Chromatogram of HIV-1 gag VLP purification eluted by a step gradient (0-25746 45% B with a hold volume of 15 bed volumes) at a flowrate of 1 mL/min. 50 mL of 0.8 ^m

747 filtered CHO cell culture were loaded. Equilibration buffer was 50 mM HEPES, 350 mM

748 NaCl, pH 7.2 and elution buffer 50 mM HEPES, 2M NaCl, pH 7.2. (B) SDS-PAGE and (C)

749 Western blot analysis of collected fractions. The amount of loaded protein per lane was 1234

750 ± 66 ng and 289 ng for the regenerate. Western blot detection was performed using a primary

751 mouse monoclonal antibody against HIV-1 p24 (1:1000) and anti-mouse IgG conjugated with

752 alkaline phosphatase (1:1000) as secondary antibody. Visualization of bands was carried out

753 by a chemiluminescent substrate detecting alkaline phosphatase. M: molecular mass marker,

754 S: cell culture supernatant; L: load, filtered cell culture supernatant; FT: flow through, W:

755 wash, P1-P2: fractions of eluting peaks, R: regenerate.

757 Figure 7. (A) Dynamic binding capacity of filtered CHO cell culture supernatant containing

758 HIV-1 gag VLPs. Flow through fractions (1-10) were measured by NTA to monitor the VLP

759 breakthrough and analysed by Picogreen assay to monitor dsDNA breakthrough. (B) Western

760 blot analysis of collected flow through fractions (5 ^L of fractions 1-10 in (A) were loaded).

761 Western blot detection was performed using a primary mouse monoclonal antibody against

762 HIV-1 p24 (1:1000) and anti-mouse IgG conjugated with alkaline phosphatase (1:1000) as

763 secondary antibody. Visualization of bands was carried out by a chemiluminescent substrate

764 for detecting alkaline phosphatase.

767 Figure 1

768 (A)

1.20 -

CD □

1.05 -

X Density @ 22°C X

—•— Absorbance @ 280 nm

/ \ * *

\ . , A X ,

770 (B)

Fraction number

- 0.06

- 0.05

o o c CO

0.03 £ o en

- 0.02

- 0.01

Figure 3 (A)

M 5 L FT W PI P2 P3 M P2 P3 fkDa) [kDa}

98 - 98

62 -49 - 62 " ^49 rita

38 « 38m

28 * 28|¿

17 -14 17 14

€ m 6

(55 kDa)

795 Figure 5

796 (A)

799 (B)

801 802

807 (A)

816 817

820 821 822

Figure 7 (A)