A novel method for recombinant mammalian‐expressed S‐HBsAg virus‐like particle production for assembly status analysis and improved anti‐HBs serology

Abstract

The Hepatitis B surface antigen (HBsAg) as the only lipid‐associated envelope protein of the Hepatitis B virus (HBV) acts as cellular attachment and entry mediator of HBV making it the main target of neutralizing antibodies to provide HBV immunity after infection or vaccination. Despite its central role in inducing protective immunity, there is however a surprising lack of comparative studies examining different HBsAgs and their ability to detect anti‐HBs antibodies. On the contrary, various time‐consuming complex HBsAg production protocols have been established, which result in structurally and functionally insufficiently characterized HBsAg. Here, we present an easy‐to‐perform, streamlined and robust method for recombinant S‐HBsAg virus‐like particle (VLP) production by transient expression in mammalian cells and purification from the cell lysate with the aim of displaying uniform antigenic epitopes on the surface to improve serological detection of anti‐HBs antibodies. We not only compare assembly status and particle composition by transmission electron microscopy and mass photometry of our S‐HBsAg and of commonly used HBsAg reference samples, but also assess their antigenic quality and functional suitability for anti‐HBs antibody detection to identify the best performing sample for serological screenings. While we found that serum‐isolated and recombinant HBsAg VLPs are assembled differently, our S‐HBsAg VLPs detected anti‐HBs antibodies with the highest sensitivity and specificity in multiplex serology when compared to yeast or serum HBsAg making it the most suitable antigen for analysis of HBV immunity through anti‐HBs serostatus.

1. INTRODUCTION

Despite the availability of a highly efficient vaccine, Hepatitis B virus (HBV) remains the major cause of acute and chronic liver disease with an estimated number of 300 million people suffering from chronic hepatitis and over 800,000 deaths in 2019 more than HIV, tuberculosis, and malaria combined (Stanaway et al. 2016).

HBV produces both mature and virus‐like particles (VLPs) as part of its replication cycle, but VLPs lacking genomic DNA are secreted in great excess (Hu and Liu 2017; Lamontagne et al. ²⁰¹⁶). The major component of these VLPs the Hepatitis B surface antigen (HBsAg) exists in three versions: small (S), medium (M), and large (L). While all HBsAg versions share the same C‐terminal part, the M‐ and L‐HBsAgs are extended at the N‐terminus by the preS2‐ or preS2 + preS1‐regions, respectively (Lamontagne et al. 2016). Subviral particles come in spherical and tubular shape, exhibit a diameter of approximately 22 nm (Ho et al. 2020; Liu et al. ²⁰²²; Seitz et al. ²⁰²⁰; Tsukuda and Watashi ²⁰²⁰), and contain predominantly S‐HBsAg which represents the minimum prerequisite for particle assembly (Cornberg et al. 2017; Dubois et al. ¹⁹⁸⁰; Patient et al. ²⁰⁰⁹). Although S‐HBsAg lacks the receptor‐interacting preS1‐sequence, it contains the immunogenic determinant “a” making it the major immunogen utilized in recombinant yeast‐derived second‐generation protein vaccines developed in the late 1980s (Di Lello et al. 2022). In addition, HBsAg is essential for diagnosis or serosurveillance to detect anti‐HBs antibodies, which indicate protective immunity after a resolved infection or vaccination.

However, despite this central role in inducing protective immunity, manufacturers rarely specify in detail the source, purity, and kind of HBsAg implemented in their anti‐HBs assays (ABBOTT, ^n.d, Gerlich ²⁰¹⁵) and HBsAg structures with sub‐nanometer and near‐atomic resolution were only recently published (Liu et al. 2022; Wang et al. ²⁰²⁴). Based on heterogeneity in size and geometry of native and recombinant spherical HBsAg VLPs, structural investigations are in general complicated (Venkatakrishnan and Zlotnick 2016). Therefore, prior moderate resolution structures between 12 and 30 Å in cryogenic electron microscopy (cryo‐EM) led to contradictory conclusions in regards to particle symmetry and lipid organization (Cao et al. 2019; Gilbert et al. ²⁰⁰⁵; Mulder et al. ²⁰¹²). Even the higher (6.3 and 3.7 Å) resolution structures exhibit such differences (Liu et al. 2022). The 6.3 Å resolution structure displayed rhombicuboctahedral symmetry, lipid organization in patches and showed that ~17 nm VLPs consist of 48 HBsAg monomers (Liu et al. 2022). In contrast, Wang et al. presented two stable VLP assembly symmetries (D₂‐ and D₄‐like) with a lipid bilayer, where 80 (D₂) or 96 (D₄) HBsAg monomers form the ~22 nm particles (Wang et al. 2024).

HBsAg has been recombinantly produced using most of the commonly available expression systems. The most frequently utilized expression hosts are yeast strains because of scalability and cost‐effectiveness (Diminsky et al. 1997; Gurramkonda et al. ²⁰¹³; Hardy et al. ²⁰⁰⁰; Valenzuela et al. ¹⁹⁸²). However, their inability to glycosylate HBsAg, assemble, or secrete VLPs are obvious drawbacks (Diminsky et al. 1997; Gurramkonda et al. ²⁰¹³). This in turn gave rise to a multitude of different purification and in vitro maturation protocols including many tedious steps to achieve VLPs (Gurramkonda et al. 2013; Wampler et al. ¹⁹⁸⁵) making it however ultimately possible to demonstrate that yeast‐expressed HBsAg VLPs assemble progressively during those purification steps with increasing homogeneity (Zahid et al. 2015). In particular, the treatment with highly concentrated thiocyanate salt buffers after purification results in relatively homogeneous VLPs (Gurramkonda et al. 2013; Wampler et al. ¹⁹⁸⁵; Zahid et al. ²⁰¹⁵; Zhao et al. ²⁰⁰⁶). Most HBsAg production protocols end at this point, generating VLPs with an immature surface. However, a subsequent aging step is hypothesized to assist formation and reorganization of disulfide bonds of surface epitopes and improve lipid fluidity (Zhao et al. 2006), as VLPs are also composed of host‐derived lipids (Mangold et al. 1997; Mangold and Streeck ¹⁹⁹³). Especially the incubation of HBsAg VLPs with a redox couple at 37°C was shown to improve antigenicity tremendously (Zhao et al. 2011).

While secondary structure features and orientations of HBsAg monomers have been resolved, the cytosolic and antigenic loops could only be modeled by AlphaFold‐guided predictions (Liu et al. 2022; Wang et al. ²⁰²⁴). Thus, higher resolution is needed to resolve the orientation of the side chains forming these loops. Especially the antigenic loop is of great importance and interest as it contains the major epitope (Carman et al. 1990). Further insight into the structure of this loop could aid the design of novel neutralizing antibodies and deepen the understanding of initial interaction between the virus and heparan sulfate proteoglycans (Schulze et al. 2007). In this work, we established a streamlined, easy‐to‐perform and reproducible protocol for the expression, purification and in vitro maturation of HBsAg VLPs to generate superior samples with uniform surface presentation of the antigenic loop to allow the analysis of VLP assembly and the application in serological anti‐HBs screenings. Through transient gene expression using HEK293‐6E cells, HBsAg is accumulated intracellularly. The protein is solubilized from the membranes and purified in a simple affinity chromatography step. While treatment of pure HBsAg with NH₄SCN results in VLP formation, using reduced and oxidized Glutathione (GSH/GSSG) at 37°C fully matures the surface epitopes on the VLPs. Mass photometry (MP) in combination with transmission electron microscopy (TEM) were utilized to evaluate the assembly of the produced HBsAg VLPs by mass and size. Comparison to a recombinant yeast HBsAg VLP and to one serum‐isolated sample revealed similar diameters of all particles while the mass composition between recombinant and native is distinctly different. Finally, by utilizing bead‐based multiplex technology, we compared our S‐HBsAg VLPs with serum‐ and yeast‐derived commercial HBsAg samples for their suitability to detect anti‐HBs antibodies from individual donors as well as from the international anti‐HBs immunoglobulin standard and anti‐HBs quality control samples produced for diagnostic laboratories. Therefore, we provide the only available direct comparative evaluation of different HBsAg samples in one assay format. Here, our S‐HBsAg performed best by generating the highest levels of sensitivity and specificity in the assay, making it the most suitable antigen for incorporation in serological screening assays for analysis of HBV immunity in population‐based public health research.

2. MATERIALS AND METHODS

2.1. Design of S‐HBsAg expression vector

pTT53 generated from a pTT5 vector (supplied by Yves Durocher, NRC) was used for transient gene expression in the suspension cell line HEK293‐6E (RRID:CVCL_HF20) (Durocher et al. 2002). Two oligonucleotides—pTT5 Polylinker (MCS) Forward: (5′‐GGATCCAGCT AGCACATCTA GAACACCTAG GTAACCGACC TCGACCTCTG GCT) and pTT5 Polylinker (MCS) Reverse: (5′‐GTTACCTAGG TGTTCTAGAT GTCTAGCTGG ATCCAAACTT GGACCTGGGA GTG) were designed to re‐introduce unique BamHI and AvrII restriction sites and to invert the polylinker of the pTT5 plasmid. The reverse and forward oligonucleotides with the sequence homology at the 5′ end and 3′ ends were designed to anneal at specific sites of the plasmid. The vector pTT5‐BA was finally created by using a SLIC reaction after PCR amplification of the new backbone (Li and Elledge 2012). Furthermore, the Bsu36I and SpeI sites outside the multiple cloning site (MCS) of the pTT5‐BA backbone were removed by restriction endonucleases and fill‐in using the Klenow fragment of the DNA‐Polymerase (Cat: M0210S, New England Biolabs). To insert the gene encoding the small version (S) of HBsAg_ayr subtype (genotype C, AA 175–400, GeneBank: X04615.1), a SLIC reaction was performed to add a Strep‐Tag II to the N‐terminus of the HBsAg gene and to remove the original start codon by using the following forward and reverse oligonucleotides: (N7_STII_SHBs_fw: 5′‐CCCAGGTCCA AGTTTGGATC CATGTGGAGC CACCCGCAGT TCGAGAAAGC TAGCGAGAGC ACAACATCAG GATTCCTGG; N7_SHBs_rev: 5′‐ACCTGAGGTC TTACCTAGGG GTACCTCATT AAATGTATAC CCA). For expression of cytosolic GFP as a transfection efficiency indicator, the pTTo_eGFP vector (supplied by Yves Durocher, NRC) was used.

2.2. Cultivation and transfection of HEK293‐6E cells

FreeStyle F17 expression medium (Cat: A13835‐01, Gibco) supplemented with 7.5 mM Glutamine (Cat: 49450, Fluka BioChemika), 0.1% Pluronic (Cat: A12880500, PanReac AppliChem), and 25 μg/mL G418 Geneticin (Cat: 10131‐027, Gibco) was used to culture HEK293‐6E cells at 37°C, 5% CO₂ and 60–95 rpm, depending on flask size. Cultures were split to maintain cell densities of 0.4–4.0 × 10⁶ cells/mL. Transient transfection with polyethyleneimine for gene expression was performed as previously described with minor modifications (Karste et al. 2017). Briefly, 2 days prior to transfection, cultures were split and seeded at densities of 0.45–0.6 × 10⁶ cells/mL. On the day of transfection, the cultures were diluted to 2 × 10⁶ cells/mL in expression medium without G418 Geneticin. 1.125 μg plasmid DNA consisting of 95% target vector and 5% GFP expression vector were mixed with 2.8125 μg polyethyleneimine 40 K (PEI, Cat: 24765‐1, Polysciences Inc.) per mL of transfection culture and incubated at room temperature for 15 min before adding it to the prepared cultures. Transient S‐HBsAg expression was carried out for 48 h. After 24 and 48 h, before harvest, samples were taken to determine cell count and viability using a hemocytometer and to approximate transfection efficiency by measuring % of eGFP⁺‐fluorescent cells on a Guava EasyCyte HT‐BG flow cytometer (Luminex Corporation).

2.3. Extraction and purification of intracellular recombinant S‐HBsAg

Cell pellets of HEK293‐6E expression cultures were harvested by a 30 min centrifugation step at 3000g and then liquid nitrogen flash‐frozen for storage at −80°C until further use or alternatively for the direct extraction of S‐HBsAg. Extraction was performed by thoroughly suspending the cell pellets in Tris‐buffered saline pH 8.0 (TBS) containing 0.6% Tween20 (Cat: P1379‐500ML, Sigma‐Aldrich), 1–2 tablets protease inhibitor cocktail (Cat: 11836170001, Roche) per 50 mL volume and ~0.2 μg/mL in‐house produced DNAse. After a 30 min incubation on ice, the lysate was centrifuged at 30,000g for 30 min at 4°C in a Sorvall RC 6+ centrifuge (Thermo Scientific). Prior to affinity purification, biotin in the lysate was blocked for 15 min at room temperature with Biolock (Cat: 2‐0205‐250, iba) and then removed by centrifugation at 10,000g for 20 min. Biotin‐free extracts were subsequently filtered through a 0.2 μm Sartolab P20 Plus filter (Cat: 18068, Sartorius). For affinity purification, a Strep‐TactinXT 4Flow high capacity cartridge (Cat: 2‐5027‐001, iba) was connected to an Äkta Start FPLC system (Cytiva) to obtain pure, solubilized HBsAg. Elution fractions were collected and stored at 4°C until further processing.

2.4. Maturation of recombinant S‐HBsAg VLPs

After affinity chromatography, elution fractions were combined and concentrated to a range of around 0.5–1.5 mg/mL with a Vivaspin 20, 10 kDa MWCO (Cat: VS2002, Sartorius) centrifugal concentrator. Concentrated S‐HBsAg samples were then diluted approximately 20‐fold with phosphate‐buffered saline (PBS) pH 7.4 containing 3 M NH₄SCN in the same centrifugal concentrator. To yield a buffer‐exchanged, concentrated sample, the diafiltration step was repeated once, followed by a transfer into a micro‐reaction tube for a 16 h incubation at 4°C on a roller mixer (Stuart Bibby SRT6D). Subsequently, this sample was transferred back into a Vivaspin 20, 10 kDa MWCO centrifugal concentrator (Cat: VS2002, Sartorius) and the buffer was exchanged again as described above. However, a PBS pH 7.4 containing the redox couple 0.6 mM reduced Glutathione (GSH, Cat: G4251‐100G, Sigma‐Aldrich) and 0.3 mM oxidized Glutathione (GSSG, Cat: 151193, MP Biomedicals) was used here. After a 72 h incubation at 37°C and 500 rpm on a Thermomixer C with a ThermoTop lid (Eppendorf) matured S‐HBsAg was dialyzed against PBS pH 7.4 using a 2 mL 10 kDa MWCO Slide‐A‐Lyzer MINI Dialysis Device (Cat: 88404, Thermo Scientific). Recombinant HBsAg VLP samples were then either flash‐frozen in liquid nitrogen and stored at −80°C or directly used for further analysis.

2.5. SDS‐PAGE

Protein samples were mixed with 2× reducing gel‐loading buffer (100 mM Tris, pH 6.8; 4% (w/v) SDS; 0.2% (w/v) bromophenol blue; 10% (v/v) β‐mercaptoethanol; 20% (v/v) glycerol). For non‐reducing sample preparations, β‐mercaptoethanol was omitted from the gel‐loading buffer. Prior to mixing the sample with gel‐loading buffer, total protein concentration of each HBsAg‐containing sample was determined by measuring the 280 nm absorbance with a NanoDrop spectral photometer (Fisher Scientific) to equalize loading amounts. After incubation at 95°C for 5 min, samples were loaded on precast SDS‐PAGE gels (Cat: 4569036, any kD Mini‐PROTEAN, Bio‐Rad) and run at 200 V for 30 min. After careful washing with deionized water, protein bands were visualized by incubation with Coomassie‐based staining solution (InstantBlue, Cat: ISB1L, Abcam). Stained gels were scanned using a biostep ViewPix 700 scanner (Epson).

2.6. Mass photometry

Both recombinant HEK293‐derived S‐HBsAg and commercial HBsAg VLP samples were analyzed using a Refeyn TwoMP mass photometer (Refeyn Ltd.). Particle‐free PBS buffer (2 mM KH₂PO₄; 10 mM Na₂HPO₄; 2.7 mM KCl; 137 mM NaCl; pH 7.4) which was prepared by filtration through a 3000 MWCO centrifugal ultrafilter (Vivaspin® 6, Cat: VS0692, Sartorius) was used to dilute samples to a concentration between 100 pM and 100 nM and for blank measurements. Particle‐free PBS buffer was only used for mass photometry if a blank measurement on an in focus coverslip resulted in less than 70 binding events, then diluted sample was added to the buffer on the coverslip. Binding events in a range of 1500–3000 were recorded as a 60 s video in the AcquireMP software (Refeyn Ltd.) and analyzed using DiscoverMP software (Refeyn Ltd.). Data is presented as counts‐per‐contrast histograms, fitted as Gaussian distributions to find the peak contrast. Conversion of contrast to molecular weight of the VLPs was performed by standard calibration with NativeMark™ Unstained Protein Standard (Cat: LC0725, Invitrogen).

2.7. Negative‐stain transmission electron microscopy and image processing

HBsAg VLP samples with 30–50 μg/mL concentrations were pipetted as 30–50 μL droplets onto parafilm (Cat: PM999, Bemis), then a thin carbon film was floated on top to allow adhesion. Following a 1 min incubation, the carbon film was removed from the sample droplet using a 300‐mesh copper grid (Cat: G2300C, Plano GmbH). The grid was floated consecutively on two droplets of deionized water to wash off residual sample. Subsequently, the washed grid was placed on a droplet of 4% uranyl acetate (w/v) for 1 min. Residual staining solution was removed using a filter paper and dried by heat from a 60 W light bulb. Samples were visualized utilizing a Libra 120 transmission electron microscope (Zeiss) with calibrated magnifications. Adjustments of brightness and contrast as well as image acquisition were performed using the WinTEM software (Olympus). For estimation of VLP diameters on TEM images, ImageJ software (NIH) was utilized. First, pixels were converted into nm by measuring the scale bar, then the background was subtracted, contrast maximized, and brightness adjusted until VLPs were fully visible. A threshold including all particles was set and the projection areas of the particles were analyzed. To exclude aggregates, all analyzed areas had to be between 175 and 1000 nm² and circularity had to be between 0.30 and 1.00 to reject harshly deformed particles. Further analysis restrictions were that partial particles on edges were prohibited and that holes in eligible particles were excluded. Output areas were used for calculation of VLP diameters assuming circular shape. Resulting diameters were graphed as frequency distributions and fitted utilizing Gaussian function.

2.8. Samples for serological anti‐HBs assays

A total of 163 de‐identified pre‐existing samples from either routine diagnostic or a commercial source were used for serological anti‐HBs analysis. Their previous anti‐HBs serostatus was either determined with the Architect Anti‐HBs Assay (7C18 (ABBOTT, ^n.d)) or the Alinity i Anti‐HBs Assay (07P89 (ABBOTT, ^n.d)) following the manufacturers protocol. According to the manuals, both use HBsAg_ad and HBsAg_ay subtypes expressed in mouse cells from Escherichia coli DNA as antigen. Sample characteristics are summarized in Tables 1 and S1, Supporting Information. Ethical approval was granted from the Ethics Committees of Hannover Medical School (Nr. 10305_BO_K_2022) and the Christian‐Albrechts‐University Kiel (D523/22). Citrate plasmas were purchased from Biomex GmbH.

TABLE 1.

Samples for serological anti‐HBs analysis.

Reactivity (n, %)		Median age (IQR) a	Sex (n, %)
Non‐reactive	66 (40.49)	42 (28–58)	Female (18, 27.69)	Male (29, 43.08)	n.a. (19, 29.23)
Reactive	97 (59.51)	49 (34–57)	Female (40, 40.82)	Male (47, 47.96)	n.a. (11, 11.22)

Abbreviation: n.a., not available.

^a Not available total n = 30 (18.40%).

2.9. Multiplex‐based anti‐HBs serology

A bead‐based multiplex immunoassay was established to assess the functional capacity of the utilized HBsAgs to detect anti‐HBs antibodies in sera und plasma samples. Briefly, different HBsAg antigens were immobilized on spectrally distinct populations of MagPlex beads (Cat: MC10XXX‐01, Luminex Corporation) at the indicated concentrations either by EDC/s‐NHS coupling (Becker et al. 2021) or by Anteo coupling (Cat: A‐LMPAKMM‐10, Anteo Tech Reagents) following the manufacturers instruction (Junker et al. 2022). Prior to bead coupling, the total protein concentration of each HBsAg sample was determined by measuring the 280 nm absorbance with a NanoDrop spectral photometer (Fisher Scientific) to equalize coupling concentrations. A detailed description of the bead coupling procedure can be found in Method section in Data S1. All commercially available antigens are listed in Table 2. All dilutions steps were carried out in assay buffer (1 part Low Cross Buffer (Cat: 10500, Candor Bioscience) + 3 parts blocking buffer (1× PBS (Cat: BP399‐4, Fisher Scientific) + 1% (w/v) BSA Fraction V (Cat: T844.3, Carl Roth) + 0.05% (v/v) Tween20 (Cat: P1379, Sigma‐Aldrich)). The combined MagPlex beads were then incubated with samples at a final dilution of 1:200 in a 96‐well Half Area NBS Microplate (Cat: 3686, Corning). After a wash step on a BioTek 405 TS Washer (Agilent) to remove unbound antibodies, IgG was detected with R‐phycoerythrin labeled goat‐anti‐human IgG as secondary antibody (Table 3). To verify, sample or signal system was added to all wells, human IgG, or anti‐human F_c‐specific IgGs were EDC‐NHS‐coupled to beads, respectively (Table 3). After another wash step and bead resuspension, samples were measured on Luminex 200 instrument (Luminex Corporation) using the following settings: timeout 30–90 s, gate: 7500–16,500, reporter gain: standard PMT, 100 events. Raw median fluorescence intensity (MFI) values corrected by an individual bead‐specific blank value are shown. Negative MFI values after correction were set to 0 for display purposes and calculation. A valid plate run was defined by obtaining a minimum bead count per sample, having a sample and signal system control bead value within a defined range and pass plate‐by‐plate sample controls (see Method section in Data S1 for additional details).

TABLE 2.

Commercial HBsAg samples.

Antigen	Source	Product code	Manufacturer
Recombinant HBsAgadw	Saccharomyces cerevisiae	ab167754	abcam
Third International Standard for HBsAg (HBV genotype B4, HBsAg subtypes ayw1/adw2)	Human plasma	12/226	NIBSC
HBSAgadw1 24 kDa	Pichia pastoris	PR‐1402‐S	Jena Biosciences
HBsAg protein (Subtype ayw)	Saccharomyces cerevisiae	30R‐AH018	Fitzgerald Industries
HBsAg protein (Subtype ay)	Human blood	30‐1816	Fitzgerald Industries

TABLE 3.

Reagents used for anti‐HBs multiplex serology and ELISA.

Reagent	Product code (RRID)	Manufacturer	Assay concentration
QCRHBsQC1‐Anti HBs Quality Control	15/B680	NIBSC	1:200 (Multiplex) 1:100 (ELISA)
QCRHBsGQC1‐HBsAg Quality Control	17/B717	NIBSC	1:200 (Multiplex)
QCRHBsQC1‐Anti HBs Quality Control	20/B760	NIBSC	1:200 (Multiplex)
Anti‐HBs immunoglobulin (2nd International Standard)	07/164	NIBSC	200 mIU (ELISA)
Human IgG	I2511‐10mg (AB_1163604)	Sigma Aldrich	2 μg per 2.5 × 106 beads
AffiniPure Goat Anti‐Human IgG Fc	109‐005‐008 (AB_2337534)	Jackson ImmunoResearch Labs	2 μg per 2.5 × 106 beads
Peroxidase AffiniPure Goat Anti‐Human IgG (H + L)	109‐035‐003 (AB_2337577)	Jackson ImmunoResearch Labs	0.08 μg/mL (ELISA)
R‐Phycoerythrin AffiniPure F(ab′)₂ Fragment Goat Anti‐Human IgG	109‐116‐098 (AB_2337678)	Jackson ImmunoResearch Labs	5 μg/mL (Multiplex)

2.10. Anti‐HBs ELISA utilizing HBsAg_ayr HEK

In addition, ELISAs were performed to analyze stability of HBsAg_ayr HEK and to show suitability for a second assay format should Luminex infrastructure not be available. Briefly, 96‐well plates (U96 Polysorp Nunc‐Immuno, Cat: 475434, Thermo Scientific) were coated with 50 μL HBsAg_ayr‐HEK for 16 h at 4°C. To determine the antigenic quality of the produced HBsAg, a factor 11 serial dilution from 2.5 μM (3325 ng per well) to 1.4 pM (37.2 ng per well) was coated. For detection of anti‐HBs, 400 nM (532 ng per well) were used. After coating and every subsequent incubation period, plates were washed four times with 200 μL PBST buffer (PBS + 0.05% (v/v) Tween20) using a Hydroflex plate washer (Tecan). Next, plates were blocked for 24 h at 20°C in a rotary shaker by addition of 200 μL blocking buffer (PBST +1% (w/v) BSA Fraction V). Afterwards, either NIBSC reagents at the indicated concentration (Table 3) or a 1:10 dilution of sera with pre‐defined HBs serostatus were added in a 50 μL volume to the 96‐well plate and incubated for 2 h at 20°C, 300 rpm on a Thermomixer C (Eppendorf). Corresponding species‐specific horseradish peroxidase‐conjugated secondary antibodies were also diluted with blocking buffer at the indicated concentration (Table 3) and incubated for 1 h at 20°C, 300 rpm on a Thermomixer C (Eppendorf). Subsequently, 50 μL 1‐Step Ultra TMB‐ELISA Substrate Solution (Cat: 34029, Thermo Scientific) was added per well for 5 min at room temperature before stopping the reaction using 50 μL 2 M H₂SO₄ (Cat: 9896.1, Carl Roth). Results were read out utilizing the NanoQuant infinite M200Pro plate reader (Tecan) measuring absorbance with the standard setting of 25 flashes at either 450 nm alone (antigenic quality assessment) or at 450 and 630 nm reference wavelength (anti‐HBs ELISA).

2.11. Data analysis and visualization

All data was analyzed using Microsoft Office Excel 2016 (Microsoft) or GraphPad Prism (GraphPad Software), if not stated otherwise. ROC analysis to determine anti‐HBs assay sensitivity and specificity was performed in GraphPad Prism with the recommended method of Wilson/Brown. Cut‐offs were selected to create balanced levels of sensitivity and specificity. Graphs were generated in GraphPad Prism and schematics with BioRender.

3. RESULTS

3.1. Commercially available HBsAgs vary in their ability to detect anti‐HBs antibodies

Correctly assembled and immunogenic HBsAg VLPs are essential for accurate serological classification of anti‐HBs IgG status to determine protective immunity after HBV infection or vaccination (Kim et al. 2011). Likewise, correctly assembled HBsAg preparations in sufficient quantities are crucial for further structural analysis. However, manufacturers often do not detail their production or purification strategy and provide insufficient details about the performance of the HBsAg protein in their serological assays (ABBOTT, ^n.d, Gerlich ²⁰¹⁵). Consequentially, we assessed multiple currently commercially available HBsAgs (Table 2) for their functional capacity to detect anti‐HBs antibodies in 50 blood samples with pre‐defined HBs status. Three of the tested HBsAg preparations resulted in a poor classification of anti‐HBs serostatus, with only the serum‐ and yeast‐derived HBsAg from Fitzgerald Industries reaching 80% sensitivity and specificity levels (Table 4 and Figure S1a). However, yeast cells possess a different folding machinery than human cells and are unable to glycosylate S‐HBsAg (Diminsky et al. 1997), while serum‐derived HBsAgs results in strongly varying assay performance (Brenner et al. 2019; Filomena et al. ²⁰¹⁷). Further analysis of the selected HBsAgs with a dilution curve of the 2nd international anti‐HBs immunoglobulin standard and purity analysis by SDS‐PAGE reflected the varying performances seen with the blood samples collected from individual donors (Figures S1b and S2) making all of the tested HBsAgs not optimally suited for assembly analysis or for serological anti‐HBs testing. In addition, most reported strategies to produce recombinant HBsAg VLPs require considerable hands‐on time, involve multiple purification steps and lack a standardized maturation procedure (Gurramkonda et al. 2013), leading to vastly varying sample quality. Therefore, we established a streamlined and reproducible S‐HBsAg VLP production, purification and in vitro maturation protocol (Figure 1) to overcome limitations of current methods. The here presented protocol provides optimal material for functional testing in serological assays and could serve as starting point for future structural VLP analysis.

TABLE 4.

Suitability of multiple commercially available HBsAg samples for the detection of anti‐HBs antibodies.

Antigen	AUC	% sensitivity with 95% CI	% specificity with 95% CI
HBsAgadw Pichia pastoris (abcam)	0.66	60.00 (38.66–78.12)	60.00 (42.32–75.41)
HBsAgayw1/adw2 Plasma (NIBSC) a	0.54	50.00 (29.93–70.07)	50.00 (33.15–66.85)
HBSAgadw1 Yeast (Jena)	0.53	50.00 (29.93–70.07)	50.00 (33.15–66.85)
HBsAgay Serum (Fitzgerald)	0.92	80.00 (58.40–91.93)	80.00 (62.69–90.49)
HBsAgayw Yeast (Fitzgerald)	0.88	80.00 (58.40–91.93)	80.00 (62.69–90.49)

Note: Based on ROC analysis with 50 samples with 20 pre‐classified as HBs‐reactive.

^a Performed with 15 μg of HBsAg containing sample per 2.5 × 10⁶ beads as with 5 μg only three samples resulted in blank‐corrected MFI > 0.

FIGURE 1.

A novel method for producing recombinant HBsAg with a mammalian expression system. Schematic workflow of the established protocol (a–d) with corresponding results (e–i). (a) Plasmid‐based transient gene expression using the HEK293‐6E suspension cell line. (b) Solubilization of membrane‐embedded HBsAg_ayr HEK by detergent and purification utilizing the Strep‐Tag II affinity tag at the N‐terminus of the protein. (c) Particle formation facilitated by buffer exchange for PBS with high NH₄SCN concentration and incubation overnight. (d) Epitope maturation through buffer exchange to PBS containing the redox couple GSH/GSSG and incubation as indicated. (e) Affinity chromatogram of recombinantly produced, detergent‐solubilized HBsAg. (f) SDS‐PAGE gel of HBsAg_ayr HEK purified by affinity chromatography. Lanes: F1: Soluble cell extract. F2: PageRuler Plus Prestained Protein Ladder. F3: Flow through. F4 + 5: Wash fractions 1 + 2. F6‐15: Elution fractions 2–11. (g) TEM image of HBsAg_ayr HEK VLPs formed after NH₄SCN treatment. (h) TEM image of HBsAg_ayr HEK VLPs after epitope maturation and final dialysis to PBS. (i) SDS‐PAGE loaded with 10 μg matured HBsAg_ayr HEK VLPs, stained with Coomassie‐based gel stain. Lanes: I1: PageRuler Plus Prestained Protein Ladder. I2: HBsAg_ayr HEK VLPs reducing sample preparation. I3: HBsAg_ayr HEK VLPs non‐reducing sample preparation. *: HBsAg_ayr HEK. Arrow: glycosylated and unglycoslyated HBsAg_ayr HEK monomer bands. Graphic created with BioRender.com.

3.2. A novel streamlined method for recombinant S‐HBsAg VLP production, purification, and maturation

To establish such a novel protocol, we first introduced a short, 8 amino acid, N‐terminal Strep‐Tag II as it had been observed in plant‐expressed HBsAg that GFP fusion to the N‐terminus allowed better epitope recognition compared to a C‐terminal fusion (Huang and Mason 2004). For expression, we selected mammalian HEK293‐6E suspension cells because they possess a sophisticated folding machinery and are able to perform native‐like post‐translational modifications (PTMs). While others (Diminsky et al. 1997; Ma et al. ²⁰⁰⁸; Michel et al. ²⁰⁰⁷) were able to generate intracellular assembled or secreted VLPs, our expression cell line did not yield either, making an alternative approach necessary. Here, cell pellets were harvested from transfected cells and membrane‐embedded HBsAg was solubilized with Tween20‐containing cell lysis buffer (Figure 1a). Subsequent affinity purification based on Strep‐Tag II–Strep‐Tactin XT interactions resulted in highly pure S‐HBsAg (Figure 1b,e,f). Next, to facilitate particle formation, the buffer was exchanged for an extended incubation in 3M NH₄ SCN buffer (Figure 1c). High thiocyanate concentrations give rise to small amounts of NCS‐SCN species acting as redox helper to aid disulfide bond formation (Wampler et al. 1985) in the cysteine‐rich HBsAg (Mangold and Streeck 1993). Chaotropic properties of NH4⁺‐ and SCN⁻‐ions in high concentration likely cause a mild salting‐out effect and promote hydrophobic interactions between HBsAg monomers. Those conditions then create nucleation starts for VLP assembly while not precipitating the formed particles (Hofmeister 1888; Hyde et al. ²⁰¹⁷). Instead, VLP structures of varying size are the result as demonstrated by exemplary TEM images taken after this step of the protocol (Figure 1g). Finally, to aid epitope maturation by improving inter‐ and intramolecular disulfide bond formation (Zhao et al. 2006; Zhao et al. ²⁰¹¹), the buffer was exchanged for a solution containing the redox couple oxidized and reduced Glutathione (Figure 1d). Following a 72 h incubation at 37°C, resulting VLPs are homogeneous (Figure 1h) and pure and show only glycosylated and non‐glycosylated monomer and dimer bands and do not migrate into SDS‐PAGE gels if reducing agents are omitted during sample preparation (Figure 1i).

3.3. Biophysical characterization of HBsAg_ayr HEK particles shows expected size with high molecular weight but reveals different mass compositions of recombinant and native VLPs

Next, we investigated HBsAg_ayr HEK VLPs produced by our protocol using TEM and mass photometry (MP) to compare their assembly status through size and mass to the serologically best‐performing HBsAgs from Fitzgerald Laboratories (Table 4 and Figure S1). To estimate the size of these VLPs as diameters, multiple TEM images were taken and analyzed using ImageJ software. Representative images of yeast‐, serum‐, and HEK‐derived HBsAg are shown (Figure 2a). Resulting diameters were then plotted against their frequencies and fitted utilizing a Gaussian distribution to unveil a mean diameter of 20.7 nm for our HEK‐derived HBsAg, of 21.0 nm for the yeast‐derived recombinant sample and of 21.7 nm for the serum‐isolated native HBsAg (Figure 2b). Fittingly, those diameters are close to the HBsAg VLP diameters of approximately 22 nm described in the literature by others (Cheong et al. 2012; Cregg et al. ¹⁹⁸⁷; Ganem and Varmus ¹⁹⁸⁷; Gurramkonda et al. ²⁰¹³; Mulder et al. ²⁰¹²; Valenzuela et al. ¹⁹⁸²; Zhou et al. ²⁰⁰⁶). Noteworthy is that the native particles appear more monodisperse in size than the recombinant ones, which could be due to a more organized intracellular assembly process before secretion.

FIGURE 2.

Biophysical characterization of HBsAg VLPs. (a) Exemplary unprocessed TEM images of HBsAg_ay Serum, HBsAg_ayw Yeast, and HBsAg_ayr HEK are shown (left panel). TEM images with maximized contrast and subtracted background are shown in the middle panel. TEM images analyzed for their particle cross‐sectional areas (shown as outlines) are displayed in the right panel. To enable the estimation of VLP diameters, those areas were assumed to represent perfect circles and were separately given in nm² through calibration with imprinted measuring bars. Measuring bars (bottom left corner): 500 nm. (b) Size frequency distributions of VLP diameters after analysis of TEM images are shown for the three HBsAg samples. (c) Mass frequency distributions of VLP masses are shown for the three analyzed HBsAgs. Numbers above peaks show mean diameters/masses.

Assessing the mass of single VLPs was simplified by the use of MP and revealed clear differences in the assembly of recombinant and native VLPs (Figure 2c). Frequency distributions of both recombinant particles allowed for fitting of three Gaussian distributions with nearly identical mean masses of ~1.4, ~1.8, and ~2.1 MDa indicating that three stable assembly states are present in these preparations. Interestingly, production of low molecular weight VLP assembly seems to be favored by recombinant production protocols like our own, as the lightest state is the most prevalent, while the heaviest is the least produced. Serum‐isolated particles exhibit a mass signal that fits exactly one Gaussian distribution. This signal overlays with the highest mass peak of the recombinant VLPs, signifying that recombinantly produced HBsAg samples contain native‐like assembled VLPs to some extent.

3.4. HBsAg_ayr HEK detects anti‐HBs antibodies with the highest sensitivity and specificity

After assessing assembly status and performing a biophysical analysis of our HEK‐derived HBsAg in comparison to yeast‐ and serum‐derived varieties from Fitzgerald Laboratories, we continued with a functionality analysis utilizing multiplex‐based bead technology from Luminex. Our HBsAg showed highest sensitivity and specificity compared to the two other HBsAgs in a panel of 163 blood samples with pre‐defined anti‐HBs levels (Tables 1 and 5 and Figure 3a). As part of our validation strategy, we also compared different amounts of antigens and different bead coupling techniques as initially others had reported that they failed to set up a multiplex assay with commercially available HBsAgs from Fitzgerald (Brenner et al. 2019). Here, we did not identify any changes in sensitivity or specificity levels based on either covalent EDC‐NHS or coordination chemistry‐based Anteo coupling or antigen amount for all utilized HBsAgs (Table 5 and Figure S3). This improved performance of our antigen was also reflected in its performance when utilizing two NIBSC quality control samples (QCs) and three concentrations of the second international anti‐HBs immunoglobulin standard as part of a continued quality monitoring process (Table S2 and Figure S4). Overall, we found higher signal‐to‐noise levels between the used controls and more stable MFI levels for our antigen. When using the three HBsAgs in a concentration curve with the international anti‐HBs immunoglobulin standard, performance was more comparable even with antigen‐coupled beads stored up to 12 weeks at 4°C (Figure S5). As part of our comprehensive antigen validation, we further defined a limit of detection (LOD) equaling mean MFI + 3 standard deviations from blank measurements for each HBsAg (Table S2) and assessed parallelism of selected samples (Figure S6) which aided in identifying an appropriate dilution for sample screening. To further warrant the performance of our HBsAg HEK_ayr, we performed additional analytical steps such as stability of antigen‐coupled beads, freeze–thaw stability of samples, batch‐to‐batch comparison between HBsAg_ayr productions and how different storage conditions would impact antigenicity, composition, size, and mass frequency distribution (Figures S7–S9). Here, we identified comparable serological results between batches, good stability upon sample freeze–thaw, a stable performance in recognizing the NIBSC standard 07/164, and no alterations in size and mass distribution frequency even when stored at 37°C for an extended period of 6 weeks. Although, we aim to integrate our HBsAg antigen into a multiplex‐based multi‐analyte immunoassay able to simultaneously characterize humoral immunity towards all to date identified hepatitis viruses, we last also determined the ability of our antigen in a single‐analyte assay such as ELISA should a multiplex‐bead based infrastructure not be available. Here, we found a good correlation between both assay formats and comparable levels of sensitivity and specificity in the planar ELISA set up (Figure 3b,c).

TABLE 5.

Suitability of HEK‐, serum‐, and yeast‐derived HBsAg VLPs for the detection of anti‐HBs antibodies when using different coupling methods and antigen amounts.

Antigen (μg total protein per 2.5 × 106 beads)	AUC	Cutoff MFI	Sensitivity (%) a	95% CI	Specificity (%) a	95% CI	Correctly classified
							As HBs non‐reactive (of 66)	As HBs‐reactive (of 97)
5 μg HBsAgayr HEK (Anteo)	0.93	>98.56	87.63	79.61–92.78	87.88	77.86–93.73	58	85
15 μg HBsAgayr HEK (Anteo)	0.94	>96.47	87.63	79.61–92.78	87.88	77.86–93.73	58	85
5 μg HBsAgayr HEK (EDC‐NHS)	0.96	>92.75	87.63	79.61–92.78	87.88	77.86–93.73	58	85
15 μg HBsAgayr HEK (EDC‐NHS)	0.95	>97.00	86.46	78.20–91.91	86.57	76.40–92.77	57	84
5 μg HBsAgayw Yeast (Anteo)	0.89	>167.10	80.41	71.42–87.09	80.30	69.16–88.11	53	77
15 μg HBsAgayw Yeast (Anteo)	0.88	>215.00	79.38	70.29–86.24	78.79	67.49–86.92	52	77
5 μg HBsAgayw Yeast (EDC‐NHS)	0.89	>92.79	80.41	71.42–87.09	80.60	69.58–88.29	52	78
15 μg HBsAgayw Yeast (EDC‐NHS)	0.91	>94.42	81.44	72.56–87.93	81.82	70.85–89.28	55	79
5 μg HBsAgay Serum (Anteo)	0.92	>122.20	83.51	74.87–89.58	83.33	72.57–90.43	55	81
5 μg HBsAgay Serum (EDC‐NHS)	0.95	>130.30	85.57	77.22–91.20	86.36	76.07–92.66	57	83
15 μg HBsAgay Serum (EDC‐NHS)	0.95	>143.90	86.60	78.41–92.00	86.36	76.07–92.66	57	84

Note: Analysis was performed in 163 samples, measured in three independent repeats.

^a 1 replicate of a non‐reactive sample excluded as MFI value were out of signal system control range; 2 replicates of a second non‐reactive sample excluded as MFI value out of serum control range.

FIGURE 3.

Serological performance of HBsAg_ayr HEK. (a) Five microgram of sample containing HBsAg_ayr HEK were EDC‐NHS coupled to 2.5 × 10⁶ beads to measure MFI values of anti‐HBs non‐reactive (n = 66) and reactive (n = 97) samples to define the assay's sensitivity and specificity. (b) An ELISA was performed with 0.53 μg of sample containing HBsAg_ayr HEK per 96‐well in combination with 56 anti‐HBs non‐reactive and 87 reactive samples to define sensitivity and specificity in this assay format. Mean MFI or OD values signals of three independent repeats were plotted as Box‐Whisker plot with boxes representing the median, 25th and 75th percentiles, and whiskers illustrating maximum and minimum values. The red line indicates the cut‐off value after ROC analysis to achieve balanced levels of sensitivity and specificity (a, b). (c) Correlation analysis was carried out to compare HBsAg_ayr HEK performance coupled to magnetic beads to planar assay surface in 143 blood samples from (b). Spearman's ρ is shown in the upper left corner.

4. DISCUSSION

Here, we not only present a novel method to produce pure and authentic S‐HBsAg VLPs with little hands‐on time and a simple, standardized maturation procedure, but we also provide a comprehensive biophysical and functional characterization of the resulting product. While a major advantage of our protocol is the short 8 day turnaround time from transfection to final dialyzed sample with typical yields ranging between 0.5 and 1 mg/L production culture, we were however unexpectedly not able to detect and purify S‐HBsAg VLPs from cell culture supernatants. Most publications describing the production of intracellular assembled or secreted HBsAg VLPs in mammalian cell lines used adherent cells, while others did not specify the actual mode of cell cultivation (Diminsky et al. 1997; Ma et al. ²⁰⁰⁸; Michel et al. ²⁰⁰⁷). We speculate that even though the N‐terminal Strep‐tag II is one of the shortest tags available while still exhibiting high affinity levels for affinity purification, that it still hampered the intracellular assembly and/or secretion of particles. Equally possible could be that our specific HEK suspension cell line is missing an essential factor required for efficient S‐HBsAg VLP assembly and secretion. Interestingly, it was only recently shown that a HEK293F suspension cell line is able to assemble and secrete HBsAg VLPs, which are also indistinguishable from native VLPs present in infected sera (Wang et al. 2024). Because these authors do neither report production nor purification yields, we can only speculate that their use of a Flag‐tag with the high specificity anti‐DYKDDDDK interaction might be better suited to enrich minute amounts of HBsAg VLPs from extremely dilute cell culture supernatants. In this context, we have however shown that recombinant, in vitro formed S‐HBsAg VLPs exhibit a similar size to native particles as others before (Liu et al. 2022; Valenzuela et al. ¹⁹⁸²; Zahid et al. ²⁰¹⁵), but we report for the first time that three different and stable mass assembly states exists in such VLPs. Gilbert et al. reported that HBsAg VLPs could be formed by different spacial arrangements of the protein subunits resulting in divergent sizes (Gilbert et al. 2005). In contrast, the recently published structures of Liu, Hong et al. and Wang et al. clearly showed that particles with similar size but variable mass exist (Liu et al. 2022; Wang et al. ²⁰²⁴). Therefore, it is conceivable that multiple particle assembly states result from the in vitro assembly and maturation process. We now confirm that S‐HBsAg VLPs are able to assume more than one particle architecture. However, the mass of one assembly state of our recombinant S‐HBsAg VLPs equals the mass of native VLPs, signifying the possible presence of native‐like assembled VLPs in recombinant samples. We also show that our particles exhibit better antigenic properties than native ones making it therefore plausible that the antigenic region is properly displayed independent from whether the particles are assembled native‐like or not. Using material from such assembly analyses after further VLP separation should allow the identification of homogeneous, native‐like assembled recombinant VLPs suitable for structural studies by cryo‐EM. The major advantage of correctly assembled recombinant HBsAg particles for structural investigations is their composition of S‐HBsAg alone. Thus, they present a uniform particle surface required for high resolution in cryo‐EM single particle analysis (SPA), in which thousands of particle micrographs are superimposed to elucidate the structure. In comparison, native VLPs incorporate a considerable amount of M‐HBsAg and a few L‐HBsAg molecules, complicating SPA (Ho et al. 2020; Liu et al. ²⁰²²; Seitz et al. ²⁰²⁰). Thus, native‐like assembled recombinant HBsAg VLP samples should be prime candidates for further improvements of HBsAg VLP structures (Liu et al. 2022; Wang et al. ²⁰²⁴). Detailed insights into the structure of the exposed antigenic loop would aid in the design of novel neutralizing or diagnostic antibodies (Chiu et al. 2019). Additionally, it would provide a deeper understanding on the molecular basis underlying the initial interaction of the virus particle with heparansulfate proteoglycans on the surface of hepatocytes (Schulze et al. 2007). Thus, providing a novel tool to identify HBV cell entry inhibitors.

We also analyzed functional properties of our HBsAg_ayr HEK by determining its ability to differentiate anti‐HBs reactive from non‐reactive samples. While Brenner et al. failed to establish a bead‐based anti‐HBs serology with multiple HBsAgs antigens (among them a serum‐derived equally sourced from Fitzgerald Industries as ours) and two different sera reference panels, we succeeded in setting up such an assay. We were able to classify anti‐HBs serostatus with different HBsAgs from various sources, albeit with different levels of diagnostic accuracy. Brenner et al. attribute their failure to the stringent conditions when covalently coupling HBsAgs to the beads; consequentially, we also attempted a coordination chemistry‐based approach, but did not succeed in increasing levels of sensitivity and specificity above the 90% margin. Although our S‐HBsAg VLPs and yeast‐derived HBsAg VLPs from Fitzgerald Industries seem to be assembled similarly judged by size and mass distributions, ours is the best performing for discriminating serum samples for their anti‐HBs status. We attribute this improvement of our S‐HBsAg in serological assays to the exactly defined amino acid sequence of our construct, the very high purity of final product and the unique substantially optimized maturation procedure. A second advantage of using our S‐HBsAg antigen comes from its production in a human cell line resulting in near‐native glycosylation patterns and a similar fold to native HBsAg with an epitope presentation in a natural conformation. In comparison, the yeast‐derived HBsAg from Fitzgerald Industries is not glycosylated and was expressed and folded by the yeast folding machinery. A third advantage of producing recombinant S‐HBsAg VLP with our protocol is that it uses exactly defined sequence variants compared to the non‐revealed HBsAg_adw sequences of the yeast‐derived material or the serum material which is a pool of different ay‐serotypes. Furthermore, our method allows to produce other HBsAg serotypes to potentially examine immune escape potential of those (Lazarevic et al. 2019). Last, while a genuine comparative cost analysis is difficult due to substantial differences in many factors involved in the production process, our maturation protocol should also be applicable to every other recombinant intracellularly produced HBsAg making it a prototypic approach. It could however be necessary to find a more suitable detergent for efficient HBsAg solubilization from the production hosts' membranes (le Maire et al. ²⁰⁰⁰; Lichtenberg et al. ¹⁹⁸³).

Our study has two limitations. First, our collection of 163 blood samples to validate the selected HEK‐, serum‐, and yeast‐derived HBsAgs as antigen is smaller than the panels used to set up IVD‐certified anti‐HBs assays for clinical laboratories. While we consider the sample size a limitation, we could at the time of preparing the publication not source more samples. Yet, our approach positively covers multiple analytical aspects needed to assess an antigens performance such as sample freeze–thaw stability, batch‐to‐batch stability, and reproducibility and the potential to interpolate results with the international anti‐HBs immunoglobulin standard making it valuable for other (public health) laboratories establishing anti‐HBs multiplex immunoassays. Second, the relatively low yields from the production process compared to yeast production. However, our approach results in a sufficient quantity to screen hundreds of samples from epidemiological studies from an individual production batch.

Overall, we present a novel easily implementable method for recombinant S‐HBsAg VLP production compared to previous time‐consuming and frequently not reproducible protocols. By combining this novel method with a thorough characterization of the resulting product, we demonstrate that our S‐HBsAg VLP is the most suitable sample for VLP assembly analysis and incorporation in serological immunoassays.

AUTHOR CONTRIBUTIONS

Michael Lehky: Writing – original draft; formal analysis; conceptualization; investigation; project administration; writing – review and editing; visualization; methodology. Tashveen Moonian: Investigation; writing – review and editing; formal analysis; methodology. Tanja Michel: Investigation; writing – review and editing. Daniel Junker: Investigation; writing – review and editing. Mathias Müsken: Investigation; writing – review and editing; methodology; supervision. Julia Strömpl: Investigation; writing – review and editing; resources. Patrick Nübling: Resources; writing – review and editing; investigation. Franziska Neumann: Investigation; resources; writing – review and editing; formal analysis. Andi Krumbholz: Resources; writing – review and editing; investigation; formal analysis. Gérard Krause: Writing – review and editing; funding acquisition; resources. Nicole Schneiderhan‐Marra: Writing – review and editing; funding acquisition; resources. Joop van den Heuvel: Supervision; writing – review and editing; funding acquisition; formal analysis; conceptualization. Monika Strengert: Conceptualization; investigation; writing – original draft; supervision; resources; project administration; formal analysis; methodology; visualization.

CONFLICT OF INTEREST STATEMENT

NSM was a speaker at Luminex user meetings in the past. The Natural and Medical Sciences Institute at the University of Tübingen is involved in applied research projects as a fee for services with the Luminex Corporation. The other authors declare no conflicts of interest.

Supporting information

Table S1. Sample characteristics.

Table S2. QC chart with NIBSC reagents and three concentrations of the 2nd international anti‐HBs immunoglobulin standard.

Figure S1. Suitability of commercially available HBsAg samples as antigens for the detection of anti‐HBs antibodies.

Figure S2. SDS‐PAGE of commercially available HBsAgs under reducing and non‐reducing preparation conditions.

Figure S3. Capability of different concentrations of HEK‐, serum‐, and yeast‐derived HBsAgs coupled with the EDC‐NHS or Anteo method to classify anti‐HBs serostatus.

Figure S4. QC chart with NIBSC control reagents and three concentrations of the 2nd international anti‐HBs immunoglobulin standard 07/164.

Figure S5. Recognition of HEK‐, serum‐, and yeast‐derived HBsAg samples by the 2nd international anti‐HBs immunoglobulin standard and stability of antigen‐coupled beads over time.

Figure S6. Dilution linearity of selected serum samples with HEK‐, serum‐, and yeast‐derived HBsAg samples.

Figure S7. Analysis of antigenic stability of 5 HBsAg_ayr HEK production batches with pre‐characterized blood samples and the 2nd international anti‐HBs immunoglobulin standard.

Figure S8. HBsAg_ayr HEK stability after exposure to different storage temperatures.

Figure S9. Biophysical characterization and analysis of antigenic stability of HBsAg_ayr HEK batches after exposure to different storage temperatures.

Lehky M, Moonian T, Michel T, Junker D, Müsken M, Strömpl J, et al. A novel method for recombinant mammalian‐expressed S‐HBsAg virus‐like particle production for assembly status analysis and improved anti‐HBs serology. Protein Science. 2025;34(1):e5251. 10.1002/pro.5251

DATA AVAILABILITY STATEMENT

Data sets used in this study will be made available upon publication as Supporting Information files.