Introduction

Malaria is caused by Plasmodium parasites and one of these, Plasmodium falciparum, infects 16% of children aged 2–10 years in sub-Saharan Africa and is responsible for the most deaths worldwide (1). With some success, considerable up-scaling of malaria control efforts has been done in the last decade (2). However, novel treatments that reduce the transmission of malaria between humans and mosquitos are needed to eradicate malaria (3). The life cycle of the Plasmodium parasites is complex, encompassing both sexual development in the mosquito and asexual development in the human host. A vaccine that disrupts parasite development in both mosquitos and humans would be of great value for enhanced malaria control as it would have the potential to provide direct personal protection and at the same time reduce parasite transmission from human to mosquito.

A recently developed chimeric vaccine candidate, GMZ2, consists of conserved fragments of two asexual blood stage antigens: the glutamate-rich protein (GLURP) and the merozoite surface protein 3 (MSP3). Both components of GMZ2 have been identified as targets of naturally acquired immunity against malaria as multiple immune epidemiological studies have demonstrated a strong correlation between levels of cytophilic antibodies against MSP3 and GLURP as well as protection against clinical malaria (4–7). Subsequent in vitro studies have demonstrated that affinity-purified human IgG antibodies and human monoclonal antibodies (mAb) against GMZ2 and its components mediate parasite killing in in vitro by antibodydependent cellular inhibition (ADCI) assays (6,8,9). We have recently demonstrated that GMZ2 elicits a modest but robust efficacy against clinical malaria in a phase IIb clinical trial, where the vaccine efficacy was higher in older children (10). In GMZ2-vaccinated children, the incidence of malaria decreased with increasing vaccine-induced anti-GMZ2 immunoglobulin G (IgG) concentration.

Pfs48/45 is one of the most advanced transmission blocking vaccine (TBV) antigens in the clinical pipeline. Monoclonal antibodies against native Pfs48/45 (11) and anti-Pfs48/45 antibodies generated by vaccination (12–14) have the capacity to block transmission of the parasite to the mosquito in a standard membrane feeding assay (SMFA). Pfs48/45 comprises three domains, of which domains I, II, and III contain six, four, and six disulphide-bonding cysteines, respectively (15,16). Domain III constitutes a s48/45 domain, which has been found in several proteins both in Plasmodium and other Aconoidasida. An NMR structure of a single (6-Cys) s48/45 domain of Pf12, has been solved, in which the s48/45 domain adopts a β-sandwich fold, where two disulphide-bonds are formed between each β-sheet and one disulphide bond is formed outside this core structure (17).

The C-terminal region of Pfs48/45 containing domain II (4-Cys) and domain III (6-Cys), collectively referred to as the 10C region encompasses three of the four epitopes for transmission blocking antibodies present (16). Of these, epitope I is located in domain III and the epitopes IIb and III are located in domain II (16,18). For instance, the well-established transmission blocking antibody mAb45.1, binds epitope I in domain III of Pfs48/45. The ability of these domains to elicit transmission blocking antibodies is dependent on their ability to adopt a correct native tertiary structure based on the correct cysteine connectivity (17). The Pfs48/45-10C region has been difficult to produce in a recombinant form due to the well-known inherent difficulty of refolding complex cysteine-bonded protein regions and the fact that the correct connectivity of the 10C region, that is required for antibody recognition and a transmission blocking effect, has not been verified experimentally (19–21). In particular, while the disulphide-bond connectivity of domain III (6C) may be inferred from the available NMR structure of a related s48/45 domain, the cysteine-bonding of domain II is more unclear.

Despite the inherent complexity of the Pfs48/45, several attempts have been made to combine this antigen with P. falciparum asexual blood-stage antigens in a chimeric vaccine and interestingly, it has been shown that combining domain III (6C) or domains II and III (10C) of Pfs48/45 with GLURP.R0 enables expression in Lactococcus lactis (L. lactis) of the respective fusion proteins (R0.6C and R0.10C, respectively) at relatively high yields (12,14).

Here we describe the expression, production and characterization of a new chimeric vaccine candidate composed of GMZ2’ (a version of GMZ2 in which the MSP sequence was changed from MSP3212-380 to MSP3154-249) genetically fused to the 10C region of Pfs48/45.

The resulting GMZ2’.10C protein represents a novel candidate for a multi-stage vaccine, capable of circumventing the current limitations of transmission blocking malaria vaccines (TBMV) by also providing personal protection against clinical malaria. Correctly folded GMZ2’.10C was successfully produced at yields compatible with larger scale pharmaceutical production. The purity, structure and storage stability of the produced GMZ2’.10C was assessed to validate the pharmaceutical potential of the fusion protein. GMZ2’.10C production was achieved by optimizing two approaches for refolding and purification of correctly-folded versions of the 10C region of GMZ2’.10C, following recombinant expression in L. lactis. One purification workflow employed immunoaffinity purification with an antibody specific to a correctly folded epitope of Pfs48/45. The other approach was based on a classical purification workflow circumventing the need for immunoaffinity purification, making this approach inherently more applicable for vaccine production. We furthermore map for the first time, the disulphide-bond connectivity in both domain II and III of Pfs48/45 as contained in the produced GMZ2’.10C protein and the reference proteins R0.6C and R0.10C. Our results thus provide new insights into the native cysteine connectivity of the most vaccine-relevant s48/45 epitopes of the Pfs48/45 protein to guide further TBV development.

Materials and Methods

Construction of Plasmids

Pfs48/45 477-1284 (10C) was amplified from pLEA5 (14) with primers 5′-CAATGGATCCGATAATACTGAAAAGGTTATATCAAGTATAGAAGG -3′ and 5′-CCATAGATCTTGCTGAATCTATAGTAACTGTCATATAAGC-3. Amplified PCR product was digested with BamHI and BglII (underlined) and cloned in the frame into pSS3 (22) in the BglII site to generate the GMZ2’.10C fusion construct. The expression vector is based on the L. lactis P170 promoter, which is induced by lactate accumulation in the growth medium (23). The construct was verified by DNA sequencing and transformed into L. lactis MG1363 by electroporation as described (24) for expression of recombinant proteins with 6xHis tags. The clone form for the construct with the highest expression level was selected for making a research cell bank (RCB).

Fermentation and Protein Purification

Fermentation was performed as described previously (14). Briefly, L. lactis MG1363 containing the indicated plasmids was cultured in a 1 l fermentor (BioFlo Benchtop Fermentor, New Brunswick) in LAB medium (2% yeast extract, 0.05 mM FeSO4, 2.5 μM CaCl2, 2.65 μM MgCl2, 0.5 mM citric acid, 1.38 mM ammonium sulphate, 14.7 mM sodium-acetate, 20 mM KH2PO4 buffer pH 7.4, 5% glucose, 10 μg/ml erythromycin) (14) supplemented with 10 mM cysteine and 1 mM cystine. Culture supernatants were clarified by centrifugation and concentrated tenfold by buffer exchange to phosphate-buffered saline (PBS) pH 7.4 containing 10 mM imidazole on a QuixStand Benchtop system (Hollow fiber cartridge with cut-off at 50000 Da, surface area 650 cm2, GE Healthcare, Sweden) followed by filtration through a Durapore Membrane Filter (PVDF, 0.22 μm, Millipore) and kept at −20°C until used. GMZ2’.10C was captured by immobilized metal ion chromatography (IMAC) on a 5 ml HisTrap HP column (GE Healthcare, Sweden) and eluted in tris pH 8.0 buffer (55 mM tris, 21 mM NaCl and 500 mM imidazole). After IMAC chromatography, fractions containing GMZ2’.10C were pooled and diluted into refolding buffer pH 9.0 (55 mM tris, 21 mM NaCl, 500 mM L-Arg, 0.88 mM KCl, 1 mM EDTA, 1 mM cysteine, 1 mM cystine) at 200 μg/ml protein concentration, and incubated for 16 h at room temperature. The refolded GMZ2’.10C was further purified by strong anion ion exchange chromatography (IEC) on a 5 ml HiTrap Q HP column (GE Healthcare, Sweden). The fractions containing GMZ2’.10C with high percentage (>70%) of binding to mAb45.1 were pooled and termed “conventionally purified GMZ2’.10C” (CP-GMZ2’.10C). Immunoaffinity purification of GMZ2’.10C was performed as described earlier (14) with minor modifications. Briefly; the diafiltrated material was applied to a 5 ml HisTrap HP column (GE Healthcare, Sweden) and bound protein was eluted in tris buffer (55 mM tris, 21 mM NaCl, 500 mM imidazole). Fractions containing the desired recombinant protein were refolded in refolding buffer and protein was captured by IEC on a 5 ml HiTrap Q HP column (GE Healthcare, Sweden). Fractions containing mAb45.1 immune reactive protein were further purified on a 5 ml HiTrap NHS-activated HP column (GE Healthcare, Sweden) immobilized with rat mAb 85RF45.1 (termed 45.1) against epitope I of Pfs48/45. Fractions containing the desired protein were concentrated to >1 mg/ml against tris buffer (55 mM tris, 21 mM NaCl and 0.025% polysorbate 80) pH 8.0 on a Vivaspin 20 (30,000 MWCO) centrifugal concentrator (Vivascience, Germany) and termed “immunoaffinity purified GMZ2’.10C” (IP-GMZ2’.10C). All preparatory chromatography was performed using an ÄKTAxpress system controlled by UNICORN 5.11 software (GE Healthcare, Sweden). Protein concentrations were measured using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, MA, USA).

Protein Characterization

Analytical Size Exclusion Chromatography of Intact Proteins

Size exclusion high-performance liquid chromatography (SEC) of intact proteins under native conditions was performed using an Agilent 1100 Series HPLC System (Agilent Technologies, CA, USA) equipped with a TSKgel G3000SWxl SEC column, 5 μm, 7.8x300 mm (Tosoh Bioscience, Japan). 160–210 pmol of protein was loaded on the column with a flow of elution buffer (200 mM phosphate, 0.65 g/l L-arginine (25), 0.05% NaN3, pH 6.7), set to 0.15 or 1.00 ml/min at room temperature. The absorbance was measured at 280 nm and analysed by HPLC ChemStation (Agilent Technologies, CA, USA).

Analytical Reversed-Phase HPLC of Intact Proteins

RP-HPLC of GMZ2’.10C was performed on an Agilent 1100 Series HPLC System (Agilent Technologies, CA, USA) equipped with a Vydac 214TP C4 Reversed-Phase Column, 5 μm, 4.6x250 mm (The Separations Group, CA, USA). 210 pmol of protein was loaded on the column and eluted with a linear gradient of 3–95% of 0.1% trifluoroacetic acid, 20% isopropanol and 70% acetonitrile over 30 min. The absorbance was measured at 214 nm and analysed by HPLC ChemStation (Agilent Technologies, CA, USA).

Peptide Map**

Tryptic digests of GMZ2’.10C were performed by drying down 250 pmol of protein using a Concentrator Plus (Eppendorf, Germany). 10 μl of 6 M guanidine hydrochloride, 50 mM NH4HCO3 (pH 8.00) was added and incubated for 30 min at 60°C. 5 μl of 45 mM dithiothreitol (DTT) was added and incubated for 3 h at 60°C. 5 μl of iodoacetamide (IAM) was added and incubated for 30 min at room temperature protected from light. 80 μl of 50 mM NH4HCO3 (pH 8.00) and 9.4 μl of 4 μM trypsin were added and incubated overnight at 37°C. The tryptic peptides were analysed by loading 100 μl of the tryptic peptide samples on a nanoACQUITY UPLC system (Waters, Milford, MA, USA) for desalting and chromatographic separation. Protein samples were desalted using a flow of 0.23% formic acid (Buffer A, pH 2.5) for 3 min at 2000 μl/min on a reversed-phase C18 pre-column (Acquity UPLC BEH C18 Vanguard, 1.7 μm, 2.1 x 5 mm - Waters, Milford, MA, USA). The samples were then eluted at 40 μl/min for 40 min with a chromatographic gradient of 8–40% of 0.23% formic acid in acetonitrile (Buffer B) over an analytical reversed-phase C18 column (Acquity UPLC BEH C18, 1.7 μm, 10 x 100 mm – Waters, Milford, MA, USA) and ionized by positive-mode electrospray using a Z-spray ESI source (Waters, Wilmslow, UK) mounted on a hybrid Q-TOF mass spectrometer (Synapt G2 HDMS mass spectrometer - Waters, Wilmslow, UK). MSE data was acquired in the MassLynx software (Waters, Wilmslow, UK) with a scan time of 0.3 s in the range of 50–2000 m/z. A reference lock-spray signal of [Glu1]-Fibrinopeptide B (Sigma, Schnelldorf, Germany) was acquired for internal calibration (m/z 785.8426, z = 2). The data files were loaded into ProteinLynx Global SERVER (PLGS) (Waters, Wilmslow, UK), and searched against a database containing the sequence of the protein and the protease trypsin. PLGS results were imported into DynamX (Waters, Wilmslow, UK) and lock mass corrected before creating a peptide map. Peptide exclusion criteria included mass deviation >10 ppm, <2 fragments per peptide, and <0.2 fragment ions per amino acid.

Map** of cysteine connectivity

Map** of cysteine connectivity was done with an identical UHPLC setup as for peptide map**. Tryptic digestion was done with or without prior reduction of cysteine bonds. LC-MS/MS data was acquired in a targeted manner for each individual precursor m/z range corresponding to the three identified cross-linked (XL) peptides. CID fragmentation was performed in the trap travelling wave ion guide by a collision energy ramp of 25–30 V. Scan time was set to 0.5 s in the range of 50–2000 m/z. The cysteine connectivity was identified by manually analysing MS/MS fragments using MassLynx software and by comparing the results to the known primary sequence of GMZ2’.10C.

Intact mass analysis

Intact mass analysis was analysed with a similar UHPLC setup, by loading 30 pmol of intact protein (100 μl sample of 300 mM phosphate buffer, pH 2.3) for trap** on a reversed-phase C4 pre-column (Acquity UPLC Protein BEH C4 Vanguard, 1.7 μm, 2.1 x 5 mm - Waters, Milford, MA, USA) and eluted directly onto the mass spectrometer by a chromatographic gradient of 6–97% Buffer B over 20 min at 40 μl/min. LC-MS data was recorded by MassLynx software with MS-only mode with a scan time of 1 s in the range of 300–2000 m/z. Deconvoluted mass spectra were produced by the MaxEnt 1 algorithm (Gs,0.100,750:1700,1.0,L33,R33) in the MassLynx software, after being processed with baseline subtraction (5,25) and Savitzky-Golay smoothing (1,15) parameters.

Stability Studies

Stability studies of GMZ2’.10C studies were performed over 30 days at 4°C. Samples were taken out after 0, 15 or 30 days at 4°C and analysed on the same day. Three cycles of freeze/thaw at −80°C was also performed and analysed. Analysis of each sample were performed by sodium dodecyl sulphate polyacrylamide gel electrophoresis with Coomassie staining (SDS-PAGE) and Western blotting with anti-GMZ2’ or mAb45.1 against the conformational epitope I of Pfs48/45 and by size exclusion high-performance liquid chromatography (SEC-HPLC). Using SEC-HPLC the stability of GMZ2’.10C at elevated temperatures (37 or 65°C) was additionally studied.

Immunizations

Groups of ten Wistar Hannover rats (Taconic, Denmark) were immunized with 10 μg of IP-GMZ2’.10C or CP-GMZ2’.10C in a volume of 100 μl by the subcutaneous route three times at 3-week intervals. All procedures regarding animal immunizations complied with European and National regulations. Protein preparations were emulsified in Montanide ISA 720 VG (Seppic, France) immediately before use. Responses were measured using sera taken three weeks after the third immunization (Day 63).

Enzyme-Linked Immunosorbent Assay (ELISA)

The ratio of GMZ2’.10C with correct disulphide connectivity/correct folding/percentage binding to mAb45.1 was determined by enzyme-linked immunosorbent assays (ELISA) as previously described (12,14). Gametocyte extract and recombinant antigens ELISA were performed as described elsewhere in details (14,26). The coating concentration of recombinant GLURP-R0, MSP3 or 6C was 0.5 μg/ml and bound antibodies were detected with HRP-conjugated rabbit-anti rat IgG-HRP (DAKO, Denmark).

Standard Membrane Feeding Assay

The biological activity of anti-GMZ2’.10C antibodies was assessed in the SMFA as previously described in details (14).

Statistical Analysis

All statistical analysis was conducted using GraphPad Prism 5, (GraphPad Software, USA). Antibody midpoint (EC50) values were defined as serum dilutions giving an absorbance higher than the average OD value at 450 nm of pre-immune sera. Data were analysed by a nonparametric test by comparing the medians of two groups using the Mann-Whitney test (27).

Results and Discussion

Production of GMZ2’.10C for Pharmaceutical Development

The multi-component multi-stage GMZ2’.10C fusion protein contains three P. falciparum antigens, including the asexual blood-stage antigens GLURP-R027–500 and MSP3154–249, called GMZ2’, together with the C-terminal 10C region of the sexual stage antigen Pfs48/45159–428 (Fig. 1a). L. lactis MG1363 expressing GMZ2’.10C was grown in LAB medium supplemented with a 10 mM cysteine and 1 mM cystine to enhance proper folding of the 10C region Pfs48/45.

Fig. 1
figure 1

Expression and purification of GMZ2’.10C. (a) Schematic representation of GMZ2’.10C (97.8 kDa) fusion protein, reference proteins used in disulphide-bond connectivity map** experiments, R0.10C (86.6 kDa) and R0.6C (72.2 kDa), as well as the Pfs48/45 antigen. Blue boxes illustrate protein antigen towards the asexual life stage of the malaria parasite, green boxes illustrate part of the protein from Pfs48/45, red boxes illustrate a 6xHis-tag and the black bars illustrates the presence of cysteine residues. (b) Overview of expression and purification methods used for GMZ2’.10C (c) SDS-PAGE analysis of CP-GMZ2’.10C and IP-GMZ2’.10C. Protein was loaded in each lane with (+) or without (−) reduction by DTT. The sizes (kDa) of the molecular mass markers are indicated. Coomassie Brilliant Blue stained gel (top). Immunoblot protein analysis of the same gel shown above using mAb45.1 (middle) or anti-GMZ2’ (bottom). (d) 2-site ELISA of CP-GMZ2’.10C and IP-GMZ2’.10C fusion protein, using mAb45.1.

Full size image

GMZ2’.10C was secreted into the culture supernatant of L. lactis from where it was expressed by immobilized metal ion affinity chromatography (IMAC) on a 5 ml HisTrap HP column. Since the amount of properly folded GMZ2’.10C was relatively low in the fermentation broth, IMAC-purified GMZ2’.10C was incubated for 16 h in refolding buffer. This refolding step was developed because the 10C region of GMZ2’.10C contains a total of 10 cysteines localized in two domains (II and III, Fig. 1a), and it is critical for the cysteine residues to be paired correctly in order for these domains to adopt the correct native fold. Folding conditions were carefully optimized to achieve the highest amount of correctly folded Pfs48/45 as determined by the reactivity with mAb45.1 which recognizes a conformational epitope in domain III (epitope I). Refolded GMZ2’.10C was subsequently purified by IEC. Eluate fractions from this polishing purification step were analysed by a 2-site ELISA assay, using mAb45.1 as the capturing antibody. Fractions that showed more than 70% reactivity in the mAb45.1 2-site ELISA were pooled and stored at −80°C. This batch (CP-GMZ2’.10C) represents GMZ2’.10C protein purified by a conventional approach that does not require immunoaffinity chromatography. In other words, this conventionally purified batch of GMZ2’.10C was produced using an exclusively generic purification approach highly compatible with large scale pharmaceutical production. The IEC fractions of GMZ2’.10C that displayed mAb45.1 reactivity below 70% was subjected to an immunoaffinity chromatography step, using a column immobilized with mAb45.1 (Fig. 1b). Eluted fractions corresponding to binding-capable GMZ2’.10C, were pooled and stored, representing the GMZ2’.10C protein batch purified using immunoaffinity chromatography (IP-GMZ2’.10C). 1 l fermentation gave approx. 5 and 3.5 mg CP- and IP-GMZ2’.10C, respectively (Table I). Importantly, on gels both CP- and IP-GMZ2’.10C contain a major band of monomeric protein (Fig. 1c upper panel), which reacted strongly with mAb45.1 (Fig. 1c middle panel) and anti-GMZ2’ antiserum (Fig. 1c lower panel). When tested in the mAb45.1 2-site ELISA, CP- and IP-GMZ2’.10C showed >70 and 100% properly folded Pfs48/45 conformers as compared to a reference protein (IP-R0.10C) (Fig. 1d and Table I). Although the present workflow for immunoaffinity purification of GMZ2’.10C is currently unrealistic due to the use of an uncharacterized mAb, we note that immunoaffinity purification of active ingredients has been previously used in the pharmaceutical industry for purification of recombinant blood-coagulation factors (28,29) and the production of a recombinant hepatitis B surface antigen from yeast (30). Thus, provided mAb45.1 can be produced under GMP it may be possible to apply a similar process in the clinical development of Pfs48/45-based protein-vaccine antigens. However, the successful results from the purification of GMZ2’.10C using only conventional (generic) chromatography (CP-GMZ2’.10C) makes this approach much more feasible for large scale pharmaceutical production.

Table I Purification of GMZ2’.10C. Protein Concentration and Folding Were Determined After Centrifugation (Supernatant), Immobilized Metal Ion Chromatography (IMAC) and Ion Exchange Chromatography (IEC) for CP-GMZ2’.10C). As well as Additional Immunoaffinity Chromatography with mAb45.1 for IP-GMZ2’.10C. Protein Concentration was Determined by BCA Protein Assay. The Folding (%) was Estimated by the mAb45.1 2-site ELISA Using IP-R0.10C as a Reference
Full size table

Characterization of Recombinant Produced GMZ2’.10C by Mass Spectrometry

Mass spectrometry (MS) offers a range of detailed protein analysis capabilities and has developed into a premier analytical tool in drug development science, including vaccine development (31–33). Mass spectrometry was used for accurate mass determination of the full-length fusion protein GMZ2’.10C from the two purified batches. Furthermore, tryptic digestion and LC-MS/MS was applied for peptide map** and for map** the connectivity of cysteines in the Pfs48/45 region of IP-GMZ2’.10C.

The theoretical molecular mass of full-length GMZ2’.10C is 97,849 Da as the recombinant protein contains the vector-encoded amino acid residues A-E-R-S at the N-terminal end, a 6xHis-tag at the C-terminal end and five disulphide-bonds. The experimentally determined molecular masses for both CP-GMZ2’.10C and IP-GMZ2’.10C were determined by LC-MS to be 97,844.9 Da (Fig. 2a) and 97,849.3 Da (Fig. 2b), respectively. This corresponded well to the theoretical mass, with an accuracy of 3 and 42 ppm, respectively. Overall the result indicates that the correct protein sequence has been expressed, with cysteines in the fusion protein engaging in disulphide-bonds. Furthermore, it shows that both proteins have been expressed and purified without introducing any unwanted covalent modifications to the primary structure. The difference in the molecular mass determined for IP-GMZ2’.10C and CP-GMZ2’.10C resulting from the two purification methods was <5 Da (<42 ppm), reflecting the error of the mass measurement. Incomplete formation of cysteine bonds would result in a 2 Da increase in mass per incompletely formed bond, and the apparent slightly lower mass (5 Da) for CP-GMZ2’.10C is therefore not due to incomplete formation of cysteine bonds in this protein sample. No distinct peaks were observed for any lower mass species, indicating that both IP-GMZ2’.10C and CP-GMZ2’.10C had been purified in their intact form and with no detectable degradation.

Fig. 2
figure 2

Mass spectrometry analysis of GMZ2’.10C. Intact mass analysis of full-length GMZ2’.10C: (a-b) Deconvoluted LC-MS spectra of full-length CP-GMZ2’.10C (3 ppm) and IP-GMZ2’.10C (42 ppm), respectively. In brackets is the calculated accuracy of the measurement to the theoretical mass. Measurements were done in duplicates. (c) To confirm the primary structure of IP-GMZ2’.10C, tryptic peptides were analysed by LC-MS/MS after reduction of disulphide bonds and digestion with trypsin, resulting in a peptide map with a sequence coverage of 88.3%. Analyses were performed in triplicate and only peptides identified in ≥2 replicates are shown. A peptide map for CP-GMZ2’.10C is shown in Supplementary Fig. 1.

Full size image

To further verify the primary structure of CP-GMZ2’.10C and IP-GMZ2’.10C, peptide map** analysis was performed by full reduction of disulphide bonds, followed by trypsin digestion with subsequent identification of each resulting peptide by LC-MS/MS analysis. Sequence coverages of 88.3% (Fig. 2c) and 86.2% (Supplementary Fig. 1) were obtained, for IP-GMZ2’.10C and CP-GMZ2’.10C, respectively. A comprehensive coverage of the tryptic peptide maps in combination with the correct intact mass, validates the identity and the correct primary structure of GMZ2’.10C from both purification methods.

RP-HPLC Analysis of GMZ2’.10C to Assess Sample Purity

To assess purity and the presence of degradation products in purified protein samples, reversed-phase HPLC was performed on both IP-GMZ2’.10C and CP-GMZ2’.10C (Fig. 3a and b, respectively). Intact GMZ2’.10C eluted at 18.5–18.7 min, with only few other lower intensity peaks observed in either purified batch. In the IP-GMZ2’.10C batch, there was a minor degree of overlap** species in the main form (Fig. 3b), indicating a homogenous population of species of GMZ2’.10C with identical disulphide-bond connectivity and a relative purity of 90.6%. For CP-GMZ2’.10C, a closely-related species appears as peak shoulders on the main CP-GMZ2’.10C peak (Fig. 3a), which could be due to host cell impurities or a minor component of CP-GMZ2’.10C with an incorrect disulphide-bonding and thus a more open structure with a larger hydrophobic foot (Z-value) for adsorption to the chromatographic stationary-phase. Overall, the results show that the two purification methods both give rise to GMZ2’.10C of the same primary structure, albeit at relatively higher purity in the case of the immunoaffinity approach.

Fig. 3
figure 3

Reversed-phase and size exclusion HPLC analysis of purified GMZ2’.10C. Reversed-phase HPLC-UV chromatograms was performed on full-length (a) CP-GMZ2’.10C (89.6%) and (b) IP-GMZ2’.10C (90.6%) to determine the sample purity (in brackets). The peak at 18.5–18.7 corresponds to the main form of GMZ2’.10C antigen. Size exclusion chromatography (SEC) was performed on full-length (c) CP-GMZ2’.10C (91%) and (d) IP-GMZ2’.10C (79%) at native conditions to determine the relative amount of the main form of GMZ2’.10C (RT = 41.4) and dimers of the main form (RT = 36.8–37.2) present in the samples. The calculated percentage main form to dimer is shown in brackets above. The integrated areas of the peaks are written above the peaks for both RP-HPLC and SEC-HPLC spectra.

Full size image

SEC-HPLC Analysis of GMZ2’.10C for Aggregation Analysis

The propensity of a therapeutic protein to aggregate will impact batch-to-batch variation and can ultimately cause a loss of therapeutic efficacy (34–36). Size exclusion chromatography represents a fast and robust technique for quantitative assessment of protein multimerization (37,38), and in this study the technique was applied to monitor aggregation in GMZ2’-10C samples and to provide a quantitate measure of multimerization as a function of storage time.

For SEC analysis of both purification methods, the main form of GMZ2’.10C eluted at 41.4 min (Fig. 3c-d). The molecular mass of GMZ2’.10C was estimated to 610 kDa (based on a calibration curve made using globular reference proteins), which is approximately six times higher than the molecular mass of monomeric GMZ2’.10C, possibly indicating that the main form of GMZ2’.10C found in solution is multimeric. However, it should be noted that the hydrodynamic radius during SEC using globular proteins for calibration does not correlate with the molecular weight of unstructured or partially unstructured proteins (39). It is likely that several domains of the chimeric GMZ2’.10C protein adopt or contain elongated unfolded structure, thus significantly increasing the hydrodynamic radius above that expected of a globular protein with the same molecular mass. Notably GMZ2’.10C migrated with an estimated molecular mass of around 130 kDa in the absence of DTT in SDS-PAGE and with a slightly larger mass in the presence of DTT (Fig. 1c). This shows that the large elution volume observed for GMZ2’.10C is not due to covalent multimerization through misformed cysteine bonds. Thus, we cannot accurately determine the precise oligomeric state of the main peak observed upon SEC of GMZ2’.10C.

Peaks at shorter elution times at 36.8–37.2 min correspond to a dimeric form of the main GMZ2’.10C form with twice the estimated molecular mass of the main peak. Control experiments where SEC-HPLC was performed under denaturing conditions (pH 2.7) showed only the main form at 41.4 min, indicating that the peak at 36.8–37.2 is a non-covalent dimer of the main form of GMZ2’.10C found in solution (data not shown). The relative abundance of the dimeric form of GMZ2’.10C was low (10–20%) and comparable between CP-GMZ2’.10C (Fig. 3c) and IP-GMZ2’.10C (Fig. 3d). The split nature of the peak corresponding to a dimer of GMZ2’.10C could indicate two different dimer forms arising from either differences in conformation or aggregation interfaces. Peaks at longer elution times at 55–60 min represent impurities of lower molecular weight including minor degradation products of GMZ2’.10C. However for both batches, the large majority of the protein is present in a single state in solution, and that the simpler conventional purification method does not increase the extent of aggregation/multimerization or the extent of lower mass species.

Stability Studies of GMZ2’.10C

To investigate the chemical and physical stability of produced GMZ2’.10C in terms of pharmaceutical suitability, stability studies were performed for GMZ2’.10C over three freeze/thaw cycles, and antigens from both purification methods were analysed by SEC-HPLC, SDS-PAGE, Western blotting with mAb45.1, and 2-site ELISA with mAb45.1. Importantly, all samples were stable after three freeze/thaw cycles with no or minor detectable degradation (Fig. 4d, Fig. 4f, Table II, and Supplementary Fig. 2).

Fig. 4
figure 4

Storage stability at 4°C and freeze-thaw stability of GMZ2’.10C measured by SEC-HPLC. Analytical size exclusion was used to assess the physical stability of GMZ2’.10C by measuring the amount of GMZ2’.10C present in the main form and dimers of the main form. CP-GMZ2’.10C was stored at 4°C for 0, 15 or 30 days, (a-c) respectively. IP-GMZ2’.10C was similarly stored at 4°C for 0, 15 or 30 days, (e-g) respectively). Furthermore, stability analyses after three freeze-thaw cycles were performed in triplicates for (d) CP-GMZ2’10C and (h) IP-GMZ2’.10C.

Full size image
Table II Physical Stability Studies of GMZ2’.10C at 4°C Storage and During Freeze/Thaw Cycles. The Calculated Area of the Peaks Corresponding to Main Form of GMZ2’.10C and Dimers Thereof are Shown. The Main Form Ratio is Calculated as Percentage of Main Form to Dimers, and Provides a Measure of the Degree of Aggregation. The Main Form Intensity is Calculated as Percentage of the Main Form to the Reference Analysis at Day 0, and Provides Information of the Overall Physical Stability of Intact GMZ2’.10C
Full size table

Next, the integrity of immunoaffinity and conventional purified GMZ2’.10C was assessed in a preliminary stability study over 30 days at 4°C (Fig. 4, Table II, and Supplementary Fig. 3). Samples at 0, 15 and 30 days were analysed by SEC-HPLC, SDS-PAGE, Western blotting with anti-GMZ2’ or mAb45.1, and 2-site ELISA using freshly purified IP-R0.10C as a reference (14). At day 15, CP-GMZ2’.10C (Fig. 4b and Supplementary Fig. 3a) did not show degradation whereas partial degradation of IP-GMZ2’.10C was observed (Fig. 4f and Supplementary Fig. 3e). The site of this degradation could be assigned to the GMZ2’ region, as the mAb45.1 antibody, specific for the 10C region did not show altered binding (Supplementary Fig. 3h). At day 30, CP-GMZ2’.10C also showed degradation (Fig. 4c, Table II), especially in the GMZ2’ portion (Supplementary Fig. 2a), whereas reactivity of the 10C region is still present as can be seen in mAb45.1 Western blot and 2-site ELISA (Supplementary Fig. 3b, d, f, and h).

CP-GMZ2’.10C appeared fully stable after 15 days, while IP-GMZ2’.10C showed signs of degradation at this time. At 30 days, a large amount of lower-mass species were seen in SEC-HPLC and SDS-PAGE in both batches, and only a minor fraction of the main form was left (Fig. 4c and g, Supplementary Fig. 3a and b). However, the Western blot with mAb45.1 and 2-site ELISA with mAb45.1 retained most of the intensity at 30 days, indicating that the epitope on the 10C region was still intact. In contrast, the Western blot with anti-GMZ2’ had a substantial decrease in signal over the same time period. This could indicate that the degradation is happening in the R0-region of GMZ2’.10C, marking this region as the major site for covalent degradation upon storage at 4°C. The increased stability of the 10C region could be explained by the abundance of disulphide-bonds, which confer significant stability (40). We note that it is possible that some proteolytic degradation does occur in the 10C region but dissociation into separate protein components (and thus detection) is prevented by the disulphide-bonds.

To further study the stability of IP-GMZ2’.10C, we exposed IP-GMZ2’.10C samples to stressed conditions and analysed samples by SEC-HPLC to test the physical stability at elevated temperatures (Fig. 5). Significantly, even after incubation of IP-GMZ2’.10C for 24 h at 37°C, no change in physical stability or increased aggregation is apparent (Fig. 5b). After 5 h at 60°C, roughly 50% of the sample has aggregated to a dimeric state (Fig. 5c). After 65 h at 60°C no IP-GMZ2’.10C was present in the main form (Fig. 5d).

Fig. 5
figure 5

Physical stability of IP-GMZ2’.10C under stressed conditions. Samples of IP-GMZ2’.10C from −80°C storage were incubated at either (a) no incubation (reference) with 76% main form, (b) 37°C for 24 h with 70% main form, (c) 60°C for 5 h with 34% main form or (d) 60°C for 65 h with 0% main form present in the chromatogram. Size exclusion chromatography was performed under native conditions and ratios of main form peak to dimer peak were calculated to assess the physical stability and aggregation of GMZ2’.10C under stressed conditions.

Full size image

In conclusion, the GMZ2’.10C produced by either workflow was stable for several freeze/thaw cycles, showed good physical stability at 4°C as well as at slightly elevated temperatures. The produced GMZ2’.10C, in particular CP-GMZ2’.10C, thus appears compatible with further pharmaceutical development.

Map** the Disulphide-Bond Connectivity of Purified GMZ2’.10C, R0.6C and R0.10C

To unequivocally establish the disulphide-bond connectivity within the 10C region of Pfs48/45, GMZ2’.10C (IP and CP) were analysed by LC-MS/MS after tryptic digestion. As the previously reported R0.6C and R0.10C fusion proteins similarly encompass domain III or domain II and III of Pfs48/45, respectively (12, 14), these were purified by the same immunoaffinity workflow as IP-GMZ2’.10C and analysed in parallel to support results obtained for GMZ2’.10C. Samples were digested by trypsin with and without prior reduction treatment with DTT and analysed by LC-MS/MS. The tryptic peptide maps of reduced GMZ2’.10C, R0.6C and R0.10C yielded a coverage of >83% of the sequence of either protein, including peptides covering all potentially disulphide-bonding Cys residues. As expected, these same peptides were not present in non-reduced samples, demonstrating their involvement in cross-linking to other peptides containing Cys residues in GMZ2’.10C, R0.6C and R0.10C. By manual inspection of the MS and MS/MS data, we observed three peptides (I-III) in more than three different charge states each, which were distinctively present only for the sample that had been prepared without reduction (Fig. 6a-c). Based on the peptide mass and resulting fragment ions (see Fig. 6e for an example), these peptides could be unequivocally identified as disulphide cross-linked (XL) peptides of the 10C region (see Fig. 6f for a schematic overview). In XL peptide I (consisting of three cross-linked peptide segments from domain II), C646 forms a disulphide-bond with either C692 or C694, while C653 forms a bond with the other Cys residue. In XL peptide II (consisting of two cross-linked peptide segments from domain III), C717 and C746 form a disulphide-bond. In XL peptide III (consisting of another two cross-linked peptide segments from domain III), C763 forms a disulphide-bond with either Cys829 or Cys831, and Cys771 forms a bond with the other Cys residue. Due to the close proximity of two cysteine pairs, it was in some cases only possible to resolve the disulphide-bond connectivity to either of two cysteine residues placed 1–2 residuesapart. Importantly, the same XL peptides I-III were present in similar abundance in R0.10C, indicating that the 10C region adopts the same connectivity in both fusion proteins (Supplementary Fig. 4a). To our knowledge, these findings represent the first experimental cysteine connectivity map** of the 10-region of the antigen Pfs48/45 from the P. falciparum malaria parasite. Our results appear to be in agreement with an NMR study of the 6C region of Pf12, a protein that is a member of the s48/45 protein family together with Pfs48/45 (17).

Fig. 6
figure 6

Map** of the disulphide-bond connectivity of domain II and III of Pfs48/45. IP-GMZ2’.10C with reduction (red traced spectra) and without reduction (black traced spectra) was digested with trypsin prior to LC-MS/MS analysis. Three cross-linked peptides, XL peptide I [M + 6H]6+ (a), XL peptide II [M + 6H]6+ (b) and XL peptide III [M + 5H]5+ (c) of the 10C region were absent in the mass spectrum of the reduced protein. The proposed disulphide-bond connectivity for the peptides with two disulphide bonds is indicated by the dashed red lines, and the connectivity of the peptide with a single disulphide-bond is indicated with a solid black line. MS/MS fragments observed are indicated by dashed black arrows. (d) Several peptides without cysteines were analysed simultaneously, the spectrum for one of these peptides is show here. (e) MS/MS spectrum of XL peptide I. F) Schematic of the identified cysteine connectivity in the Pfs48/45-10C region of GMZ2’.10C. The cysteine bond that are shown with a solid black line have been verified with LC-MS/MS and cysteine bonds shown as dashed red lines are one of two possibilities for cysteine connectivity.

Full size image

In good agreement, XL peptides II and III (domain III) were observed exclusively for the non-reduced sample of R0.6C, corresponding to the same disulphide-bonding in domain III of both R0.6C and R0.10C. However, XL peptide I was not observed in the non-reduced sample of R0.6C, nicely corresponding to the absence of domain II in this shorter protein antigen. Results obtained for R0.6C are thus in line with the disulphide-bond map** for GMZ2’.10C and R0.10C and acts to further validate the experimentally established disulphide connectivity of the 10C region of Pfs48/45. To verify that the same quantity of sample of non-reduced vs. reduced proteins was analysed in each LC-MS/MS experiment, we monitored the abundance of several other non-cysteine containing peptides (see Fig. 6d).

Immunogenicity of GMZ2’.10C

In order to examine the ability of produced GMZ2’.10C to present correctly folded epitopes in vivo, groups of Wistar rats (n = 10) were immunized three times at 3-week intervals with 10 μg of CP- or IP-GMZ2’.10C formulated with Montanide ISA 720 VG. Responses were measured using sera taken three weeks after the third immunization (Day 63). CP-GMZ2’.10C elicited significantly lower levels of gametocyte-specific (Mann-Whitney test, p = 0.0011) and epitope I-specific (Mann-Whitney test, p = 0.0003) antibodies than IP-GMZ2’.10C (Fig. 7a). This difference in antibody levels was also reflected in the 6C-ELISA but not in the GLURP- and MSP3-ELISAs (Fig. 7b). Collectively, these findings support the notion that a correctly folded 10C-domain is required for induction of specific antibodies that recognize the native parasite protein. Irrespective of the difference in the sexual-stage ELISAs, pooled sera (1:9 dilution) from both groups of immunized rats promoted strong TB activity in the SMFA, an assay that assesses the capacity to inhibit the development of oocysts (data not sown). This supports the conclusion that recombinant GMZ2’.10C, purified from either immune- or conventional chromatography, represents a promising vaccine candidate for further clinical development.

Fig. 7
figure 7

Immunogenicity of purified GMZ2’.10C. Rats were immunized with 10 μg of CP- or IP-GMZ2’.10C, shown in blue or red, respectively. Levels of specific antibodies were measured for CP-GMZ2’.10C and IP-GMZ2’.10C in the Gametocyte-extract ELISA and in the mAb45.1 competitive ELISA (a) and for GLURP-R0, MSP3 and Pfs48/45-6C (b). Antibody titres are expressed as EC50 values. The lines represent the median values.

Full size image

Conclusion

Here we have successfully expressed, purified and characterised a novel fusion protein GMZ2’.10C, that combines multiple antigens of the Plasmodium falciparum parasite, in terms of yield, purity, stability, and structure. We have showed that the vaccine candidate can be produced in quantity, purity and quality suitable for further pharmaceutical development using a Lactococcus based expression system and a conventional purification workflow compatible with up-scaling. The conventional purification workflow was benchmarked against an established immunoaffinity workflow and yielded GMZ2’.10C of similar primary structure and stability. The conventional purification workflow yielded GMZ2’.10C in lower purity but higher yield than the immunoaffinity purification workflow. The complex disulphide-bond connectivity of Pfs48/45 (both domain II and III) encompassed in either protein antigen was validated for the first time by tandem mass spectrometry. Overall, our results form a critical framework for spear-heading the development of GMZ2’.10C as a candidate in a combined much-needed vaccine against both the asexual and sexual life-stages of the malaria-causing Plasmodium falciparum parasite.

Acknowledgments and Disclosures

This work is funded in part by the PATH Malaria Vaccine Initiative. We also gratefully acknowledge the financial support from the Danish Council for Independent Research | Natural Sciences (Steno Grant no. 11-104,058), and the Danish Council for Strategic Research (Grant no. 13127). The authors would like to thank Tenna Jensen for technical assistance.