Optimized Method for Preparation of IgG-Binding Bacterial Magnetic Nanoparticles

In this study, the optimized method for designing IgG-binding magnetosomes based on integration of IgG-binding fusion proteins into magnetosome membrane in vitro is presented. Fusion proteins Mbb and Mistbb consisting of magnetosome membrane protein MamC and membrane associating protein Mistic from Bacillus subtilis as anchors and BB-domains of Staphylococcus aureus protein A as IgG-binding region were used. With Response Surface Methodology (RSM) the highest level of proteins integration into magnetosome membrane was achieved under the following parameters: pH 8.78, without adding NaCl and 55 s of vortexing for Mbb; pH 9.48, 323 mM NaCl and 55 s of vortexing for Mistbb. Modified magnetosomes with Mbb and Mistbb displayed on their surface demonstrated comparable levels of IgG-binding activity, suggesting that both proteins could be efficiently used as anchor molecules. We also demonstrated that such modified magnetosomes are stable in PBS buffer during at least two weeks. IgG-binding magnetosomes obtained by this approach could serve as a multifunctional platform for displaying various types of antibodies.


Introduction
The systems of antibodies conjugated to the surface of magnetic nanoparticles (MNPs) are increasingly used in diagnostics and therapy. Many studies have previously demonstrated their efficiency for cancer cell detection, magnetic separation of stem cells, magnetic immunoassay and as a carrier for targeted drug delivery [1,2]. Recently, an interesting alternative to these synthetic MNP, called magnetosomes, was found in magnetotactic bacteria. Magnetosomes are intracellular magnetic crystals produced by magnetotactic bacteria (MTB) and also referred to as bacterial magnetic nanoparticles (BMPs) [3,4]. The advantages of magnetosomes in comparison with artificial MNPs are: i) uniform species-specific size (30-120 nm) and shape; ii) magnetic crystal is coated with a lipoprotein membrane, making BMPs easily dispersed in aqueous suspension and providing an opportunity to modify a surface by genetic engineering; iii) high crystallinity; iv) low cytotoxicity [5,6]. Due to these features, magnetosomes attract significant interest as biogenic MNPs, which could be used in a number of biomedical applications. For instance, magnetosome chains were shown to be highly efficient for cancer therapy when they are exposed to an alternative magnetic field [7], magnetosomes have been proposed as potential carriers for drugs in tumor treatment and for DNA in genetic transformation [8,9].
Three general approaches have been proposed to magnetosomal membrane modification: subsequent chemical alterations of purified magnetosomes [10,11], transformation of MTB with genetic constructs encoding magnetosome membrane proteins fused to foreign proteins (in vivo modification) [12][13][14] and insertion of recombinant fusion proteins into magnetosomal membrane in vitro [15,16]. Magnetosome membrane proteins Mms16, MamC, MamF and MamG were proposed as anchor molecules for foreign proteins display on the magnetosome surface during genetic manipulations with MTBs magnetosome. MamC was used in most of such studies as an efficient anchor protein [13,17,18]. In GFP-based analysis MamC-GFP displayed the highest expression and fluorescence levels comparing with GFPtagged magnetosome proteins MamF and MamG. However, genetic manipulations of MTB are still hampered due to the difficulties in cultivation and fastidiousness of these organisms. At the same time, magnetosome membrane proteins can be easily overexpressed in E. coli and purified according to the standard procedures, i.e. immobilized metal ion affinity chromatography. Thus, Matsunaga and co-authors have demonstrated insertion of heterologously expressed recombinant MagA-Luc fusion protein consisted of integral magnetosome protein MagA and firefly luciferase into the membrane of purified magnetosomes [16]. This approach seems to be an efficient and simple way for magnetosome surface modification. In this study the role of NaCl concentration and sonication time was investigated but not the mutual influence of such factors as NaCl concentration, pH value and the mode of mechanical action (sonication vs vortexing).
In this study we presented an optimized method for the IgG display on the surface of BMP. Chimeric proteins containing double IgG-binding B-domains of Staphylococcus aureus protein A fused with anchor proteins were integrated in vitro into the membrane of magnetosomes extracted from the magnetotactic strain Magnetospirillum sp. SO-1 by means of simple vortexing procedure. Highly hydrophobic and small (12.4 kDa) protein MamC was chosen as an anchor molecule for introduction of fused proteins into magnetosomal membrane. As another promising protein for this purpose was chosen Mistic, an unusual membrane-associated protein (13 kDa) from Bacillus subtilis which was recently found to be capable of autonomous integrating into the membrane [19]. For this study, two genetic constructs, mbb and mistbb, coding the fusion proteins, were synthetized. Both constructs contained double B domain of Staphilococcus aureus protein A as immunoglobulin-binding region and differed by their membrane-anchoring domains. In mbb it was MamC protein from Magnetospirillum magnetotacticum MS-1, the corresponding domain in mistbb was Mistic proteins from Bacillus subtilis. Whereas MamC is the commonly used protein for the functionalization of bacterial magnetosomes [13,20], Mistic protein in this study was used as anchoring domain for this purpose for the first time. In the previously published studies Mistic was used as a fusion partner for the overexpression of histidine kinase receptors (HKR) in E. coli [21], eukaryotic membrane protein (pkjDes4) and a prokaryotic membrane protein (pkjLi) in Lactococcus lactis [22], as well as for expression of Gprotein coupled receptors (GPCR) in E. coli [23]. The insertion procedure was optimized using Response Surface Methodology (RSM). IgG-binding activities of magnetosomes modified by fusion   proteins contained MamC and Mistic as anchor domains were compared. The selectivity of obtained BMP was demonstrated using embryonic kidney cells extract contained Kaiso/GFP protein and anti-GFP IgG conjugated with IgG-binding BMP. Finally, stability of modified BMP after storage at +4uC was analyzed.  The medium for Magnetospirillum sp. SO-1 consisted of (per liter of medium): 1 ml mineral solution [24], 0.7 g KH2PO4, 0.5 g sodium succinate, 0.1 g yeast extract, 0.35 g NaNO3, 10 ml 0.01 M ferric citrate, 0.05 g sodium thioglycolate. pH was adjusted to 6.75 with NaOH. The cells were cultivated at 28uC under microaerobic conditions in a 15-L fermenter for 3-4 days.

Magnetosomes Extraction and Purification
After achieving growth stationary phase Magnetospirillum sp. SO-1 cells were centrifuged 10,000 g for 10 min at +4uC, resuspended in 20 mM HEPES buffer, pH 7.4, contained 4 mM

Bacterial Strains and Growth Media
We used Magnetospirillum magnetotacticum strain MS-1 (DSM 3856), a Staphylococcus aureus strain (VPKM 1899), and Escherichia coli strains XL-1 Blue (Stratagene, United States) and BL21 (DE3) (Novagene, United States). A pET23a(+) vector (Novagene, United States) was used for genetic engineering manipulations. E. coli XL-1 Blue cells were grown in an LB medium [25]. Expression of recombinant proteins was performed in a TB medium [26]. A solid medium for the cultivation of single E. coli colonies on Petri dishes was prepared by adding 2% agar to the LB medium.

Construction of the pET23a(+)/mbb and pET23a(+)/mistbb Expression Vectors
Preparation of competent E. coli cells, transformation of cells with plasmid DNA, isolation of plasmid DNA, hydrolysis with restriction endonucleases, phosphorylation, electrophoresis in agarose gel, and other standard procedures were performed as described in [25] and as recommended by the manufacturers of the enzymes used. pET23a(+)/mbb performed as described previously [27]. PCR fragment of mistic gene was obtained from Bacillus subtilis genomic DNA using specially designed sequencemodifying oligonucleotide primers MistF 59-AGAGGAGATAT-CATATGGGCTTT-39 and MistR 59-CAGAACCGGATC-CTTCTTTTTCTC-39. After digestion with NdeI and BamH1 restrictases mistic PCR product was cloned into vector pET23a(+)/ mbb replacing mamC in genetic construct mbb [27]. The presence of the histidine tag at the C-terminus of heterologous expressed proteins allows for the easy identification of proteins and their purification on chelating sorbents. XL-1 Blue E. coli cells were transformed by the pET23a(+)/mbb and pET23a(+)/mistbb recombinant plasmids. Clones harboring plasmids with target inserts were selected using PCR screening and subsequent sequencing of inserts using a BigDye Terminator v3.1 Cycle Sequencing Kit (App1ied Biosystems, United States). Nucleotide sequences were determined on an ABI 3730 automatic sequencer (App1ied Biosystems, United States).

Analysis of the Total Protein from BL21 (DE3) E. coli Cells
Cells were collected from 1 mL of medium, resuspended in 100 mL of TED buffer (10 mM tris-HCl, pH 6.8, 1 mM EDTA, 1% SDS), and then incubated at 100uC for 5 min. The lysate obtained was analyzed by denaturing electrophoresis in a polyacrylamide gel according to the method of Laemmli. The protein concentration in the solution was determined by the method of Bradford; BSA solutions were used to build a calibration curve.

Fractionation of Soluble Cellular Proteins of E. coli
Cells harvested from 1 mL of medium were resuspended in 50 mL of buffer I (100 mM Tris-HCl, pH 8.0, 0.5 M sucrose, 0.5 mM EDTA). Phenylmethylsulfonyl fluoride (PMSF) was added to the solution to a final concentration of 0.1 mM; 2.5 mL of 2 mg/ml lysozyme solution in the same buffer was also added. The mixture was incubated at room temperature for 20 min, and then 100 mL of buffer I and 100 mL of water were added, mixed, and incubated for 10 min. An equal volume of 0.2% aqueous solution of Triton X-100 was added, and the mixture was incubated for 1 h at room temperature. The cell suspension was frozen at 220uC, thawed at room temperature three times, and then centrifuged at 12,000 g for 3 min. The supernatant contained the soluble protein fraction, and the pellet contained the insoluble protein fraction. The precipitate was resuspended in 100 mL of lysis buffer supplemented with Triton X-100 (to a final concentration of 0.1%). A membrane fraction was obtained in the following way: 150 mL of overnight culture was centrifuged at 12,000 g for 10 min. Cells were suspended in 15 mL of buffer I, and 3 mL of 2 mg/mL lysozyme solution in the same buffer was added to the suspension. The mixture was incubated at room temperature for 30 min, and then 135 mL of buffer I, 150 mL of 0.1 mM PMSF, and 300 mL of 0.5 M EDTA were added to it. The sample was sonicated for 10 min using a Sonopuls UW2070 device (Bandelin, Germany) at a frequency of 20 kHz. The cell debris was removed by centrifugation at 6,000 g for 30 min, and the membrane fraction was precipitated by supernatant centrifugation at 100,000 g for 2 h at +4uC.

Purification of Fusion Proteins
The presence of histidine tag in both fusion proteins substantially allowed to perform their purification by metal chelate affinity chromatography. The membrane fraction preparation obtained was resuspended in buffer A (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 5% glycerol, 10 mM b-mercaptoethanol, 10 mM imidaz-

Optimization of Fusion Proteins Insertion into Magnetosome Membrane
The optimization of Mbb and Mistbb insertion into magnetosome membrane was carried out using Response Surface Methodology (RSM). Variation due to model inadequacy was evaluated by Lack-of-fit test. The analysis of variance (ANOVA), which was carried out by Fisher's statistical test, was employed for the determination of the significance of the models. The quality of the model was evaluated by the coefficient R 2 . Several parameters possessing the significant influence on the integration of fusion proteins in the magnetosome membrane were tested in a series of pilot experiments: pH value, NaCl concentration, and the type of stirring of suspension (vortexing or sonication). The respective levels of fusion protein integration vs coded levels for the factors are listed in Table 1. Counting data were obtained using Minitab 15.0. The graphical representations of the regression model plots and their corresponding contour plots were obtained using Design-Expert software (version 9.0.1.0, Stat-Ease Inc., USA). In all experiments 10 mg of magnetosomes and 50 mg of each fusion protein were used. Total volume of reaction mixture was 1 ml.

Atomic Force Microscopy
Visualization and size of biogenic magnetic nanoparticles were performed on NTEGRA Prima microscope (NT-MDT, Russia). Semicontact mode was selected for scanning, NSG01 probes (resonant frequency 87-230 kHz, force constant 1,45-15,1 N/ m)were used, scanning speed was 1 Hz. 10 ml of sample was incubated on freshly cleaved mica for 1 min which was subsequently rinsed thoroughly with milliQ water and dehydrated in a dry air stream. The resulted preparations were scanned immediately after drying at room temperature and ambient humidity with no additional treatment.

Expression and Purification of Fusion Protein Kaiso-GFP
The expression vector for methyl-DNA-binding protein Kaiso was generated by inserting human cDNA Zbtb33 (Kaiso) [28] into the pFLAG-CMV-2 contained GFP at the C-termini. Human embryonic kidney (HEK 293) cells were transiently transfected with pFLAG-Kaiso-GFP using Calcium phosphate transfection method (Promega) according to the manufacturer's protocol, and cells were used for experiments after 48 h. The nuclear localization of protein Kaiso and transfection efficiency were confirmed using immunofluorescence assay. Cells were seeded on a 24-well, transfected of Kaiso-GFP and after 48 h fixed with 4% paraformaldehyde (15 min, 37uC), and washed with PBS (16). The preparations were mounted in Mounting medium for fluorescence with DAPI (VECTOR, USA) and visualized on an Olympus BX51 microscope (Olympus, Japan). Then the nuclear extraction was prepared from transfected HEK 293 cells according to standart protocol.

Enzyme-linked Immunosorbent Assay
The comparison of IgG-binding activities of Mbb and Mistbb was performed by ELISA. Human insulin (1 mg/well) was preadsorbed in the wells of the ELISA plate overnight at +4uC. The residual sorption was blocked by incubation with a 1.5% solution of BSA in a PBS-Tween buffer (PBS, 0.05% Tween 20) for 1 h (200 ml/well). Monoclonal mouse IgG antibodies (0.1 mg/well) against human insulin (Imtek, Russia) were added to the wells and incubated for 1 h at room temperature. The wells were washed four times with PBS-Tween buffer, and then the fusion proteins were added at the pre-requisite dilutions and incubated for 1 h at room temperature. After a similar washing procedure, the plates were incubated with 0.1 mg/well of mouse IgG antibodies against the His-tag (Imtek, Russia) for 1 h and detection was performed using a hydrogen peroxide/horseradish peroxidase detection system with TMB (Sigma, USA) as a chromogenic substrate. The adrenocorticotropic hormone carrying a His-tag at the Cterminus was used as a negative control. All assays were carried out in at least triplicate.

IgG-binding Activity Assay
The ability of modified magnetosomes to bind IgG was tested by magnetic ELISA. 10 mg/well BMP-Mbb or BMP-Mistbb was incubated with 100 ml of horseradish peroxidase-labeled rabbit IgG (100 mg/ml) for 1 h at room temperature. Then magnetosomes were separated on magnetic stand (Promega, USA) and washed 5 times with 100 ml PBST (PBS, 0.05% Tween 20, pH 7.0). Washed magnetosomes were incubated with TMB Liquid Substrate System for ELISA (Sigma-Aldrich, USA) for 1 min, the reaction was terminated by 50 ml 1 M HCl solution. Wild-type magnetosomes were used as negative control. All assays were carried out in at least triplicate.

Comparison of IgG-binding Activities of Mistbb and Mbb
After the fusion proteins have been expressed and purified (Figure 2), we compared the IgG-binding activities of Mistbb and Mbb by ELISA. Taking into account, that B-domain of staphylococcal protein A is preferentially bonded with IgG Fc fragment [30], human insulin was absorbed at the surface of immunoplate wells and then primary mouse anti-insulin antibodies were bonded with the absorbed insulin thus providing the proper orientation of their Fc fragments on the well surface. The results of detection of fusion proteins (Mbb or Mistbb) coupled with Fc fragments of primary antibodies are shown at Figure 3. According to the ELISA data, Mistbb exhibited IgG-binding activity similar to Mbb. For the negative control sample no signal was observed. Thereby both proteins were used in further experiments.

Optimization of the Artificial Insertion
For the proof-of-concept, first, Mbb and Mistbb were inserted into magnetosome membrane by sonication in the presence of 300 mM NaCl according to a previously described procedure [16].
Immunoglobulin-binding ability of fusion proteins integrated into magnetosome membrane was tested by magnetic ELISA as described in Material and methods section. The data of ELISA are given at Figure 4. As it can be concluded from the data, both proteins kept their ability to bind immunoglobulin, but the IgGbinding activity of magnetosomes with membrane-integrated Mbb protein was higher. When wild-type magnetosomes incubated with antibodies OD 450 values were significantly low, that indicated negligible non-specific adsorption of antibodies on the surface of BMPs. As the result of preliminary testing (data not shown), the following parameters affecting the integration of fusion proteins into the magnetosome membrane were chosen -pH, NaCl concentration and mechanical mode of integration (vortexing or sonication). The Response Surface Methodology (RSM) with Box-Behnken design (BBD) was employed to determine the optimal values of selected parameters. It was proved, that this methodology is an effective tool for prediction of optimal process condition for multiple parameter systems [31][32][33]. The respective levels with the coded levels for the factors are listed in Table 1.
The capacity of BMP-Mbb and BMP-Mistbb to bind IgG was tested in assay with horseradish peroxidase-labeled rabbit IgG. OD 450 values were used for activity rating. Experimental design and results are shown in Table 2. The highest OD 450 for Mbb was observed under the following conditions -with no NaCl added, pH 8.0, 55 s of vortexing or sonication. In the case of Mistbb integration the highest OD 450 was achieved for sonication mode with no NaCl added, pH 8.0, 5 s sonication and for vortexingunder 250 mM NaCl, pH 11.0 and 55 s vortexing.

RSM Analysis of Mbb Protein Integration through Sonication
The second-order polynomial was as follow: The equation was sustainable with R 2 = 0.937, what means that 93.7% of variations could be explained by the chosen model. The ''Lack-of-Fit F-value'' of 7.43 implies the Lack of Fit is not significant relative to the pure error. There is a 12.09% chance that a ''Lack of Fit F-value'' this large could occur due to noise.
To estimate the statistical faithfulness F-test and ANOVA analysis were performed ( Table 3). The Model F-value of 8.27 implies the model is significant. There is only a 1.58% chance that an F-value this large could occur due to noise. Between the faithful correlations were: A 2 , C2, A6C, B6C. 3D plots based on equation (1.1) are given at Figure 5.
The optimal predicted parameters for sonication-mediated integration of Mbb protein into magnetosome membrane were the following: pH 5.97, 500 mM NaCl, sonication for 5 s.

RSM Analysis of Mistbb Protein Integration through Sonication
The second-order polynomial was as follow:  The equation was sustainable with R 2 = 0.9431. Lack of fit = 0.317, there is a 31.74% chance that a ''Lack of Fit F-value'' this large could occur due to noise, however this value is not significant relative to the pure error ( Table 4). The Model F-value of 9.20 implies the model is significant. There is only a 1.25% chance that an F-value this large could occur due to noise.
Between the faithful correlations were: A, B, C, B 2 , A6B, B6C (p-value ,0.05). It means that the efficacy of Mistbb integration is effected by (pH value)/(NaCl concentration) and (NaCl concentration)/(sonication time). 3D plots based on equation (1.2) are given in Figure 6. The optimal predicted parameters for ultrasonic integration of Mistbb protein into magnetosome membrane were the following: pH 8.9, 0 mM NaCl and sonication for 5 s.

RSM Analysis of Mbb Protein Integration through Vortexing
The second-order polynomial was as follow:  The equation was sustainable with R 2 = 0.9507, lack of fit value = 0.186 supported the adequateness of this model ( Table 5). The Model F-value of 10.72 implies the model is significant. There is only a 0.89% chance that an F-value this large could occur due to noise. Between the faithful correlations were: A 2 , A6B, B6C (p-value ,0.05). 3D plots based on equation (1.3) are given in Figure 7. The optimal predicted parameters for vortex-mediated integration of Mbb protein into magnetosome membrane were the following: pH 8.78, 0 mM NaCl, 55 s of vortexing.

RSM Analysis of Mistbb Protein Integration through Vortexing
The second-order polynomial was as follow: The equation was sustainable with R 2 = 0.956, lack of fit value = 0.762 supported the adequateness of this model ( Table 6). The Model F-value of 10.72 implies the model is significant. There is only a 0.89% chance that an F-value this large could occur due to noise. Between the faithful correlations were: A, C, A 2 , B 2 , A6C. 3D plots based on equation (1.4) are given in Figure 8. The optimal predicted parameters for vortex-mediated integration of Mistbb protein into magnetosome membrane were the following: pH 9.48, 323 mM NaCl, 55 s vortexing.

Testing the Optimized Conditions
To verify the optimal predicted parameters for IgG-binding proteins, a validation experiment was performed in triplicate, where the quantity of magnetosomes per reaction was decreased to 20 mg/well. OD 450 values obtained with RSM-predicted and BBD maximum output parameters ( Figure 9) were compared. According to the results of magnetic ELISA, the reliable difference between both values was detected. The highest OD 450 values (0.60560.020 at Mbb and 0.56860.021 at Mistbb) were achieved after vortex-mediated integration of both proteins under RSMpredicted parameters. Since the absolute levels of integration for Mbb and Mistbb were similar, both proteins are equally applicable as anchor molecules. According to the data obtained under these parameters vortexing was chosen as the most appropriate mode of integration. Since under optimized conditions there were no faithful differences between BMP-Mbb and BMP-Mistbb IgGbinding activities, both proteins could be successfully used for magnetosome membrane modification.

AFM of Modified Magnetosomes
The morphology of magnetosomes was analyzed by AFM at the various stages of their modification. The intact magnetosomes were 50-60 nm in diameter ( Figure 10A). Being conjugated with fusion proteins (Mbb or Mistbb), the resulted magnetosomes possessed the same diameter as intact ones ( Figure 10B, C), whereas incubation of fusion protein coupled magnetosomes with IgG their diameter increased up to 95-105 nm and the surface of IgG-bound magnetosome became hilly ( Figure 11B, C). Immunoglobulins were seen through AFM as 25 nm knobby corpuscles ( Figure 11A).
To examine the ability of modified BMP-IgG to bind selectively to the respective antigen, we used BMP conjugated with anti-GFP IgG and transformed human embryonic kidney (HEK 293) nuclear extract, contained modified Kaiso fused with GFP (Kaiso/GFP). The expression of Kaiso/GFP genetic construction was proven by GFP fluorescence of the transformed cells nuclei ( Figure S1). After exposure of IgG-bind magnetosomes conjugated with anti-GFP IgG with Kaiso/GFP containing nuclear extract, magnetosome diameter increased up to 120 nm ( Figure 12A, B). No size increase of magnetosomes occurred when wild-type magnetosomes with anti-insulin antibodies immobilized on their surface were incubated with HEK 293 nuclear extract (Figure 12C, D).

Permanence of BMP-Mbb and BMP-Mistbb
Magnetosomes with Mbb and Mistbb inserted in their membranes were stored in PBS buffer at +4uC for 3 weeks. During this period modified magnetosomes were tested on their ability to bind IgG by magnetic ELISA test. According the results given in Figure 13, BMP-Mbb and BMP-Mistbb retained their IgG-binding activity up to 14 day storage under above conditions. Again, no difference was found between BMP-Mbb and BMP-Mistbb in regard to activity retaining. The decrease of IgGbinding activity at 21 day of storage is presumably accounted for by membrane degradation.

Conclusion
A simple and efficient procedure for the integration of fusion proteins into magnetosome membrane by vortexing of magneto-some/protein mixture was proposed. The optimal parameters of Mbb integration were determined as follows: pH 8.78, with no NaCl added, and vortexing for 55 s. The highest level of Mistbb integration was achieved at pH 9.48, 323 mM NaCl and 55 s of vortexing. For the first time, we showed that not only original magnetosome membrane proteins but also foreign membrane associating Mistic protein could be effectively used as anchor molecules for integration of hybrid proteins into the BMPs membrane. Also we demonstrated that the resulted modified magnetosomes were stable in PBS buffer for at least two weeks. The optimized procedure for design of IgG-binding BMP in vitro could further facilitate the development of methods for functionalizing BMP via protein display on their surface.