The GEMP contained four components: a truncated phage lambda lysis cassette, a non-specific nuclease NucA, an intein, and an anti-TBBPA Nb. The phage lambda lysis cassette consisted of four genes: S, R, Rz and Rz1. The S gene encodes holin and its inhibitor. The R gene encodes the endolysin. Rz and Rz1 are nested genes encoding Rz and Rz1 proteins, respectively, involved in resolving the oligopeptide linkages. Holin is a small membrane protein, which forms μm-scale holes in host cytoplasmic membrane (inner membrane). These holes result in the release of R endolysin, Rz and Rz1 proteins, thus accessing to their substrate (host cell wall), and at last, the disruption of host cell [30]. The surface attached extracellular nuclease (Nuc) of Staphylococcus aureus is a secreted enzyme that possesses a long 60-residue Sec signal sequence. The secreted form of Nuc, known as NucB, is processed by most S. aureus strains to a shorter form called NucA [31]. The fusion of NucB and the signal sequence of E. coli’s major outer membrane protein (OmpA) resulted in accumulation of NucA in the periplasm of E. coli [32]. NucA can hydrolyze DNA and RNA, whether double or single strand, into oligo- and mononucleotides. NucA was here responsible for reducing extract viscosity by completely digesting genomic DNA [33].
To avoid cell lysis during cultivation, the elements mentioned above need to be expressed at different subcellular locations. The truncated phage lambda lysis cassette and the NucA were expressed in cytoplasm and periplasm, respectively, while the intein and Nb were expressed on the surface of magnetosome (Fig. S2). These elements were cloned into different plasmid vectors, introduced into E. coli S17-1 via transformation, followed by function verification of the first two elements. All of vectors was then transferred into MSR-1 via bacterial biparental conjugation.
Transfer of nucB and RRz/Rz1 into E. coli S17-1
In the present work, only R, Rz and Rz1 in the phage lysis cassette were employed and cloned into a plasmid vector pBBR1MCS-2 to construct recombinant strains. The S gene was omitted to avoid host lysis during cultivation. The native promoter PR’ from λ phage was cloned and employed for the expression of RRz/Rz1 genes. Besides, the DNA fragment encoding OmpA signal peptide was fused with nucB, and then cloned into the plasmid vector pBBR1MCS-2, along with the RRz/Rz1. The lac promoter from pBBR1MCS-2 was employed for the expression of NucA. The resulting plasmid was referred to as pBBRONL (Fig. 2a). Another recombinant plasmid pBBRPNL was also constructed, which was the same with pBBRONL, except that the signal peptide sequence was from a phaZ1 gene instead of ompA (Fig. 2a). The PhaZ1 here was an “intracellular” poly(3-hydroxybutyrate) (PHB) depolymerase of Rhodospirillum rubrum, which is a periplasm-located protein with specificity for native PHB and with structural similarity to extracellular PHB depolymerases [34]. R. rubrum is closely related to Magnetospirillum spp., of which 90% of the 16S rRNA sequence is identical to that of WT MSR-1 [35]. The plasmids pBBRONL and pBBRPNL were then transferred into S17-1 and the resulting recombinant strains were referred to as ONL and PNL, respectively (Fig. 2a).
The target genes nucB and RRz/Rz1 in recombinants were identified by colony PCR and sequencing. On IPTG/DNase agar plates, transparent zones surrounding the colonies of ONL and PNL were observed, while those of controls (S17-1) did not occur (Fig. 2b). NucA was presumably transported across cytoplasmic membrane into periplasmic space under the direct of signal peptide from OmpA or PhaZ1. To confirm this, two recombinant strains and their host strain (S17-1) were cultivated in shaking flasks. Their cell pellets were harvested by centrifugation and resuspended into 3 mL of Tris-HCl (20 mM, pH 8.0), followed by the addition of CHCl3 (20 μL). The majority of ONL and PNL cells was disrupted within 0.5 h, while no apparent changes were observed from S17-1 cells (Fig. 2c), indicating the successful expression of RRz/Rz1 in the cytoplasm of two recombinant strains. Compared to the viscosity of the homogenate from modified S17-1 strains which expressed RRz/Rz1 only, the viscosity from disrupted S17-1 expressing both NucA and RRz/Rz1 was lower (Fig. S3).
Construction of An Engineered MSR-1 Harboring nucB, RRz/Rz1 and Nb
RRz/Rz1 and nucA were successfully expressed in E. coli S17-1, and their protein products exhibited specific activities in the process of cell self-lysis and nucleic acid hydrolysis. We assumed that RRz/Rz1 and nucA could be expressed in other Gram-negative bacteria including magnetotactic bacteria. Herein, a recombinant MSR-1 was developed for automatic downstream processing. The plasmid pBBRPNL with the RRz/Rz1 and recombinant nucB was employed and transferred into MSR-1. The signal peptide sequence of the recombinant nucB gene in this plasmid was from phaZ1 gene of R. rubrum, which was closely related to MSR-1. A fusion gene of the anti-TBBPA Nb, an intein, and MamC was also introduced into MSR-1 via another plasmid vector. The intein gene was cloned from DNA polymerase II of Pyrococcus abyssi, abbreviated as the Pab PolII intein [36]. MamC was the most abundant protein on the magnetosome surface, which helped to express the anti-TBBPA Nb and the Pab PolII intein and anchor them on the surface of magnetosomes [37]. Different from the plasmid pBBRPNL, a suicide plasmid vector pK18mobSacB was employed for cloning the fusion gene of Nb and intein, resulting a recombinant plasmid pKTBInC. After transferred into MSR-1, the anti-TBBPA Nb and Pab PolII intein genes were integrated into the host chromosome. The recombinant MSR-1 strain was referred to as TBInCPNL (Fig. 3a). An additional control strain was also constructed and referred to as TBInC, which contained the fusion gene of anti-TBBPA Nb and Pab PolII intein in its chromosome but did not harbor the plasmid pBBRPNL (i.e., without the RRz/Rz1 and nucB genes in this strain).
The western-blot analysis showed that the proteins RRz/Rz1, NucA and Nbs-Intein-MamC extracted from TBInCPNL migrated as expected (Fig. 3b-d), illustrating successful expression of each exogenous gene in respective compartments of the recombinant strain.
Culture of TBInCPNL
The growth and magnetic response of TBInCPNL in shaking culture: To evaluate the impacts of exogenous genes on the growth of recombinant strains, the OD565 value and Cmag of TBInC and TBInCPNL growing in 100 mL of SLM were detected and compared with those of WT MSR-1 (Fig. 4a). The growth curves of three strains almost overlapped. The maximum OD565 values of MSR-1, TBInC and TBInCPNL were 1.53, 1.46 and 1.49, respectively. After 53 h of cultivation, cells were harvested via centrifugation to yield 0.92, 0.96, and 0.90 g (wet weights, ww) of MSR-1, TBInC, and TBInCPNL, respectively. Although the cell yields varied slightly, the strains showed different magnetic responses. Remarkably, the average Cmag values of TBInC and TBInCPNL were approximately 1.5- and 2-fold higher than that of WT MSR-1, respectively.
Magnetosome’s characteristics: The morphology of magnetosomes was characterized by TEM. Magnetosomes from different strains all appeared in a chain (Fig. 4b (i-iii)). The size and yield of magnetosomes were determined with the ImageJ (Fig. 4b (iv-v)). The diameters of magnetosomes were mostly distributed in a range of 20-50 nm (Fig. 4b(iv)). No significant differences were observed in the size of magnetosomes biosynthesized by MSR-1, TBInC, and TBInCPNL, with an average diameter of 32.7 ± 7.9, 33.1 ± 7.7, and 33.0 ± 7.4 nm, respectively (Fig. 4b(iv)). However, dramatic differences were observed in the numbers of magnetosomes biosynthesized by single cell (Fig. 4b(v)). The numbers of magnetosomes in a single cell of WT were distributed in a range of 5-15 and those in a single cell of TBInC and TBInCPNL were distributed in a range of 5-25 (Fig. 4b(v)). The average number of magnetosomes biomineralized in a single cell of WT, TBInC, and TBInCPNL was 11 ± 5, 14 ± 6, and 16 ± 7, respectively (Fig. 4b(v)). These results demonstrated that the transfer of these exogenous genes into MSR-1 exhibited little inhibition on the proliferation of cells but promoted the biomineralization of magnetosomes. It is implied a positive correlation between the magnetic responses and the number of magnetosomes from various strains.
Suitability of high cell density cultivation: Subsequently, TBInCPNL was incubated in fed-batch cultivation in a 7.5-L fermenter. Fig. 4c showed a typical growth curve including the lag, exponential, stationary, and decline phases. The peak value of OD565 was 19.6, appearing at 114 h. The curve of magnetic response from growing TBInCPNL could be divided into two parts: a decreasing curve and a parabolic curve. Typically, MSR-1 biosynthesizes magnetosomes under a low concentration of dissolved oxygen (dO2 < 1%). When TBInCPNL cells were transferred into the fermenter, dO2 of the culture medium was enhanced and the biosynthesis of magnetosomes in cells was temporarily inhibited, leading to the initial decrease of Cmag. After around 25 h culture, dO2 was gradually driven down to a level suitable for the biomineralization of magnetosomes and thus, Cmag values started to increase. The Cmag value reached to the peak 1.11 and thereafter declined again with the increase of dO2, due to the high stirring rate in partial. After 120 h, TBInCPNL was harvested and the yields of cells and magnetosomes were 106.3 g and 6.8 g (ww), respectively. Hence, in spite of exogenous genes, with nucB and RRz/Rz1 in particular, it was not a problem to carry out a high-density culture of TBInCPNL at a large scale.
Cascade Cell Lysis of TBInCPNL and Hydrolysis of Nucleic Acids
The function of RRz/Rz1 and nucB had been demonstrated in the recombinant E. coli strains (PNL and ONL) above in the section of 3.1. We investigated whether these genes worked in the recombinant MSR-1 strain (TBInCPNL). TBInC, the recombinant strain harboring Nb but not RRz/Rz1 and nucB, was employed as a control. Cells of TBInC and TBInCPNL were harvested from a shake-flask culture (100 mL) and resuspended in 3 mL of PBS (pH 7.4), followed by the addition of 20 µL of CHCl3. After 2 h incubation, the majority of TBInCPNL cells were broken, whereas TBInC cells changed slightly (Fig. 5a). The lysis rate of TBInCPNL cells was over 90% within 0.5 h and approximately 99% within 1 h, as determined with a blood counting chamber (Fig. 5b). Another method also showed that RRz/Rz1 worked well in TBInCPNL. When the frozen TBInCPNL cells were transferred from liquid nitrogen to room temperature, they had almost been completely lysed within 10 min (Fig. 5c). These results indicated that RRz/Rz1 was functional in TBInCPNL, and cascade cell lysis would occur at the suitable condition (e.g., CHCl3 or liquid nitrogen).
Extracts from the periplasmic space of TBInCPNL by osmotic shock were able to hydrolyze the plasmid pBBRPNL and genomic DNA of MSR-1 at 37 °C (Fig. 5d and e), indicating the expression of the functional NucA. After the addition of lysozyme into the cell suspension of TBInC and TBInCPNL (details in Methods), the mixtures were incubated at room temperature for 2 h. In the absence of Ca2+, nucleic acids in homogenates of both TBInC and TBInCPNL could be detected within 2 h. While in the presence of Ca2+, nucleic acids in the homogenate of TBInCPNL were hardly detectable after incubation for 1 h, but detectable in the homogenate of TBInC within 2 h (Fig. 5f). These results illustrated that NucA was expressed in the periplasmic space of TBInCPNL and showed a strong non-specific hydrolysis capability for nucleic acids in the presence of Ca2+ at room temperature.
The activities of NucA and RRz/Rz1 were further evaluated in TBInCPNL cells cultivated in a 7.5-L fermenter. Here, WT MSR-1 was used as a control. Cells were harvested at the later stage of exponential phase of the growth curve and then treated with CHCl3. One hour after the addition of CHCl3, over 75% WT cells were intact, while approximately 90% TBInCPNL cells were disrupted (Fig. 5g and h). In a gravity flow experiment [25], the homogenate of disrupted TBInCPNL cells had a lower viscosity than that of WT cells (Fig. 5i), indicating that the cascade cell lysis and hydrolysis of nucleic acids occurred in TBInCPNL from the large-scale culture.
Extraction and Isolation of Nbs
In general, the cells should be harvested before a drastic drop of Cmag values to ensure a high yield of magnetosomes. Herein, when the Cmag values of TBInCPNL dropped to approximately 1.0 from the peak, cells were harvested even though they were still in the exponential phase of growth. According to the curves of magnetic response and cell growth (Fig. 4c), a 5-L cultivation of TBInCPNL was harvested at 72 h (OD565=8.28, Cmag=1.07). Cells were separated from the culture medium by centrifugation. The medium supernatant was then concentrated to approximately 25 mL, containing proteins at a concentration of 9.05 mg/mL. The proteins in the supernatant showed no binding activity to TBBPA or its hapten T5 conjugated with HRP (T5-HRP) by ELISAs, suggesting that Nbs were hardly secreted into medium. Afterwards, the cascade-amplified lysis of cells suspended in PBS (with Ca2+) was carried out using CHCl3. Magnetosomes were separated from cell broth under a magnetic field and cleaned up by washing with PBS (pH 7.4). The resultant magnetosomes exhibited a strong binding activity to T5-HRP by a non-competitive ELISA (Fig. 6a), and to TBBPA by a competitive ELISA (Fig. 6b). The results indicated the attachment of Nbs to magnetosomes. The cell broth debris was removed via centrifugation. The binding activities of proteins in the supernatant and precipitant to antigens (TBBPA or T5-HRP) were evaluated. No obvious binding activities were observed. Then the ratios of Nbs in supernatant (soluble proteins) and precipitant (insoluble proteins) with whole cell were further analyzed at two different stages: 1) In exponential phase (OD565=8.28, Cmag=1.07), only ~ 2% Nbs were detected in the supernatant and ~8% Nbs were detected in the precipitant (Fig. S4a); 2) In decline phase (OD565=18.72, Cmag=0.54), only ~ 1% Nbs were detected in the supernatant and ~14% Nbs were detected in the precipitant (Fig. S4b). These results supported that the majority of Nbs was attached to magnetosomes.
In the present study, the Nb was immobilized on the surface of magnetosomes, via a Pab PolⅡ intein as a bridge of Nb and MamC, to form a fusion protein on magnetosome, MamC-Intein-Nb, which replaced native MamC (Fig. 6c). Hence, the ratios of Nb is equal to native MamC on the magnetosome. Based on a LC-MS/MS analysis, the content of MamC in membrane proteins of magnetosome was 3.5 %. The total proteins of magnetosome membrane extracted from 1 mg magnetosome complexes were ~3.3 μg (ww). Therefore, the amount of Nbs in 1 mg magnetosome complexes (ww) was calculated as 0.12 μg. Pab PolII intein could promote protein splicing in vitro at high temperature [36]. The temperature controlled self-splicing of intein would be accompanied with the separation of Nbs from magnetosomes. Under the optimized cleaving conditions: 0.1 g magnetosome (MamC-Intein-Nb) complexes were suspended in 300 µL of PBS (pH 6.0) containing 200 mM DTT and incubated at 50 °C for 30 min. Nbs were detected in PBS by SDS-PAGE analysis (Fig. 6d) and showed binding activities to T5-HRP and TBBPA (Fig. 6e and f). Magnetosome complexes showed a slight binding capability to TBBPA after the splicing treatment (data not shown), indicating that Nbs were split from the complexes. The yields of Nbs released from 1 mg magnetosome complexes were approximately 0.11 μg. Thus, the recovery of Nbs was 92%.