Analysis and application of Bacillus subtilis sortases to anchor recombinant proteins on the cell wall

Bacterial strains and culture conditions

Construction of strains NDH20 and NDH21

To measure expression of the yhcS and ywpE genes, transcriptional fusions between their promoter regions and the lacZ reporter gene were constructed. The 5' coding region of yhcS including the start codon was amplified using the primers ON59 and ON60 (Table 2), treated with Eco RI and Bam HI and ligated into the integration vector pMUTIN4 (Vagner et al. 1998), cleaved with the same enzymes resulting in pNDH26. In a second experiment, the complete ywpE gene including its start codon was amplified using the ON61/ON62 primer pair and inserted into pMUTIN4 yielding pNDH27. Both plasmids were transformed into B. subtilis 1012 resulting in the strains NDH20 and NDH21, respectively (Figure 1). Correct integration at the yhcS locus was confirmed by PCR using ON57 and ON63 and at the ywpE locus using ON55 and ON63 (Figure 1A). These PCR products were verified by sequencing using ON63. One correct transformant each was kept for further studies.

Name	Sequence (5' to 3')	Description
ON29	GGCCATGGATCC ATGATTCAAAAACGAAAGCGGACAG	5' end of amyQ
ON42	GGCCATGACGTC TTTCTGAACATAAATGGAGACGGAC	3' end of amyQ
ON47	GGCCATGACGTC TTGGAAGCGACAGTTGAGTACG	5' end of yhcR
ON48	GAATAAGATATC TCACGTTCTGGAGGCGCTCCT	3' end of yhcR
ON49	GGCCATGACGTC CGCATGTTTGATATTGAAGAAGC	5' end of yfkN
ON50	AGCAGCGATATC TTATGCCTGATTCGCTCTATTCTG	3' end of yfkN
ON54	GGCCATTTCGAAGACCTCTTTAGCTCCTTGGAAGC	3' end of erm
ON55	GACCTGAATGTGGAACGAGTGGAC	5' end of ywpF
ON56	GGCCATTTCGAACCGACTGTAAAAAGTACAGTCGGCA	3' end of cat
ON57	CGTCTTGATCAGGATACATCTGGC	5' end of yhcT
ON58	GAGAGCCATAAACACCAATAGCCTT	5' end of neo
ON59	GGCCATGAATTC AAAGGAGGAACTCCAGAACGTGAAAAAAGTTATTC	5' end of yhcS
ON60	CTAATACGACTCACTATAGGGAGAGGATCC CGACACCTTTTTCTAAATCA	3' end of yhcS
ON61	GGCCATGAATTC AAAGGAGGAACAACAATGCGCCGGGATCA	5' end of ywpE
ON62	CTAATACGACTCACTATAGGGAGAGGATCC TCTTCGTGCTTCACTCTTGC	3' end of ywpE
ON63	TCTACATCCAGAACAACCTCTGC	5' end of Pspac
ON64	GGCCATAGATCT ATGCGCCGGGATCAAAAAATG	5' end of ywpE
ON65	GGCCATAGATCT ATGAAAAAAGTTATTCCACTATTCATCATTGC	5' end of yhcS
ON66	GGCCATAGATCT AGAATGAAGAAAAGCCGCAGGCACT	3' end of yhcS
ON67	CCAGAGATCT CAAAGGAGGAACTCCAGAACGTGAAAAAAGTTATTC	5' end of yhcS
ON68	AGTAAAGTTATCGGAATCGACTTAG	5' end of dnaK
ON69	CTAATACGACTCACTATAGGGAGAAAAGTATGCAGGAACTGTGAT	3' end of dnaK

Restriction sites used for cloning are underlined

Construction of strains SZ59, SZ60, NDH30, NDH31 and NDH32

To inactivate the genes coding for the two putative sortases, their coding sequences were replaced by two different antibiotic resistance markers. To obtain this goal, yhcS was replaced by a chloramphenicol resistance marker resulting in strain SZ59 (yhcS::cat) and ywpE by an erythromycin resistance marker (SZ60: ywpE::erm) as shown in Figure 2A and 2B. To be able to use these knockout strains with plasmids that carry a chloramphenicol resistance gene, the cat cassette in strain SZ59 was replaced by a neo cassette. First, the cat 5-neo-cat 3 cassette was cloned into plasmid pBluescript II KS resulting in plasmid pB-cat 5-neo-cat 3. This plasmid was treated with Pvu II and then transformed into strain SZ59, neomycin-resistant colonies were screened for chloramphenicol sensitivity, and correct integration at the cat cassette was confirmed by PCR using ON57 and ON58 (data not shown), and one transformant was kept for further studies (NDH30). Second, chromosomal DNA of strain SZ60 was transformed into the strain NDH30; recombinants were selected on LB plates containing erythromycin and neomycin. Correct integration at the ywpE locus was confirmed by PCR using ON54 and ON55, resulting in strain NDH31. Strain NDH32 was generated by transformation of chromosomal DNA of NDH31 into WB800, followed by selection for chloramphenicol, neomycin and erythromycin resistance (strain NDH32).

Construction of B. subtilis strain WB800N

WB800 (Wu et al. 2002) is an eight-fold protease-deficient B. subtilis strain that is used for the production of secreted heterologous proteins. This strain is resistant to chloramphenicol. To be able to use it with plasmid pNDH33 derivatives all carrying a chloramphenicol resistance gene (Phan et al. 2006), a neo cassette was inserted in the middle of cat cassette resulting in strain WB800N. The Pvu II-treated plasmid pB-cat 5-neo-cat 3 was transformed into WB800 and plated on indicator medium, calcium caseinate plates with neomycin. Colonies without halos (as compared with strain 1012) were checked for sensitivity to chloramphenicol and resistance to neomycin. One transformant was kept for further study (WB800N).

Construction of plasmids able to overexpress the two putative sortases separately and together

To be able to overexpress yhcS and/or ywpE in B. subtilis under the control of the IPTG-inducible promoter P _grac (Phan et al. 2006), three different plasmids were constructed. First, the coding sequence of the ywpE gene including its start codon was amplified by PCR using ON64 and ON62, the amplicon was cleaved with Bam HI and Bgl II and ligated into pNDH33 (Phan et al. 2006) at its unique Bam HI site resulting in pNDH33-ywpE. The ywpE gene was transcriptionally fused to P _grac and a strong ribosome-binding site (RBS) present on pNDH33. Next, gene yhcS was amplified using ON65 and ON67 containing its own RBS; the PCR product was then cleaved with Bgl II and ligated into pNDH33-ywpE resulting in pNDH33-ywpE-yhcS. The gene yhcS was also amplified using ON65 and ON66, the amplicon was treated with Bgl II and ligated into pNDH33 at its unique Bam HI site resulting in pNDH33-yhcS.

Construction of plasmids pNDH88, pNDH89 and pNDH90

In order to study whether the putative B. subtilis sortases could recognize potential sorting sequences, two plasmids that allow anchoring of amyQ coding for an α-amylase (Palva 1982) on the cell wall were constructed. In a previous report, it has been suggested that a 123-amino-acids spacer between AmyQ and the sorting sequence is optimal to anchor AmyQ on the cell wall (Nguyen and Schumann 2006). Therefore, plasmids were generated, in which amyQ was translationally fused to the putative sorting sequences with the 123-amino-acids spacers encoded by yhcR (YhcR123) and yfkN (YfkN123) under the control of the IPTG-inducible promoter Pgrac. First, the amyQ gene was generated by PCR using pKTH10 (Palva 1982) as template together with ON29 and ON42, the amplicon was treated with Bam HI and Aat II and ligated into pHT01 (Nguyen et al. 2007) cut with the same enzymes resulting in pNDH88. Next, the coding regions of the 3' ends of yhcR and yfkN including the sorting motif and the additional 123 codons, the spacer regions, were amplified using ON47/ON48 and ON49/ON50, respectively. The amplicons were cleaved with Aat II and Eco RV and inserted into pNDH88 treated with Aat II and Sma I resulting in pNHD89 and pNDH90, respectively.

Determination of sortase-dependent cell wall proteins

The B. subtilis strains were inoculated to an OD₅₇₈ of 0.05 - 0.08 in LB medium and grown at 37°C in a shaking water bath. After 1 h of growth, 0.1 mM IPTG was added to induce expression of yhcS and/or ywpE and cells corresponding to 200 OD₅₇₈ units were collected after about 8 h. After sedimentation by centrifugation, the cells were resuspended in 1.5 ml of water (final volume) containing a cocktail of protease inhibitors (Roche Diagnostics), 2 mM EDTA and 100 mg/ml DNase I and disrupted by sonication (12 W, 10 × 30 pulses with 30 sec intervals) on ice. The unbroken cells were removed by low-speed centrifugation (980 × g) at 4°C for 10 min. The supernatants were then centrifuged at higher speed (21 000 × g) at 4°C for 15 min to obtain a pellet containing the envelope material. These pellets were washed three times in water containing protease inhibitors. Finally, the pellets containing peptidoglycan with cell wall proteins were resuspended in 100 μl of lysozyme (1 mg/ml), incubated at 37°C for 45 min and shaken occasionally. Samples were mixed with 3 × loading buffer and applied to SDS-PAGE (Figure 3). The target protein bands were extracted from the gel, and proteins were identified by MALDI-TOF mass spectrometry.

Enzyme assays

B. subtilis strains NDH20 and NDH21 (Figure 1A) containing the transcriptional fusions PyhcS-lacZ and PywpE-lacZ were grown in LB medium at 37°C. When an OD578 of 0.6 was reached (set as t = 0) and samples were collected at the indicated time points. β-Galactosidase activity assays were performed in triplicate with soluble extracts using α-nitrophenyl-β-D-galactoside as substrate (Miller 1972) and yielded comparable results. The activities of one representative experiment are presented. β-Galactosidase activities are given in units, where one unit is defined as ΔA₄₀₅ min^-1 × OD₅₇₈ ^-1 × 10^-3, in which OD₅₇₈ is the optical density of the growth culture.

To measure α-amylase activity, the B. subtilis strains 1012/pNDH37, 1012/pNDH37-amyQ, NDH30/pNDH89, NDH03/pNDH19 and 1012/pNDH89, three different clones for each strain were grown in LB medium containing chloramphenicol (10 μg/ml) at 37°C. When the OD₅₇₈ of the cultures reached to mid-log phase (OD₅₇₈ 0.6), 0.5 mM IPTG and 0.5% xylose were added to all cultures to induce production of sortase A in the NDH03 strains, amylase (AmyQ, from pNDH37-amyQ) and the hybrids AmyQ-FnbpB123 (from pNDH19) and AmyQ-YhcR123 (pNDH89). Cells were separated from the growth medium by centrifugation, washed twice with the medium and once with PBS buffer (pH 7.4) and finally resuspended in PBS buffer. Cells corresponding to OD₅₇₈ of 0.2 in 100 μl were used to measure the α-amylase activities. As a control, the enzymatic activity secreted in the supernatant from the strain 1012/pNDH37-amyQ was also determined and was set at 100%.

RNA extraction and Northern blot analysis

B. subtilis cells were grown and induced as described under enzyme assays. Strains containing plasmids pNDH33-yhcS and pNDH33-ywpE-yhcS were induced by 0.1 mM IPTG, and the cells were killed by addition of "killing buffer" (5 mM MgCl₂, 20 mM NaN₃, 20 mM Tris-HCl; pH7.5). Total RNA was extracted using the protocol for isolation of RNA from yeast with modification (Robert 1998). The cell walls were digested by addition of lysozyme (1 mg/ml) on ice and the samples were then heated at 95°C for 5 min before addition of phenol. The RNA concentration was measured at 280 nm and 10 μg of total RNA was loaded in each well. Northern-blot analyses were performed as described (Roche Company 2003) with antisense RNAs produced against the putative sortase mRNAs. Hybridizations specific for the putative sortase genes were carried out with digoxigenin (DIG)-labelled riboprobe RNAs synthesized in vitro with T7 RNA polymerase from PCR products equipped with a promoter recognized by that polymerase (DIG RNA labelling kit; Roche Diagnostics, Mannheim, Germany). Pairs of primers ON59/ON60 and ON61/ON62 were used to amplify an internal part of the yhcS and the complete ywpE gene, respectively. The ON68/ON69 primers were used to amplify dnaK used as a loading control (Homuth et al. 1999).

Natural expression of yhcS and ywpE

B. subtilis yhcS codes for a putative sortase of 198 amino acids with one predicted transmembrane domain, while ywpE encodes a predicted cytoplasmic protein of only 102 amino acid residues (Comfort and Clubb 2004; Pallen et al. 2001). The latter exhibits 23% sequence identity with the C-terminal domain of SrtA. To follow expression of the two genes during growth, each promoter was fused to the lacZ reporter gene (Figure 1A), and the β-galactosidase activity was measured during growth. First, expression of the ywpE gene turned out to be lower than that of the yhcS gene during exponential and early stationary phase, but both activities were comparable during late stationary phase (Figure 1B). Second, expression of both genes increased over time to be highest during late stationary phase. We conclude from this result that both putative sortase genes are preferentially expressed after cells have entered the stationary phase.

Next, we analyzed transcription of the two genes in the B. subtilis 1012 wild type strain directly by Northern blot to confirm these results. Total RNA was isolated at different time points during growth and hybridized against gene-specific DIG-labelled anti-sense RNA. A transcript of about 4.5 kb could be detected after 6 h of growth with a further increase at 8 h (Figure 4). The 4.5-kb transcript corresponds by size to the bicistronic yhcR-yhcS operon (Price et al. 2005). The smaller bands below the 4.5-kb transcript most probably represent processing or/and degradation products. The failure to detect a ywpE-specific transcript even when using a large amount of RNA (30 μg) could indicate instability (data not shown). When both genes were expressed artificially from an IPTG-inducible promoter, their bicistronic transcript was produced in high amounts (Figure 4, left lane) indicating full stability under these conditions. In conclusion, putative sortase yhcS gene is expressed preferentially at the late stationary phase while ypwE expression could not be detected in the Northern blot analysis. This might point to a role of at least one of these two enzymes (YhcS) in anchoring proteins during stationary phase to the cell wall.

Search for yhcS and/or ywpE-dependent surface proteins

To identify putative sortase-dependent substrate proteins, the three knockout strains SZ59 (ΔyhcS), SZ60 (ΔywpE) and NDH31 (ΔyhcS and ΔywpE) were constructed (Figure 2A and 2B). All three mutant strains together with the isogenic wild-type strain were incubated in LB medium for 8 h corresponding to the late stationary phase and analyzed for the presence of cell wall anchored proteins as described under Materials and methods. No difference in the protein pattern could be detected (data not shown). We conclude that the amount of proteins anchored is not sufficient to be detected either due to the low amount of sortase enzymes or due to these two enzymes, or due to a mixture of both. Therefore, we decided to repeat this experiment with strains, where either ywpE or yhcS or both genes could be expressed together using the IPTG-inducible promoter, Pgrac from plasmid pNDH33. Expression of these genes was analyzed by Northern blot. While induced expression of yhcS and ywpE yielded the expected RNAs of about 1 and 0.5 kb, respectively, the artificial bicistronic operon led to the detection of an RNA larger than 1 kb (Figure 2C). These experiments clearly demonstrate that both genes can be expressed if fused to a strong promoter.

Interestingly, strains that restored expression of ywpE and/or yhcS exhibited two YhcS-dependent proteins with molecular weights between 130 kDa and 170 kDa which appeared in the strains that express yhcS (Figure 3, lanes 2 and 4). In addition, when the strain WB800N, deficient for eight different proteases (Wu et al. 2002), carrying plasmid pNDH33-ywpE-yhcS was analyzed several bands became visible. Among them a band running with a molecular mass of 140 kDa seems to be a doublet (Figure 3, lane 6). These protein bands were then extracted and proteins were determined by MALDI-TOF mass spectrometry. As we expected one of these proteins is YhcR and the other is YfkN, both containing the secretional sequence and potential sorting signal.

Displaying AmyQ on the surface using the sorting sequences

Next, we asked whether the two putative sortases YhcS and YwpE could anchor the proteins YhcR and YfkN on the surface of B. subtilis using their sorting signal. We fused the amyQ-encoded α-amylase to the potential sorting signals of the two proteins together with an 123-amino-acid spacer region resulting in pNDH89 (AmyQ-YhcR123) and pNDH90 (AmyQ-YfkN123). These plasmids were transformed into strains SZ60 (ΔywpE), NDH30 (ΔyhcS), NDH31 (Δyhc, ΔywpE) and 1012. To determine the amylase activity on the cell surface, strain NDH03/pNDH19 that has been described to immobilize amylase on the surface (Nguyen and Schumann 2006) and strain 1012/pNDH37-amyQ (Phan et al. 2006) that secretes the amylase into the culture medium were used as positive and negative control, respectively. Cells of these strains were grown as described under Materials and methods for Western-blot (Figure 5), and samples were collected at the appropriate time points for measuring the amylase activities (Figure 6).

When the α-amylase carrying the YfkN123 sorting sequence was tested, anchored protein was hardly detected in the strain expressing either yhcS or yhcS and ywpE 8 h after induction (data not shown). When the YhcR123 motif was tested, a strong anchoring occurred in the presence of YhcS with some further increase upon additional synthesis of YwpE (Figure 5, 8 h and 12 h). But since a substantial amount of AmyQ-YhcR123 is already present in the absence of both putative sortases, these hybrid protein molecules might be retained in the cytoplasmic membrane due to the presence of a hydrophobic region being part of the sorting sequence. The presence of α-amylase attached to cells in the absence of potential sortases could also be observed with whole cells 4 h after induction (Figure 6 NDH30/pNDH89, 5% activity) when compared with the negative controls (0% activity for both 1012/pNDH37 and 1012/pNDH37-amyQ), and the positive control (76% activity for NDH03/pNDH19). Additionally, α-amylase activity of the sample that produces both potential sortases and the hybrid protein, AmyQ-YhcR123 (Figure 6, 1012/pNDH89) was as high as the positive control (Figure 6, NDH03/pNDH19); and the same results could be measured for samples collected after 2 h and 8 h of induction (data not shown). This activity-based measurement confirmed that the fusion YhcR123-AmyQ could be displayed on the surface of B. subtilis. In summary, these results strongly suggest that the yhcS gene codes for a true sortase able to anchor at least YhcR on the cell wall of B. subtilis cell and we suggest renaming it to srtD. This work could also propose an alternative way to immobilize a heterologous protein on the cell wall of B. subtilis using a fusion form of YhcR sorting sequence.