Lactococcus lactis secreting phage lysins as a potential antimicrobial against multi-drug resistant Staphylococcus aureus

Bacterial strains, plasmid and conditions of growth

The plasmid and strains used in this study are listed in Table 1. L. lactis, NZ9000 strains were cultured in M17 (Merck) agar and broth (Terzaghi & Sandine, 1975) supplemented with 0.5% (w/v) glucose (GM17) and 7.5 µg/mL chloramphenicol whenever needed. The L. lactis NZ9000 strains were incubated statically at 30 °C for 16-18 h. When screening L. lactis NZ9000 transformants, M17 agar supplement with 0.5% (w/v) glucose, 0.5 M sucrose and 7.5 µg/mL chloramphenicol was used. L. lactis NZ9000 harbouring pNZ8048 was grown in GM17 medium supplemented with chloramphenicol, Cm to a final concentration of 7.5 µg/mL. For the secretion of endolysin (NCBI GenBank Accession number: YP_240699) and VAPGH (NCBI GenBank accession number: YP_240695), SPK1, a signal peptide from Pediococcus pentosaceus (Baradaran et al., 2013) was used. Both lysins are from Bacteriophage 88 (ATCC 33742-B1), and were tested primarily against its host strain S. aureus subsp. aureus PS 88 (ATCC-33742). Top 10 E. coli strain (Invitrogen, USA) was grown in LB broth and cultured at 37 °C at 250 rpm. When screening for E. coli transformants, LB agar supplemented with ampicillin (100 µg/mL) of 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) was used. All Staphylococcus sp. and Streptococcus agalactiae strains were cultured in BHI (Brain Heart Infusion) broth or agar. S. aureus and S. epidermidis were incubated at 37 °C with agitation at 180 rpm while S. agalactiae species were grown at 30 °C without agitation.

Antibiotic susceptibility of S. aureus PS88

The antimicrobial susceptibility test was performed by disc diffusion technique using Mueller-Hinton agar (Merck, Darmstadt, Germany) according to the CLSI guidelines (CLSI , 2016). Antibiotics tested were cefoxitin, penicillin, tetracycline, streptomycin, gentamicin, erythromycin, amoxicillin, vancomycin, kanamycin, meropenem, cefotaxime and chloramphenicol.

Bioinformatics analysis

The sequence of VAPGH and endolysin (named VAH88 and Endo88, henceforth) were obtained from the GenBank database (accession numbers YP_240695 and YP_240699) for multiple sequence alignment and homology modeling. The amino acid sequences were then compared with other well-studied staphylococcal phage endolysins by MAFFT (Multiple Alignment using Fast Fourier Transform) and the amino acid sequence similarity and identity were determined using Needleman-Wunsch pairwise alignment in EMBL-EBI server (https://www.ebi.ac.uk/Tools/psa/emboss_needle/) (Madeira et al., 2019). Then, the conserved domain was predicted by NCBI Conserved Domain Database (NCBI CDD) and Pfam database. The tertiary structure of the Endo88 was generated by MODELLER 10.1 (Sali & Blundell, 1993; Fiser, Do & Sali, 2000) using a template retrieved from the SWISS-MODEL online server (https://swissmodel.expasy.org/). Twenty models were generated from each run during optimization and 5 models with the lowest Discrete Optimized Protein Energy (DOPE) score was selected for further evaluation (Shen & Sali, 2006). PROCHECK (Laskowski et al., 1993) was used to evaluate the quality of the top 5 models. The tertiary structure of Endo88 was subsequently visualized by the PyMOL Molecular Graphics System, Version 1.2r3pre, Schrödinger, LLC. (https://pymol.org/2/).

Vector construction

The VAH88 gene of S. aureus bacteriophage 88 was amplified using oligonucleotides F-VAH88 and R-VAH88 (Table 2) while the SPK1 signal peptide was amplified using F-SPK1 and R-SPK1. The VAH88 gene was subjected to single digestion with BamHI at the N-terminus and ligated in frame with the signal peptide SPK1, digested with the same restriction enzyme (RE) at the C-terminus to obtain the full fusion gene. The ligation of the fusion gene (SPK1-VAH88) was amplified using F-SPK1 and R-VAH88. The resulting PCR product was subjected to double digestion with Xba I and Pst I and a His-tag sequence was also incorporated into the reverse primer, R-VAH88 to generate a C-terminal His-tag fusion protein. The fusion protein was cloned into pNZ8048 to generate the recombinant plasmid pNZ-SPK1-VAH88. The electroporation method was used for transformation into L. lactis competent cells (Holo & Nes, 1989). Clones containing the ligated plasmids were selected using chloramphenicol. A similar method was applied to amplify and clone the SPK1-Endo88 gene into L. lactis NZ9000 competent cells. Both recombinant plasmids were confirmed by standard restriction enzyme digestion and DNA sequencing analysis.

Protein expression and purification

For protein expression and secretion of Endo88 and VAH88 by L. lactis NZ9000, both intracellular and extracellular protein fractions were analyzed. Overnight culture of recombinant NZ9000 (pNZ-SPK1-VAH88/Endo88) were sub-cultured into fresh GM17 medium and grown to an optical density at 600 nm (OD₆₀₀) of 0.4 and induced with 10 ng/mL nisin A (Sigma Aldrich, USA) for 12 h at 30 °C. Non-induced cells were used as a negative control. Bacterial cells were harvested (4,000× g, 10 min, 4 °C) and the pellet was resuspended in 1X PBS buffer and the cells subjected to sonication using Omni Ruptor 4000 (Omni International, GA, USA) for 1 min by a 30 s pause. This was repeated 10 times. The extracellular protein was extracted and concentrated using 1/10 culture volume of 100% trichloroacetic acid for 2 h. Then, the protein was pelleted down by centrifugation and was washed twice using ice cold acetone, dried and resuspended in 1X PBS. Both intracellular and extracellular proteins were quantified by Bradford assay and subjected to SDS-PAGE and Western blot analysis with mouse anti- His Tag^® monoclonal antibody (Novagen, USA). The secondary antibody used was Goat Anti-Mouse IgM Horseradish Peroxidase Conjugated (Thermo Scientific, USA) and detection performed using Advansta Reagent (Matrioux, USA).

For recombinant protein purification, the intracellular and extracellular VAH88 and Endo88 samples were purified, respectively, using complete™ His-Tag Purification Resin following the manufacturer’s instruction (Merck Darmstadt, Germany). For the intracellular protein, the culture of VAH88 and Endo88 were induced, and the cells sonicated and centrifuged as described earlier. Then supernatants were purified using a gravitational dripping method. For the extracellular protein, the recombinant expression cultures of Endo88 and VAH88 were harvested by centrifugation and the supernatants (spent media) were filtered using a 0.22 µm syringe filter. Then supernatants were purified using the same gravitational dripping method as above. The buffer in the final eluted fraction for both intracellular and extracellular of VAPGH and endolysin samples were then exchanged for Tris-HCl (50 Mm NaCl, pH 7.5) using a Pierce™ Protein Concentrator with a polyethersulfone membrane and a 5 K molecular cutoff (Thermo Fisher Scientific, Waltham, MA, USA). The concentration of purified proteins was determined by spectrophotometry using the Bradford assays. Each elution fraction was collected and analyzed by SDS-PAGE analysis.

Plate assays to assess antimicrobial activities of recombinant VAPGH and endolysin

The lytic activity against S. aureus PS 88 cells on agar plate were determined using a method previously described with slight modifications (Delisle, Barcak & Guo, 2006). Plates were prepared as follows: PS 88 was grown in 10 mL of TSB to the mid-log phase (OD_600nm 0.4–1.0). Then, the cell culture was harvested by centrifuging at 7,000× g for 10 min at 22 °C. The supernatant was discarded while the pellets were resuspended with one mL of 1X PBS buffer. Next, 1% agarose gel was used in the assay and was prepared by dissolving 1.0 g of agarose powder (Vivantis, Malaysia) in 50 mL of 1X PBS and 50 mL of Tryptic Soy broth (TSB). The mixture was heated using a microwave oven until fully dissolved. After the gel was slightly cooled, 10 μL of the PS 88 was added into the gel and then swirled slowly to mix well. Then, the mixture was poured into three different plates for triplicates and eight small wells were punched into the seeded agar. Finally, 10 µL of intracellular or extracellular protein extract with varying concentrations (1, 5, 10, 25, 50 and 100 µg) were loaded into the wells. Subsequently, the plate was incubated at 37 °C for 24 h. Plates were checked for lysis zones the next day. Ten microlitre of 1X PBS buffer was also spotted onto the plate as negative control while 5 µL chloroform was used as a positive control for halo zone formation. These experiments were performed in triplicates. Plate assays were also performed using 50 ug protein on six different kinds of clinical MRSA strains previously isolated from Hospital Serdang, Selangor, Malaysia (MRSA 6, MRSA 7, MRSA 8, MRSA 12, MRSA 10 and MRSA 20), Staphylococcus epidermidis, Streptococcus agalactiae serotype 2, S. agalactiae serotype 3, E. coli and L. lactis NZ9000 to test the host range of the recombinants.

Turbidity reduction assays of PS 88 by recombinant VAH88 and Endo88

To assess the ability of the recombinant VAH88 and Endo88 to lyse PS 88 strains quantitatively, the extracellular purified protein sample was tested in turbidity reduction assays. PS 88 cells were grown to log phase (OD₆₀₀ nm = 0.4–0.6) at 37 °C in Tryptic Soy broth. Next, the culture was centrifuged, and the pellet was resuspended in a reaction buffer consisting of 100 mM Tris-HCl buffer pH 7.5 and the OD₆₀₀ nm was adjusted to 0.75 to normalize the number of cells. To measure lytic activity, 100 µL of each purified protein of extracellular protein sample at the concentration of 50 µg/mL was mixed with cell substrate suspension and the change in OD₆₀₀ nm was recorded every 5 min for 1 h. Untreated PS 88 strain and PS 88 treated with protein from uninduced cells were used as a negative control. These experiments were performed in triplicates. Experiments were carried using a method that was previously described with slight modifications (Kashani et al., 2017b).

Antimicrobial activity of recombinant L. lactis secreting phage lysins

Recombinant L. lactis expressing pNZ-SPK1-VAH88 and pNZ-SPK1-Endo88 were grown in media. Next, the cells were pelleted at 4,000× g for 10 min by centrifugation. Spent media (9-mL) was sterilized by filtering through a 0.22 µm filter and neutralized by adding 1.5 mL of 1 M Tris-HCl (pH 9). Mid exponential phase growing PS 88 cells was washed with TSB broth and then resuspended into the same medium. Ten microliters of PS 88 cells were inoculated into the spent media from the earlier lactococcal cultures. The growth of the cells was followed for 12 h at 30 °C, by determining viable counts using TSB agar. These experiments were performed using three biological replicates and three technical replicates each. Experiments were carried using a method previously described with slight modifications (Turner et al., 2007).

Statistical analysis

The statistical significance for all histograms shown was set at the threshold of P < 0.05 and was assessed using the Student’s t-test.

Antibiotics susceptibility testing of S. aureus PS 88

Bacteriophage 88 was purchased from ATCC together with its host S. aureus PS 88. According to ATCC, bacteriophage 88’s application is for the assay of MRSA strains. However, upon testing, its host S. aureus PS 88, a clinical isolate from New York, was found to be susceptible to cefoxitin (alternative to methicillin). However, it was resistant to multiple other antibiotics including penicillin, tetracycline, streptomycin, gentamicin and erythromycin but was susceptible to amoxicillin, vancomycin, kanamycin, meropenem, cefotaxime and chloramphenicol, other than cefoxitin. Since multi-drug resistant bacteria is defined by resistance to at least three classes of antibiotics (Neyra et al., 2014), PS 88 can be considered a MDRSA. Penicillin is a β-lactam antibiotic, tetracycline is a class of its own, streptomycin and gentamicin are aminoglycosides while erythromycin is a macrolide.

Bioinformatics analysis

The endolysin of the stapylococcal phage 88 is encoded in ORF006, consisting of 491 amino acids. NCBI CDD classifies Endo88 in Peptidoglycan recognition proteins (PGRPs) and src-homology 3 (SH3_5) domain containing protein. It consists of a Cysteine, Histidine-dependent Amidohydrolase/Peptidase (CHAP) domain at amino acid residues 31–113 (Accession number: pfam05257; E-value: 2.96e−08), an Amidase (Ami_2) domain at amino acid residues 198–322 (Accession number: smart00644; E-value: 1.05e−24), and a SH3_5 domain at amino acid residues 395-460 (Accession number: pfam08460; E-value: 2.03e−21). On the other hand, the VAH88 is encoded in ORF004, consisting of 624 amino acids and classified as a glucosaminidase domain-containing protein. It consists of a CHAP domain at amino acid residues 31-119 (Accession number: pfam05257; E-value: 6.09e−05), and a Beta N-acetylglucosaminidase (LytD) domain located at residues 431–624 (Accession number: COG4193; E-value: 9.65e−68).

Multiple sequence alignment was performed to compare Endo88 and VAH88 with other well-studied staphylococcal phage lysins (Fig. 1). Endo88 has high amino acid sequence similarity with MVL (BAF33253.1, 99.8%) (Rashel et al., 2007) and LysH5 (YP_002332536.1, similarity: 96.5%) (Rodríguez-Rubio et al., 2012a) but not with LysGH15 (ADG26756.1, 50.6%) (Gu et al., 2011), LysK (YP_009041293, 50.8%) (Becker, Foster-Frey & Donovan, 2008) and Twort (AAX92311,55.3%) (Becker et al., 2015) (Fig. 1A). However, the amino acids forming the catalytic site in CHAP (C32, H95, E111 & N113) and the peptidoglycan binding site (N390, Y392, G393 & T394) are conserved (Gu et al., 2014). On the other hand, VAH88 has high amino acid sequence similarity with another two previously reported VAPGH proteins (BAF79638.1, 99.5%; YP_002332533.1 HydH5, 71.9%) (Rashel et al., 2008; Rodríguez-Rubio et al., 2012b) (Fig. 1B). The LytD domain, including the catalytic site (E509, F528, Y590 and W596) is highly conserved in these VAPGH proteins. The catalytic site for the CHAP domain (C38 and H11) is also highly conserved.

Homology modelling for Endo88 and VAH88 was performed using the SWISS-Model online server and MODELLER (Fig. 2). Three templates for each domain were chosen from SWISS-Model. The templates were 6ist.C (Identity: 21.68%) for the CHAP domain, 4ols.A (Identity: 51.22%) for the Amidase domain, and 2mk5.A for the SH3 domain (Identity: 45.08%). The quality of the model was evaluated by Discrete Optimized Protein Energy (DOPE) value and Ramachandran plot. The model with the lowest DOPE value and highest number of amino acids in the most favoured region was chosen for final loop refinement using MODELLER (Fig. S1). For VAH88, templates 6ist.C (Identity: 30.34%) and 6fxo.A (Identity: 42.48%) were retrieved from SWISS-MODEL for the CHAP and LytD domains respectively. The 3D structure of the amino acid residues between the domains were not built because there were no template structures available.

Cloning of endolysin and VAPGH gene into L. lactis NZ9000

The VAH88 and Endo88 genes of S. aureus bacteriophage 88 were ligated in frame with SPK1 secretion signal peptide and cloned into pNZ8048 to generate plasmids pNZ-SPK1-VAH88 and pNZ-SPK1-Endo88, respectively (Figs. S2A and S2B). His-tag was placed at the C-terminal of each sequence to facilitate the detection of the expressed proteins. The putative transformants of pNZ-SPK1-VAH88 and pNZ-SPK1-Endo88 were screened by plasmid extraction, single digestion and double digestion analyses on agarose gel electrophoresis (Figs. S2C and S2B). The expected product size for SPK1-VAH88 and SPK1-Endo88 genes were 1,988 bps and 1,557 bps, respectively (Figs. S2C and S2B) inclusive of the restriction enzymes (REs) sites and His-tag sequence. Both genes were successfully cloned into pNZ8048 and were further confirmed by sequencing.

Expression of SPK1-VAPGH and SPK1-Endolysin from L. lactis

Intracellular and extracellular crude protein extracts from NZ9000 (pNZ-SPK1-VAH88/Endo88) were analyzed by SDS-PAGE (data not shown) and western blotting (Fig. 3). Bands of expected sizes were seen for both clones. For clones harbouring pNZ-SPK1-VAH88, the expected sizes of intracellular and extracellular protein samples were observed at ∼74.3 kDa and ∼71.3 kDa (Fig. 3A and 3B) respectively, which were not found in the negative controls. For clones harbouring the pNZ-SPK1-Endo88, the expected sizes of intracellular and extracellular protein samples were observed at ∼57.3 kDa and ∼54.3 kDa (Fig. 3C and 3D), respectively, while no bands were expressed in the negative control. The size of the extracellular recombinant VAH88 and Endo88 is smaller compared to the intracellular protein as the SPK1 signal peptide was cleaved from VAPGH and endolysin respectively when it was secreted extracellularly to form the mature protein (Gueniche et al., 2009; Roslan et al., 2020). This analysis showed that both recombinant proteins were successfully expressed and secreted from the recombinant strains.

Protein purification

Purified protein sample of intracellular and extracellular fractions of VAH88 and Endo88 were verified using SDS-PAGE. The detection of intracellular (∼74.3 kDa) and extracellular (∼71.3 kDa) protein band confirmed the purification of VAH88 (Figs. 4A and 4B). For endolysin, the intracellular and extracellular purified proteins were detected at ∼57.3 kDa and ∼54. 3 kDa respectively. Although the intracellular fractions were not purified to homogeneity as other bands were still visible, we did not further purify the protein as the intracellular fraction was not active in the antimicrobial assays described below.

Plate assay analysis

The plate assay results (Fig. 5A) demonstrated that the intracellular crude protein of VAH88 did not show any halo zones but the extracellular crude protein of VAH88 exhibited antibacterial lysis activity against PS 88 cells. The same results were shown by Endo88. These results showed that the extracellular protein was functional after the SPK1 signal peptide has been cleaved, leaving the mature protein. The lytic activity appears to be dose-dependent, as the size of halo zone formation was parallel to the increasing amount of proteins. However, the lytic activity of extracellular VAH88 was not observed when using only 1 µg of protein in the well, indicating that the effective dose for VAH88 is higher than endolysin. It was also apparent that Endo88 formed bigger halo zones compared to VAH88 which was confirmed once the zone of inhibition was measured in triplicates (Fig. 5B). Chloroform was used as a positive control in the plate assay to show the inhibitory zone on PS 88 cells while PBS buffer was used as the negative control.

Turbidity reduction assays

Turbidity reduction assays were performed to quantitatively assess the antimicrobial activities of recombinant L. lactis against PS 88. In this experiment, only the extracellular proteins from both strains were tested since there was no inhibition zone observed for intracellular protein in the plate assay experiment. As shown in Fig. 6, both extracellular Endo88 and VAH88 were able to reduce the viability of PS 88 when compared to the control (untreated PS 88 and cells treated with uninduced Endo88 and VAH88). However, Endo88 was able to reduce the viability of PS 88 more compared to VAH88. Therefore, this result further supported the observation from the plate assays that endolysin showed greater inhibition towards PS 88. However, both Endo88 and VAH88 was unable to completely inhibit the growth of PS 88 cells as growth became stagnant after 75 min with no further reduction in OD. Meanwhile, untreated PS 88 and PS 88 treated with uninduced protein fractions continued growing and did not show any inhibition as expected.

Antimicrobial effect of recombinant L. lactis secreting endolysin and VAPGH

Since the ultimate aim of the study is to use whole live recombinant lactococcal cells as potential antimicrobials and not the purified proteins, the ability of the recombinant lactococcal cultures to inhibit growth of PS 88 was analyzed. The spent media of both induced and uninduced recombinant strains were inoculated with PS 88 and their growth observed for 12 h. Induced L. lactis recombinant strain expressing VAH88 and Endo88 were able to reduce the viability of PS 88 compared to PS 88 in both uninduced recombinant strains until 12 h (Fig. 7). As in the plate assay and turbidity reduction assay, L. lactis strain expressing Endo88 was able to significantly reduce the viability of PS 88 better than L. lactis expressing VAH88. Overall, treatment with spent media containing L. lactis secreted Endo88 resulted in a 3.5-log reduction of PS 88 over 12 h while treatment with spent media containing VAH88 only resulted in a 1-log reduction.

Lytic activity of crude extracellular endolysin and VAPGH against other bacterial strain

Plate assay was conducted using six different kinds of clinical MRSA strains (MRSA 6, MRSA 7, MRSA 8, MRSA 12, MRSA 10 and MRSA 20), S. epidermidis, S. agalactiae serotype 2, S. agalactiae serotype 3, E. coli and L. lactis NZ9000. From the observation, after incubation for 24 h at 30 °C with 50 µg protein, it was observed that extracellular Endo88 inhibited the growth of all tested MRSA strains as well as S. epidermidis which can be observed by the development of halo zone around the wells. Degree of inhibition varied as shown by the difference in zone of inhibition, depending on the strain (Fig. 8). However, none of the zones of inhibition were as big as for its own host, PS 88 which averaged at 17 mm (Fig. 5B). In addition, there were no inhibition zones observed against S. agalactiae serotype 2, S. agalactiae serotype 3, E. coli and L. lactis. Inhibition zones were not expected in L. lactis NZ9000 since it was used as the host to produce the recombinant phage lysins which did not seem to show any inhibitory effect on the host’s growth. As for VAH88, it did not show any zone of inhibition for all strains tested, inclusive of the other MRSA strains (data not shown) which meant that the VAH88 could only lyse its own host, PS 88.