Bacteriocin-like peptides encoded by a horizontally acquired island mediate Neisseria gonorrhoeae autolysis

We searched for pathogen-specific genes in Neisseria spp. to further understand the basis of the invasive phenotype of some species in this genus that usually colonises mucosal surfaces asymptomatically. Using MaGe [17], we identified a ~3.4 kb region in the gonococcal genome, which we designated the nap island (Figs 1A and S1). This region is found in N. gonorrhoeae, with a related locus present in N. meningitidis. However, the nap island is largely absent from noninvasive Neisseria spp. (S2 Fig). The GC content of the nap island (38%) is significantly lower than rest of the genome (53%, Fig 1A), indicating that it was likely acquired by horizontal gene transfer (HGT) from an extant source. Consistent with this, napC, a gene in the centre of the nap island, encodes a protein with 68.9% of identity with a protein from Inoviridae phages (S1 Fig); these phages are associated with Neisseria spp. as 57 Inoviridae sequences are found in Neisseria CRISPR spacers (GenBank accession number DAS81996.1). One end of the island is flanked by napP, encoding a potential C39 peptidase predicted to process GG-containing peptides prior to secretion [14]. At the other end of the island, napF encodes a homologue of FapF, an outer membrane protein in Pseudomonas aeruginosa involved in the secretion of amyloidogenic peptides processed by a C39 peptidase [16]. Of note, both napP and napF have a different GC content compared with the rest of the nap island (respectively, 52% and 50%), suggesting that these 2 genes might have a different origin from the central portion of the island.

Fig 1. The nap island organisation and the absence of mNaps antimicrobial activity.

(A) Organisation of the nap island in Neisseria gonorrhoeae FA1090. Grey arrows, flanking genes (NEIS0808, pseudouridine synthase; NEIS0794, histidyl-tRNA synthase); blue arrows, genes coding for GG-containing peptides; orange arrows, genes predicted to be involved in peptide processing and secretion; green arrow, gene of unknown function; pink boxes, Correia elements. Percentage of GC content was plotted with Snap Gene Viewer. (B) Table detailing predicted function of genes in the nap locus. Details are available in S1 Fig. (C) Conservation of the nap island and flanking genes among Neisseria spp. Colours are scaled in percent of strain representatives possessing a homologue within each species. The data underlying this figure can be found in S2 Table. (D) mNaps (50 μm) were added to disks which were then placed on lawns of N. gonorrhoeae FA1090 (Ng), N. cinerea (Nc) CCUG 346T or 27178A, L. crispatus (Lc) ATCC 33820, and E. coli (Ec) MG1655 on solid media. Water and polymyxin B (50 μm) were used as negative and positive controls, respectively. Plates were incubated for 24 h and zones of growth inhibition (dotted lines) recorded. No growth inhibition was observed for any mNap (n = 3). Source data is in the file S1 Data. (E) MICs were determined by incubating bacteria in rich medium for 24 h with antibiotics (controls) or mNaps, starting from 256 μg/ml to 0.5 μg/ml in 2-fold dilutions. Bacteria were then spotted onto agar plates, which were incubated for 24 h. MBCs were the lowest concentration at which no growth was visible on plates. The mNaps did not inhibit the growth of any strain (shown by /, n = 3). The data underlying this figure can be found in S1 and S2 Data files.

https://doi.org/10.1371/journal.pbio.3003001.g001

The gonococcal nap island contains 3 genes (napABC) coding for predicted α-helical cationic peptides with GG-motifs (Fig 1B). After cleavage at the GG sequence, mature mNapA, mNapB, and mNapC are predicted to contain 36, 16, and 31 amino acids, with pIs of 9, 10, and 10.1, respectively (S1 Fig). Downstream of napABC (Figs 1A and S1) is napI which shares 30% of amino acid identity with a domain in LagC, the immunity protein for the bacteriocin, lactococcin G [18] (S1 Fig). napR is upstream of napABC and encodes a peptide with a putative DNA-binding domain (S1 Fig) so is a potential regulator of the island. Finally, napH is predicted to encode an inner membrane protein with a high methionine content with 3 transmembrane α-helices seen in some bacteriophage pore-forming holins [19] (S1 Fig).

Of note, napR contains a poly-G tract of 6 to 12 residues in its 5′ region, indicating that it is likely to be phase variable (S1 Fig and S1 Table). In N. gonorrhoeae FA1090, napR has 7 guanines, the most common allele (in 6,548 out of 10,854 isolates), leading to loss of the GG-leader sequence, so the peptide is predicted to be cytoplasmic. In addition, napC is also probably phase-variable, as it contains a poly-T tract towards its 5′ end resulting typically in a cationic (9xT) or non-cationic (8xT) peptide (S1 Fig and S1 Table).

The distribution of the nap island within Neisseria spp. was examined in more detail using PubMLST [20]. Whole genome sequences of >24,700 Neisseria spp. isolates were inspected, with results confirming that most commensal species lack the nap island with 2 exceptions (Fig 1C). The nap island is found in a proportion of isolates of Neisseria bergeri (38/40 have napA) and Neisseria lactamica (138/604 have napA), but no other noninvasive species (Figs 1C and S2 and S2 Table for details).

Most class II microcins are antibacterial [21]. To determine whether the cationic Naps possess antibacterial activity, mature versions of the peptides, mNapA, mNapB, and mNapC, were synthesised and their ability to kill competitor bacteria assessed. First, we performed disc diffusion assays against N. gonorrhoeae, Neisseria cinerea, Escherichia coli, and the gram-positive bacterium Lactobacillus crispatus, an inhabitant of the genital tract [22]. As microcins are bactericidal in the nanomolar range [23], assays were performed using discs with 50 μm of each mNap, and with polymyxin B as a positive control. As expected [24], growth of N. cinerea and E. coli was inhibited around discs containing polymyxin B, while there was no clearance of N. gonorrhoeae or L. crispatus around these discs (Fig 1D). In contrast, none of the mNaps inhibited the growth of any strain. We also measured the minimal bactericidal concentration (MBC) of each mNap by broth dilution against the bacteria. Bacteria were incubated with micromolar concentrations of mNaps or polymyxin B for 24 h. Of note, the mNaps failed to inhibit bacterial growth even at the highest concentration (256 μm, Fig 1E).

As some bactericidal mechanisms are contact-dependent [25], we performed co-culture experiments to see whether the survival of prey bacteria (i.e., N. cinerea, E. coli, and L. crispatus) was affected in the presence of wild-type N. gonorrhoeae or an isogenic mutant unable to express Naps. We constructed a markerless N. gonorrhoeae FA1090 ΔnapRABC mutant (eliminating all Nap peptides) with a pheS* counter-selection marker [26] to avoid fitness costs associated with antibiotic resistance cassettes (S3 Fig). After 3 and 24 h of co-culturing prey and N. gonorrhoeae at a 1:1 ratio, we recovered prey strains by plating to selective media. No significant difference could be observed in the recovery of prey strains in the presence of wild-type or ΔnapRABC N. gonorrhoeae (multiple paired t tests, n = 3, S4 Fig). Taken together, these results demonstrate that Naps do not have detectable antimicrobial activity.

To better understand the function of the Naps, we examined the regulation of genes encoding nap island peptides as the expression of many microcins and bacteriocins is regulated during growth [13] (Figs 2A and S5A). The mRNA levels of genes encoding the nap peptides was measured by RT-qPCR of bacteria grown in liquid media over 24 h. The mRNA levels for napA, napB, napC, and napR were lowest in the early stationary phase of growth, then rose to their highest levels in the late autolytic phase. The similar gene expression profile of napA, napB, and napC suggests that they are organised in operon. This is consistent with the absence of a predicted promoter directly upstream of napB (S5B Fig) and the detection of a transcript overlapping all 3 genes in N. gonorrhoeae MS11 [27]. The nap locus of N. gonorrhoeae FA1090 also harbours Correia elements (CEs) in the napR/napA and napI/napH intergenic regions containing potential promoters [28] (Figs 1A and S5B); these CEs harbour integration host factor (IHF) binding sites (S5C Fig), which can influence transcriptional regulation [29].

As nap mRNA levels were elevated during the late stationary phase, we considered whether the Naps are involved in PCD. Gonococcal autolysis involves initial peptidoglycan remodelling by LtgA and other enzymes [4,30], followed by cell lysis. As polyamines can increase the resistance of bacteria to cationic peptides [31,32], we assessed the activity of mNaps on gonococcal survival in protein/spermidine-free media. mNaps were added to N. gonorrhoeae at the mid-log phase, and survival examined during the stationary phase until autolysis occurred [33]. N. gonorrhoeae was grown for 3 h in 96-well plates from a starting OD₆₀₀ of 0.25, then exposed to mNaps. While mNaps had no effect on survival of gonococci during the exponential phase of growth, micromolar concentrations of mNapA and mNapC significantly reduced survival when bacteria reached the late stationary phase of growth (Fig 2B, p N. cinerea at any stage of growth (S6A Fig), while scrambled versions of mNapA and mNapC (mNapA_SCR and mNapC_SCR, respectively) did not affect survival of N. gonorrhoeae (S6B Fig). We also checked whether mNaps interacted with each other in this assay. Mixtures of Naps were tested on N. gonorrhoeae and results compared with effect of individual mNaps. There was no evidence of antagonism or synergy between the Naps in these conditions (S6C Fig).

Fig 2. Regulation of the nap island and the effect of mNaps on bacterial survival.

(A) RT-qPCR of genes in the nap island from N. gonorrhoeae FA1090 grown in GW liquid cultures. Samples were taken from flasks at different times (CFU, left panel) and mRNA levels of target genes determined by RT-qPCR (right panel). Gene expression results were normalised to samples at T0. (B) Bacteria were grown in 96-well plates in protein-free spermidine-free GW medium for 3 h before mNaps were added at various concentrations (arrow). At time points indicated, bacterial viability was determined by plating to solid media and incubation overnight. CFU were counted (dotted lines, limit of detection). Bottom panels represent the same data but grouped according to the growth phase (exponential and autolytic). Error bars, SD, two-way ANOVA (n = 4; p ≥ 0.033, not significant, not represented; p S3 Data.

https://doi.org/10.1371/journal.pbio.3003001.g002

As a control for autolysis, we next constructed a markerless N. gonorrhoeae FA1090 ΔltgA mutant; ltgA is involved in the first step of gonococcal autolysis [4]. Importantly, both the ΔnapRABC and ΔltgA mutants displayed increased survival during the late stationary/autolytic phase of growth compared with wild-type bacteria, whether in presence or absence of exogenously added mNaps (Fig 3A, at t, 36 h, p = 0.002, ** for ΔltgA, and p = 0.03, * for ΔnapRABC in the absence of Naps, and p

Fig 3. The nap island is involved in autolysis.

(A) Long-term survival of wild-type N. gonorrhoeae FA1090 and the ΔnapRABC mutant in the presence (arrow) or absence of mNaps (5 μm). The markerless ltgA deletion mutant was a control for a strain with reduced autolysis. Dotted lines, limit of detection. (B) Autolysis in rich medium. Bacteria were streaked on plates, grown overnight then resuspended in GCB at OD₅₄₀ of ~0.3 and left in cuvettes. After 24 h at room temperature, bacteria were gently resuspended and the OD₅₄₀ measured. One-way ANOVA with Dunnett’s multiple comparison against the wild-type strain (n = 3; p p p S4 Data. (C) Autolysis in HEPES buffer without (left panel) or with (right panel) 10 mM MgCl₂. A representative experiment is shown (last replicate is in the corresponding Suppl. source data files). (D) Transmission electron microscopy images from bacteria in HEPES (T = 75 min). Some wild-type bacteria appear normal (indicated with arrow numbered 1), with many others having a dense (arrow 2) or very dense (arrow 3) cytoplasm, periplasmic expansions (arrow 4), or appearing as ghost cells (arrow 5). Extracellular vesicles (arrow 6) and released cellular content (arrow 7) are also visible. Periplasmic expansions are not observable in the ΔltgA or ΔnapRABC strains. Cytoplasm condensation appears more homogenous in the ΔltgA mutant. The images underlying this figure can be found in https://doi.org/10.6084/m9.figshare.28077749.v1.

https://doi.org/10.1371/journal.pbio.3003001.g003

To establish whether the nap island directly contributes to PCD, we examined wild-type, ΔnapRABC, and ΔltgA N. gonorrhoeae in 2 independent assays of autolysis; N. cinerea was included as a control. First, we assessed changes in the OD₅₄₀ of strains incubated overnight in liquid GCB (Fig 3B); a reduction in OD provides a measure of autolysis [34,35]. While the OD₅₄₀ of wild-type bacteria fell overnight by 34.3% (±4.9, indicating autolysis), the OD₅₄₀ of the ΔltgA and ΔnapRABC mutants only dropped by 10.8% (±8.9) and 6.5% (±10.6), respectively (Fig 3B, p 540 of N. cinerea increased over this time (Fig 3B, +12.4% ± 4), consistent with the lack of PCD in this species. The rate of autolysis of these strains was also estimated by following the OD₅₄₀ of bacteria incubated in HEPES, as starvation in this buffer can induce autolysis [5]. Again, the ΔltgA and ΔnapRABC mutants exhibited decreased rates of autolysis compared with wild-type bacteria (Fig 3C), as shown by the turbidity of cultures at the mid-time point (75 min, p p 7A Fig). Autolysis of 2 strains of N. cinerea was also significantly less than N. gonorrhoeae in this assay (Figs 3C and S7C). Consistent with the ability of divalent cations to impair autolysis [34], addition of MgCl₂ (final concentration, 10 mM) markedly decreased autolysis (Fig 3C) and abolished any significant difference in autolysis between the wild-type, ΔltgA and ΔnapRABC strains (S7B Fig). To understand the role of NapI during autolysis, we attempted to generate a napI mutant by replacing its open reading frame with a kanamycin resistance cassette. Despite multiple attempts, this was unsuccessful in wild-type bacteria presumably because of the toxicity of Naps in the absence of their immunity protein. Consistent with this, we were able to generate a napI mutant in the ΔnapRABC strain. The napI mutant was incubated in HEPES and the OD₅₄₀ measured over time. Surprisingly, the ΔnapRABCI mutant was even more resistant to autolysis than the ΔnapRABC strain (S7A Fig), suggesting that NapI might contribute to autolysis or bacterial fitness in absence of other Naps.

Finally, we examined the ultrastructure of the wild-type, ΔltgA, and ΔnapRABC strains by transmission electron microscopy (TEM) after incubation in HEPES buffer to induce autolysis. Before resuspension in HEPES, the strains were indistinguishable with clear nucleoids visible (S8 Fig). After 75 min, wild-type bacteria showed a mix of phenotypes (Fig 3D), some with dense cytoplasm, while other cells displayed an expanded periplasm, or were empty “ghost” cells with evidence of nearby debris. The cytoplasm of the ΔltgA mutant appeared less dense than wild-type bacteria, with no visible periplasmic expansions (Fig 3D). Similarly, the ΔnapRABC mutant lacked periplasmic expansions (Fig 3D), with many cells having aberrant shapes (Fig 3D). Based on the OD₅₄₀ of bacteria in buffer, both mutants do undergo autolysis. However, TEM images indicate that cell death proceeds differently for the ΔltgA and ΔnapRABC mutants based on differences seen in the expansion of their periplasm. Taken together, our data provide evidence that the Naps contribute directly to autolysis of N. gonorrhoeae.

As the nap island is largely limited to the gonococcus and meningococcus which can elicit marked inflammatory responses [36], we examined whether the mNaps are toxic to host cells. mNaps were tested for their ability to lyse human RBCs over 1 h at 37°C. Interestingly, mNapC had marked haemolytic activity even at low concentrations (Fig 4A), while neither mNapA or mNapB were toxic. To check for interactions between the peptides, we added combinations of mNaps in a 1:1 ratio to cells. Surprisingly, mNapA protected RBC from lysis by mNapC (Fig 4B). To examine whether mNaps are toxic to other host cells, we also measured the release of lactate dehydrogenase (LDH) by THP-1-derived macrophages following exposure to 1 or 5 μm of each mNap (S9 Fig). No cytotoxicity was detected against THP-1 cells with any mNap, suggesting that their effect is cell-type specific.

Fig 4. Effect of Naps on human red blood cells and proposed model.

(A) Different concentrations of mNaps were added to human RBCs for 1 h and the OD₄₁₄ measured. Data were normalised to cytotoxicity with melittin (100%), or PBS (0%). Right panel, cytotoxicity induced by 1.56 μm of each mNap (n = 4; p (B) mNapA or mNapB were mixed with mNapC at a 1:1 ratio and cytotoxicity measured. Right panel, results with 3.15 μm of each peptide, n = 3, p S5 Data. (C) Proposed model of gonococcal autolysis. Peptidoglycan (PG) remodelling enzymes, including LtgA (pink) which localises at the septum, will release PG fragments that are recycled (purple arrows). When bacteria reach stationary phase, PG synthesis stops, but LtgA remains active, leading to the condensation of the cytoplasm (Fig 3D). Meanwhile, Naps are processed and secreted by NapP and NapF (orange), respectively, leading to their accumulation in the extracellular environment. Vacuolation is triggered, potentially through the action of inner membrane destabilising actors such as holins/toxins. Eventually, the outer membrane (OM) breaks when a critical concentration of mNaps (blue) is reached, leading to release of cellular contents and the appearance of ghost cells. For diplococci, cells undergoing PCD will benefit first-degree relatives by direct and indirect mechanisms.

https://doi.org/10.1371/journal.pbio.3003001.g004

Bacterial strains used in this study are listed in S3 Table. N. gonorrhoeae and N. cinerea were grown on GCB agar plates (1.5% wt./vol. proteose peptone number 3, Becton Dickinson, 0.1% starch, 0.4% K₂HPO₄, 0.1% KH₂PO₄, 0.5% NaCl, 1% Vitox, Oxoid, 1.5% agar Oxoid) [54] at 37°C with 5% CO₂. L. crispatus was grown on MRS agar plates (ATCC medium 416) and incubated at 37°C with 5% CO₂ for about 36 h. E. coli was grown on lysogeny broth (LB) agar plates at 37°C.

All primers are shown in S3 Table. For allelic replacement, overlap PCR was performed to obtain resistance cassettes (aph(3)-I or ermC, for kanamycin or erythromycin resistance, respectively) flanked by regions (approximately 1,000 bp) surrounding the target gene. Briefly, overlap PCR consisted in mixing 3 PCR products (“upstream PCR 1” including the START codon of the targeted gene; “downstream PCR 2” including the STOP codon of the targeted gene; “resistance cassette PCR 3”) in equal ratios. For markerless deletions, 2 distinct overlap PCR products were generated: (a) a PCR product consisting of homologous regions (“upstream PCR 1” and “downstream PCR 2”) flanking selection and counterselection cassettes controlled by a constitutive promoter (“p_opaB-kanR-pheS* PCR 3”, S3 Fig); (b) a markerless PCR product consisting of homologous regions directly bound to each other (“upstream PCR 1” and “downstream PCR 2”). N. gonorrhoeae was transformed as previously [54]. For selection erythromycin (0.5 μg ml⁻¹), kanamycin (80 μg ml⁻¹), or 4CP (8 mM) were added to media. Individual transformants were screened by PCR and confirmed by sequencing.

The nap island was identified by MaGe [17] by comparing synteny maps between the reference N. gonorrhoeae FA1090 and N. gonorrhoeae NCCP11945, N. gonorrhoeae FA6140, N. gonorrhoeae 35/02, N. gonorrhoeae PID24-1, N. meningitidis 053442, N. meningitidis FAM18, N. meningitidis MC58, Neisseria lactamica ATCC 23970, N. cinerea ATCC 14685, Neisseria flavescens NRL30031/H210, N. flavescens SK114, Neisseria subflava NJ9703, Neisseria sicca ATCC 29256 and Neisseria mucosa ATCC 25996. Using PubMLST [20], the ORF loci and alleles associated to the nap island were manually curated, as described in https://bigsdb.readthedocs.io/en/latest/curator_guide.html and https://www.youtube.com/watch?v=09g5YdrCtDc. Briefly, using the Neisseria isolates database curator’s interface, the “sequence tags scan” tool allowed us to retrieve alleles in given batches of isolates (see below for isolates selection criteria). New alleles were then validated through sequence alignment and added to the database using Neisseria typing database curator’s interface, “sequences (batch) add” tool. This process was repeated until at least 93.5% of the selected isolates had an allele number attributed to each nap gene. To determine gene conservation (Fig 1C), manual strain selection from the isolate collection of each species on PubMLST was done using the following criteria: a complete rMLST to confirm the species, and the number of contigs S2 Table.

Bacteria were grown in 50 ml of protein- and spermidine-free GW medium [55] in vented flasks (Corning, 431144), by inoculation at OD_600nm 0.025 from bacteria grown on GCB plates (T₀). RNA was purified from 2 ml of culture pelleted for 4 min at 4,500 rpm and resuspended in TRIzol for 5 min then frozen at −80°C. Tubes were then thawed on ice and mixed with 200 μl of chloroform by vigorous shaking of tubes. After 2 to 3 min incubation at room temperature, tubes were centrifuged at 4°C for 15 min at 12,000 x g and the aqueous phase transferred to 500 μl of ice-cold isopropanol. Tubes were incubated overnight at −20°C, then centrifuged at 4°C for 30 min at 20,000 x g, and pellets were washed with 75% ice cold ethanol, before being resuspended in 80 μl diethylpyrocarbonate-treated H₂O. Samples were treated with ezDNase (Invitrogen) and first strand cDNA synthesis was performed with SuperScript IV reverse transcriptase (RT) (Invitrogen) according to manufacturer’s instructions. Note that to obtain specific cDNA, reverse primers (available in S3 Table) were used (alongside the one for the housekeeping gene recA), and a no-RT control was done in parallel. Samples were treated with RNase H for 20 min at 37°C before qPCR was performed using SYBR Green PCR master mix (Applied Biosystems). Primer efficiency was evaluated using serial dilutions to generate a standard curve from PCR products (pre-screen) and mixes of cDNA (to reflect real qPCR conditions). The slopes of standard curves and efficiency values for each primer pair was calculated. StepONEPlus real-time PCR software was used to collect RT-qPCR data and the ΔΔCt method [56] was used to analyse the data, using recA Ct values as reference gene and T₀ Ct values as reference condition for Fig 2A. Experiments were done with at least 3 biological replicates, each time with duplicate or triplicate technical repeats.

Bacteria were freshly harvested from plates and resuspended in GW medium [55] at 10⁸ bacteria/ml then spread on agar plates. L. crispatus was plated on MRS agar plates, while all other bacteria were plated on GW agar plates, prepared by mixing (50:50) 2×-concentrated liquid GW medium (filtered) with warm autoclaved 2% agar in water. Plates were allowed to dry for 10 to 15 min, then 6 mm Whatman paper disks soaked with 10 μl of peptide (50 μm) were added. Polymyxin B (50 μm) and water were used as controls. Plates were incubated for 24 to 48 h at 37°C and 5% CO₂ as required. The efficacy of compounds was assessed by the presence of a zone of growth inhibition around disks.

All bacteria were harvested from plates and resuspended in FB medium [57]. Peptides were prepared in FB medium as 2×-concentrated solutions at 512 μg/ml and 2-fold dilutions (to 10 μg/ml) were prepared in untreated U-bottom polypropylene 96-well plates (Corning, 3879) to 50 μl per well. Penicillin G and polymyxin B were used as controls. Bacteria (50 μl of 10⁵ CFU/ml) were added to each well, then incubated at 37°C with 5% CO₂ with shaking at 180 rpm for 24 h, before spotting 10 μl of each well to plates. After 24 h of incubation, MBC values were attributed to the lowest concentration which gave no growth.

Bacteria were harvested from plates, resuspended in FB medium [57], and diluted to 10⁵ CFU (colony-forming unit)/ml. Aliquots of 250 μl of prey bacteria were mixed with 250 μl of either wild-type N. gonorrhoeae FA1090, or the ΔnapRABC mutant, or FB medium alone. Cultures were incubated at 37°C with 5% CO₂ with shaking at 180 rpm for 3 and 24 h, before plating to selective agar plates; for N. cinerea, E. coli, and L. crispatus, media were GCB plates supplemented with 0.01% Congo Red [58], LB agar plates, and MRS agar plates, respectively. The number of prey bacteria were normalised to control wells without added gonococci.

Human blood (1 ml; K2EDTA; Cambridge Bioscience, United Kingdom) was washed 3 times with 4 ml PBS prior to centrifugation at 700 × g for 8 min as previously [59]; erythrocytes were pelleted at 1,000 × g for 10 min, and diluted to 0.5% v/v. Peptides were added to 96-well V-bottomed polypropylene plates at a starting concentration of 100 μm. Mellitin (2.5 μm) and PBS were added as positive and negative controls. Erythrocytes were incubated at 37°C for 1 h, pelleted at 1,000 × g for 10 min; the OD_{415 nm} of supernatants (60 μl) was read, and results normalised to the controls.