nagZ Triggers Gonococcal Biofilm Disassembly

Bacterial-bacterial interactions play a critical role in promoting biofilm formation. Here we show that NagZ, a protein associated with peptidoglycan recycling, has moonlighting activity that allows it to modulate biofilm accumulation by Neisseria gonorrhoeae. We characterize the biochemical properties of NagZ and demonstrate its ability to function as a dispersing agent for biofilms formed on abiotic surfaces. We extend these observations to cell culture and tissue explant models and show that in nagZ mutants, the biofilms formed in cell culture and on human tissues contain significantly more biomass than those formed by a wild-type strain. Our results demonstrate that an enzyme thought to be restricted to peptidoglycan recycling is able to disperse preformed biofilms.

glycoside phosphorylase (GH3) superfamily. Glycosidases from GH3 are retaining enzymes and cleave their substrates in an acid/base-catalyzed two-step double-displacement mechanism involving a covalent glycosyl-enzyme intermediate in which a fully conserved aspartic acid functions as the catalytic nucleophile 17 . Since ORF0135 is homologous to NagZ in a wide variety of gram-negative bacteria and contains the conserved aspartic acid (the homology to E. coli and Pseudomonas aeruginosa is shown in Supplemental Fig. 1) we propose renaming ORF0135 nagZ. This gene is present in all neisserial genomes currently available in NCBI, with a high degree of conservation (> 97% identity) at the nucleotide level among GC strains ( Figure S1). The gene synteny is identical in the pathogenic strains, but diverges significantly in commensal strains (data not shown).
The DNA encoding NagZ was cloned into the expression vector pET28a and the expressed protein purified. The molecular mass of NagZ as determined by SDS-PAGE (47 kDa) was consistent with the predicted mass for this protein (Supplemental Fig. 2). We determined the optimal enzymatic conditions for NagZ using p-Nitrophenyl N-acetyl-β -D-glucosaminide (pNP-GlcNAc) as a substrate. The optimal conditions for the assay were determined to be 37 °C, in KPO 4 buffer (200 to 400 mM) at a pH of 8.0 (Supplemental Fig. 2B,C). The enzyme retained 100% of activity after 48 hr at room temperature (data not shown). The kinetic parameters K m and V max against (pNP-GlcNAc) were 3.2 mM and 64 μmol min −1 mg −1 respectively, and these values are in the same order of magnitude as those for several other hexosaminidases 18 . We calculated the specific activity of our purified protein as 1600 nmol min −1 mg −1 . We screened for activity of this enzyme using p-Nitrophenyl conjugated to other sugar moieties (Supplemental Fig. 2D). Of the substrates that produced detectable cleavage products, NagZ had about 20% maximal activity on p-nitrophenyl-beta-D-N,N'-diacetylchitobiose versus that measured with PNP-GlcNAc .
Because N. gonorrhoeae lacks the genes needed to encode a carbohydrate capsule 19 , cell surface N-acetyl-glucosamine is only found associated with LOS. However, this would not preclude NagZ possessing the ability to cleave N-acetyl-glucosamine linked through other conformations. This led us to test Staphylococcus aureus, which produces an extracellular polysaccharide that is a β -1-6-linked polymer, predominately composed of N-acetyl-glucosamine 20 that is required for biofilm formation 21 . When this polymer is degraded by treatment with dispersin B, the biofilm is disrupted 22 . We tested to see if NagZ could disrupt a biofilm formed by S. aureus SH1000. NagZ treatment of a S. aureus biofilm had a negligible impact on its biofilm (Supplemental Fig. 2E). Conversely, treatment with Dispersin B significantly reduced the biomass of the S. aureus biofilm. Dispersin B treatment of gonococcal biofilms produced a negligible impact on its biofilm (data not shown). Given that prolonged treatment of a S. aureus biofilm with NagZ failed to have an impact on it, it is likely that NagZ lacks exo-and endo-β -1-6-N-acetyl-glucosaminidase activity.
Because NagZ is needed for the formation of monosaccharides from the released disaccharides during the cytosolic steps of the muropeptide-recycling pathway in E. coli 23 , we hypothesized that the accumulation of peptidoglycan fragments that would occur in a nagZ mutant could affect gonococcal viability. We used a recombination strategy to delete this gene and verified that the deletion had been incorporated properly in the genome by DNA sequence analysis of the region containing the deletion. We measured the growth properties of the deletion mutant and the data indicate that in standard liquid growth media, there is no difference in growth between the parent and deletion strain (Supplemental Fig. 3A).
NagZ activity could be detected in the supernatant as the culture enters stationary phase (Supplemental Fig. 3A). We could not detect any NagZ activity in culture supernatants of Log-phase cells, but could readily detect its activity in the cell lysate. Bioinformatic programs to predict the cellular location of nagZ (Signal p4.0 24 , PSORTb v3.0 25 , PSLPred 26 and MESSA 27 ) suggested a cytosolic or periplasmic location for NagZ (data not shown). Using both bacterial lysates and broth supernatants isolated from overnight cultures from FA1090∆ nagZ, we were unable to detect beta-hexosaminidase activity, suggesting that NagZ is the only beta-hexosaminidase in N. gonorrhoeae FA1090. NagZ acts on peptidoglycan. As gonococci grow they recycle the peptidoglycan fragments removed from the sacculus. However, they release a significant portion of these fragments into the medium, and characterization of these fragments produces a reproducible chromatographic profile 28 . We tested NagZ for its ability to digest metabolically-labeled peptidoglycan fragments. Using a boiled preparation of NagZ as a negative control, we generated the expected profile (Fig. 1). The data in Fig. 1A show that the addition of NagZ altered the profile of peptidoglycan fragments, where the peptidoglycan monomer peak shifted to later fractions (smaller size) from the sizing columns, consistent with the removal of N-acetylglucosamine from the peptidoglycan monomers. Similarly, the free disaccharide peak was eliminated and a peak appeared for free 1,6-anhydro-N-acetylmuramic acid. No significant changes were seen to the peaks for peptidoglycan dimers and tetrasaccharide-peptide, suggesting that these molecules are not substrates of NagZ. These data are consistent with the known activity of NagZ in other bacteria, acting to remove N-acetylglucosamine from peptidoglycan monomers and free disaccharide 29 . NagZ does not act upon LOS. Because NagZ was able to cleave substrates containing N-acetylglucosamine, and neisserial LOS contains N-acetylglucosamine in a variety of conformations, we assayed NagZ for its ability to act upon neisserial LOS. Purified LOS from N. gonorrhoeae F62 was digested with NagZ and its mobility compared with undigested LOS. Because F62 expresses several different isoforms of LOS containing terminal and internal N-acetylglucosamine, and NagZ was unable to act on any of these components, whether at a terminal or internal location (Fig. 1B), we concluded that NagZ lacks β 1-3 exoglycosidase activity and α 1-2 endoglycosidase activity. We also performed these assays by adding NagZ to intact cells, and analyzing the LOS after treatment. The data obtained was the same as that obtained using purified LOS, indicating that the enzymatic activity is not linked to the presence of a specific LOS conformation (Data not shown). The nagZ deletion strain forms a thicker biofilm. Gonococcal strains growing in biofilms are under significant metabolic stress 30 . As such, we would expect that the extracellular milieu of a biofilm would contain cytosolic enzymes because stressed cells are dying and dying cells are lysing. We hypothesized that a nagZ mutation could alter the structure of a gonococcal biofilm. The wild type and ΔnagZ strains of N. gonorrhoeae FA1090 were compared for their ability to form a biofilm under static and dynamic conditions. Dynamic biofilms were developed at the liquid air interface on the inner surface of culture tubes. After 24 hr, the bacterial culture was aspirated and the biofilm dried. The data (Fig. 2, panel A) is a visual representation of the biofilm that was formed on a culture tube. The data in Fig. 2, panel B is a quantification of the biomass of this biofilm, using a crystal violet staining procedure 31 . Statistical analysis using Student's t-test showed a significant increase in biomass of the knockout strain over the wild type strain (n = 6, p < 0.001) of approximately 4 fold. A similar phenotype was observed when biofilms were made under static conditions (Fig. 2C). This biofilm enhancement was lost when the mutant was complemented (Fig. 2C). We performed confocal microscopy on static biofilms grown for 36 hr, using Hoechst stain to visualize the biofilm. The mutant strain shows a denser and taller biofilm (~4 fold), consistent with quantification of biomass using crystal violet (Fig. 2D). We compared the ability of FA1090 and FA1090∆ nagZ to form biofilms over time. The biomass associated with FA1090∆ nagZ continued to accumulate over 3 days, while the biomass of a biofilm produced by FA1090 peaked around 48hr (Fig. 2E).
Effect of NagZ on gonococcal biofilm formation. We analyzed biofilm evolution over time using scanning electron microscopy. After 12 hrs, the biofilms produced by FA1090 and FA1090ΔnagZ looked different. The biofilm produced by FA1090 appeared patchy with numerous nucleation sites while the biofilm produced by FA1090ΔnagZ appeared diffuse across the entire field of view with no obvious nucleation sites (Fig. 3A). These biofilms are dramatically different in their appearance at 12 hrs. FA1090ΔnagZ biofilms showed an increase in cell density over time, while the FA1090 biofilm density seemed to peak at 48 hrs. This difference cannot be explained by differences in the growth rate as both strains give the same growth profile (see Supplemental Fig. 3A) or due to the presence of other extracellular components as the overall biomass of the mutant biofilm is greater.
We hypothesized that the differences seen in the biofilms made by FA1090 and FA1090∆ nagZ were due to the localized presence of NagZ in the extracellular milieu. For this to happen, we hypothesized that NagZ could gain access to the extracellular milieu through autolysis. We analyzed the impact of exogenously added NagZ on biofilms produced by FA1090∆ nagZ. Static biofilms were formed and after treatment with NagZ protein, the biofilms were visualized using SEM. NagZ treatment reduced the thickness and density of the biofilm formed by FA1090∆ nagZ (Fig. 3B). The reduction in biomass was quantified, and the data indicate more than 50% of the biofilm was removed with the addition of exogenously added NagZ (Fig. 3C). Addition of NagZ to the bacterial suspension at the beginning of biofilm formation also resulted in a similar reduction in biofilm formation after 24 hr (data not shown). After NagZ treatment, strands of nucleic acid were still apparent in the biofilm (Fig. 3B); treatment with DNase not only removed the extracellular DNA but also degraded the biofilm (data not shown). The data in Fig. 3D show the presence of extracellular DNA in biofilms produced by both strains. The lack of nagZ does not alter blebbing. N. gonorrhoeae generates outer membrane vesicles 32 and alterations in vesicle formation have been observed in bacteria with defects in peptidoglycan metabolism 33 . Since NagZ is involved in peptidoglycan recycling, it is possible that NagZ could affect biofilm formation indirectly by increasing the release of outer membrane vesicles. Using high resolution SEM to visualize the biofilm architecture and the ultrastructure of bacteria within the biofilm, we determined the number and size of Blebs made by FA1090 and FA1090∆ nagZ. We found that there was no difference in bleb size or distribution between the two strains, ruling out a membrane blebbing defect as the cause for difference in biofilm formation (See supplemental Fig. 4). Viability of cells within a biofilm. SEM images showed what appears to be DNA entangling the bacterial aggregates in a random fashion. Since FA1090 lacks the gonococcal genetic island implicated in DNA secretion 34 , the source of DNA could be through bacterial lysis or leakage of DNA from dead cells. The data in Fig. 4A show the double-stained (Hoechst and propidium iodide (PI)) biofilms visualized by confocal microscopy (Hoechst stains both dead and live bacteria whereas PI only permeates the membranes of dead bacteria). Figure 4B shows the quantification of these images by comparing the mean fluorescence intensity ratio (FIR) of Hoechst staining to PI staining. No significant difference in the proportion of live to dead bacteria was observed, although the amount of dead bacteria increased over time as seen by an increase of fluorescence intensity of PI. By 72 hrs, both biofilms contain a significant portion of dead cells.

Effect of NagZ expression on Biofilm formation on human cells.
We determined the ability of FA1090 and FA1090∆ nagZ to form a biofilm on polarized T84 cells. Biofilms formed on the apical surface were visualized by SEM (Fig. 5A). The surface area covered by the biofilms was quantified using ImageJ and the percent of the area covered by biofilm determined. The data (Fig. 5B) show that the mutant strain covered significantly more surface area than the wild type strain, consistent with experiments performed on glass coverslips. Although we could not quantify the total biomass of these two biofilms, it appeared that the FA1090∆ nagZ biofilm was taller than the FA1090 biofilm.
To establish the clinical relevance of NagZ in biofilm formation, we obtained cervical tissues from women undergoing uterine surgeries and sourced by the National Disease Research Interchange (NDRI). Tissue samples were infected with ~10 9 bacteria per tissue sample, and after 48 hr incubation, the presence of GC on the sample visualized by confocal microscopy. The Map tiling function was used to take images of small sections of the tissue is an SEM of FA1090∆ nagZ biofilms, treated and untreated with exogenously added NagZ (Magnification 5000 X ). Arrows in the panel indicate that NagZ treatment did not remove DNA strands. Panel (C) Quantification of biofilm formed by untreated mutant strain, and mutant strain biofilm treated with purified NagZ, using crystal violet method. There is a statistically significant reduction in biofilm on treatment with NagZ (p < 0.001). Two-tailed t-test assuming unequal variance was employed to determine statistical significance. Panel (D) shows the presence of extracellular DNA in both strains' biofilms (magnification 20000X). The top panel is FA1090 and the lower panel is FA1090∆ nagZ. as a Z-stack, and all the images were tiled using automated software to create a Z-stack of the entire cervix specimen. The stacks were Z-projected as shown in Fig. 6A. The relative number of cervical cells in a given sample area was defined based on the fluorescence intensity of the Hoescht stain (blue), which stains nuclei. The overall shape of the explant was shown by actin staining (green). The relative number of GC of in a sampled area was determined based on the red fluorescence, which was the result of immunostaining GC. The biofilm quantity was defined as the ratio of red fluorescence intensity to blue fluorescence intensity. This was then normalized to 100 (Fig. 6C). This data indicates that the nagZ strain produced a biofilm with four times the number of GC compared to wildtype strain, which is in agreement with experiments performed on abiotic surfaces.

Discussion
While N. gonorrhoeae can form biofilms in vivo and in vitro 2 , the mechanism by which they are formed and the processes that mediate their dispersal remain unclear. We show that when GC lack the ability to make NagZ, they produce more robust biofilms that continue to accumulate over time (Fig. 2) and that exogenously added NagZ reduces these biofilms (Fig. 3). Because NagZ plays a role in peptidoglycan (PG) turnover in other organisms 23 , and is able to act in an analogous way on gonococcal peptidoglycan fragments, NagZ can be considered another example of a moonlighting enzyme 35 .
Electron microscopic studies of cervical biopsy specimens from patients with culture-proven N. gonorrhoeae infection have revealed evidence of biofilms; these biofilms appeared to be only a few bacteria thick 36 . It is unclear why these biofilms do not take on the qualities of those seen on abiotic surfaces or on tissue culture cells. Because  The graph on the right is a quantification of the biofilms. The percent of the surface covered by biofilm was measured using ImageJ analysis. Multiple images (n = 9) were evaluated using SEM and the significance of the differences determined with Student's t-test. (***p < 0.001).
FA1090∆ nagZ biofilms are significantly thicker on tissue culture cells and cervical explants than FA1090, we suggest that the release of NagZ during autolysis is limiting the accumulation of biomass.
It has been hypothesized that GC biofilms produced during colonization of the cevix results in asymptomatic carriage 36 . How N. gonorrhoeae suppresses development of an adaptive immune response during natural infection in the presence immunostimulatory peptidoglycan fragments is unclear 37 . We suggest that removal of N-acetylglucosamine from toxic peptidoglycan monomers 38 renders them immunosilent, and that the released N-acetylglucosamine suppresses neutrophil functions 39 and proinflammatory cytokine induction 40 . Hence, the liberation of N-acetylglucosamine from peptidoglycan fragments by NagZ can serve to suppress the immunological signaling when the gonococcus colonizes cervical mucosa.

Experimental Procedures
Bacterial strains and plasmids. N. gonorrhoeae were grown in standard gonococcal medium (Difco, MI), designated GCP if used as phosphate-buffered broth, and designated GCK if used with agar, supplemented with 1% Kellogg's supplement 41 , at 37 °C and 5% CO 2 . Strain F62∆ 8-1 has been previously described and is a strain that is genetically deleted for lgtA 42 . All GC strains were phenotypically non-piliated and Opa negative. E. coli BL21(DE3) and E. coli DH5α MCR (Life Technologies, MD) were grown in L broth (LB) at 37 °C or 30 °C 43 . Antibiotics included in media were used at the following final concentrations (μg ml −1 ): ampicillin 100, kanamycin 50. Plasmid pUC19 (New England Biolabs) and pET28a(+ ) (Novagen) were used to make deletions in various N. gonorrhoeae strains and to express NagZ protein, respectively. S. aureus SH1000 strain was obtained from Dr. Jeffery Kaplan, George Washington University.

Chemicals and Reagents.
All reagents were purchased from Sigma-Aldrich, unless stated otherwise.
Polymerase chain reactions were performed using Pfu polymerase (Thermo Scientific Molecular Biology, Pittsburgh, PA) or GoTaq polymerase (Promega, Madison, WI), and carried out according to the manufacturer's instructions. All restriction enzymes were sourced from New England Biolabs (Beverly, MA). A sample of Dispersin B was obtained from Dr. Jeffery Kaplan, George Washington University. Table 1. A DNA fragment carrying the nagZ gene of N. gonorrhoeae FA1090 along with ~850 bp of right and left flanking DNA sequences was amplified using primers hexoF and hexoR. The resulting amplicon (~2700 bp) was cloned into the pUC19 EcoRI and HindIII sites and transformed into E. coli DH5α MCR, resulting in the formation of pUC19::nagZ. One of the isolated clones was used for further experiments. To disrupt nagZ in the gonococcus, a Kanamycin (Kan) resistance gene was excised from pK18 44 with SmaI, the DNA fragment purified from agarose gel and inserted into the EcoRV site of pUC19::nagZ. A plasmid from a presumptive E. coli clone carrying pUC19::nagZ::Kan was isolated and the construct was verified by DNA sequencing (Macrogen, USA). This plasmid DNA was cleaved with EcoRV and ligated with double-strand oligo DNA prepared from DUSF and DUSR, which contain two EcoRV sites and N. gonorrhoeae DNA uptake sequence, after cleaving with EcoRV. Kanamycin resistant transformants were selected, the plasmid DNA isolated and checked for the presence of the uptake sequence. Plasmid DNA from such clones was used for transformation of N. gonorrhoeae strains by liquid transformation method 45 and selected on GCK plates containing kanamycin (50 ug/ml). nagZ was deleted by a spot transformation technique 46 in N. gonorrhoeae FA1090. The nagZ complemented strain was constructed using a PCR amplicon generated with primers hexoF and hexoR to transform the deleted strain, using a spot transformation technique 43 . Expression and purification of NagZ protein. DNA fragment encoding nagZ isolated from N. gonorrhoeae FA1090 was amplified using primers Ngo0135F and Ngo0135R and the resulting amplicon (1300 bp) was cloned into pET28a(+ ) using NheI and XhoI, resulting in the formation of plasmids pET28a(+ )::nagZ. To purify NagZ, a single colony generated by transformation of E. coli BL21(DE3) with pET28a(+ )::nagZ was used to inoculate 100 ml of LB broth with kanamycin. Cultures were incubated at 37 o C, and when the OD 600 of the culture reached 0.6, IPTG was added to a final concentration of 1 mM. Incubation was continued at 30 °C for an additional 12 hr. Cultures were centrifuged and bacterial pellets were resuspended in 10 ml buffer A containing 50 mM NaH 2 PO 4 (pH 8.0), 300 mM NaCl, 20 mM imidazole, 10 mM -mercaptoethanol, 0.1% (v/v) Tween 20, 100 μM PMSF. After sonication, the cellular debris was removed by centrifugation at 40 000 g for 1 hr and then the supernatant applied to a 3 ml Ni-NTA agarose column previously equilibrated with 100 ml of above buffer. LOS purification and analysis. LOS was purified from broth-grown cells by hot phenol-water method 47 . LOS was then concentrated by lyophilization and extraction with hot phenol-water continued until the preparation obtained minimal absorbance at 200 nm. LOS was analyzed on a 16.5% Tris-Tricine gel at a fixed current of 0.03 mA in an ice-cooled chamber, and visualized by silver staining 48 . Hexosaminidase assay. Enzyme activity was assayed using various p-nitrooligosaccharides as substrate. Biofilm formation. To prepare biofilms, bacteria from overnight cultures grown on GCK agar plates were suspended in GCP with sodium bicarbonate and Kellogg's supplement to a final concentration of 10 8 bacteria per mL. Dynamic biofilms were grown in new heat-sterilized glass tubes by adding 1 mL of the bacterial suspension per tube, and incubating in a rotary shaker for 48 hr. Static biofilms were obtained by making a suspension of bacteria to a Klett of 100 in GCP and adding an aliquot to culture flasks using published methods 49 . The biofilms were quantified by staining with 1% w/v crystal violet. The crystal violet stained biofilm was eluted out by dissolving in 30% acetic acid and the absorbance read at 590 nm in a spectrophotometer.

Construction of mutants. A list of all primers used is provided in
The same protocol was used to study biofilm formation on polarized epithelial cells. T84 epithelial cells were seeded at 10 5 per transwell insert in 24 well plates and grown for 10 days. The cells were considered polarized once the transepithelial electrical resistance (TEER) levels reach above 1500 Ω.cm 2 . Bacteria was added on the apical surface and biofilm formed for different time points. For scanning electron microscopy, the transwell membrane was fixed in glutaraldehyde and the membrane cut and removed from the insert before staining. Static  for 60 min at RT and then overnight at 4 o C. The coverslips were fixed the next day using 1% osmium tetroxide, dehydrated by a series of washes with increasing concentrations of ethanol, dried by critical point drying method, and finally coated with gold-palladium alloy. Samples were visualized with Amray 1820D microscope (20 kV) and Hitachi S4700 microscope (5 kV).

Cervical infections.
Women undergoing uterine surgeries provided informed consent for the use of cervical tissue samples, which were obtained from NDRI (National Disease Research Interchange). The University of Maryland Institutional Review Board approved of the experiments and methods, which were carried out in accordance with the approved guidelines. This endocervical tissue was received within 24 h post-surgery. Samples were cut into ~2.5 cm (L) × 0.6 cm (W) × 0.3 cm (H) pieces, incubated in CMRL-1066 (GIBCO) plus antibiotics for 24 h, and switched to antibiotic-free media for 24 h. Figure 6 the tissue was infected with approximately 10 9 bacteria per tissue sample. After a 48 hr incubation period, the tissues he tissue was fixed with paraformaldehyde, cryopreserved in gelatin, cryosectioned, and immunostained. Antibody used to label GC was as previously described 50 . Images were acquired by confocal microscopy using a map tiling function was used to take images of small sections of the tissue as a Z-stack, and all the images were tiled using automated software to create a Z-stack of the entire sample.