AmiC Functions as an N-Acetylmuramyl-l-Alanine Amidase Necessary for Cell Separation and Can Promote Autolysis in Neisseria gonorrhoeae

Abstract

Neisseria gonorrhoeae is prone to undergo autolysis under many conditions not conducive to growth. The role of autolysis during gonococcal infection is not known, but possible advantages for the bacterial population include provision of nutrients to a starving population, modulation of the host immune response by released cell components, and donation of DNA for natural transformation. Biochemical studies indicated that an N-acetylmuramyl-l-alanine amidase is responsible for cell wall breakdown during autolysis. In order to better understand autolysis and in hopes of creating a nonautolytic mutant, we mutated amiC, the gene for a putative peptidoglycan-degrading amidase in N. gonorrhoeae. Characterization of peptidoglycan fragments released during growth showed that an amiC mutant did not produce free disaccharide, consistent with a role for AmiC as an N-acetylmuramyl-l-alanine amidase. Compared to the wild-type parent, the mutant exhibited altered growth characteristics, including slowed exponential-phase growth, increased turbidity in stationary phase, and increased colony opacity. Thin-section electron micrographs showed that mutant cells did not fully separate but grew as clumps. Complementation of the amiC deletion mutant with wild-type amiC restored wild-type growth characteristics and transparent colony morphology. Overexpression of amiC resulted in increased cell lysis, supporting AmiC's purported function as a gonococcal autolysin. However, amiC mutants still underwent autolysis in stationary phase, indicating that other gonococcal enzymes are also involved in this process.

Neisseria gonorrhoeae is a gram-negative bacterium and the etiological agent of the sexually transmitted disease gonorrhea. An inflammatory response induced during infection is the hallmark of gonorrhea, which is manifested as urethritis in men or cervicitis in women. Gonococcal infection in women commonly causes pelvic inflammatory disease (PID), which may result in ectopic pregnancy, tubal-factor infertility, or chronic pelvic pain. How gonococci promote inflammation in the host is central to understanding the development of symptomatic gonococcal infections and the tissue damage of PID. Constituents of the neisserial cell, including porin, lipooligosaccharide, and peptidoglycan (PG), released during infection will induce inflammation or tissue damage in the host (13, 32, 33, 36). PG fragments are released both when gonococcal cells lyse and during growth (44). To allow for expansion of the cell wall, lytic transglycosylases degrade strands of PG, converting them to 1,6-anhydro PG monomers that are then released (7, 22, 24). In the fallopian tube organ culture model of PID, PG monomers alone or in synergy with lipooligosaccharide caused death and sloughing of ciliated cells (35, 36). Large PG fragments of N. gonorrhoeae have been shown to induce systemic arthritis when injected into rats (12), mimicking the arthritis seen in patients with disseminated gonococcal infection. Additionally, host lysozyme will degrade approximately 50% of large gonococcal PG fragments to PG monomers (9, 46) with toxicity similar to that of the monomers released during N. gonorrhoeae growth (36).

The release of large PG fragments and other proinflammatory cell constituents occurs during autolysis and has been attributed to the action of peptidoglycanases. Multiple peptidoglycan-degrading activities have been described in N. gonorrhoeae, including that of lytic transglycosylases (7, 10, 44), endopeptidases (6), and an N-acetylglucosaminidase (14), but the major autolytic activity has been ascribed to an N-acetylmuramyl-l-alanine amidase (19). Hebeler and Young characterized lysis of N. gonorrhoeae transferred to buffer and found that an amidase was responsible for lysis under these conditions (17-19). Amidase-specific degradation of PG was based on the observation that cell extracts added to purified PG resulted in the production of N-terminal l-alanine from peptide side chains without a concomitant increase in C-terminal amino acid (18, 19). N-Acetylmuramyl-l-alanine amidases specifically cleave the amide bond between N-acetylmuramic acid and l-alanine in PG and can act as potent autolysins when overexpressed or upon addition of antibiotics (20, 21). In Escherichia coli, amidases were shown to act in cleavage of the septum shared by daughter cells to allow for cell separation (20). This class of amidases has been shown to act in both cell separation and autolysis in other bacteria, including Bacillus subtilis and Streptococcus pneumoniae (3, 48).

Here we show that a deletion mutation in a gene encoding a putative N-acetylmuramyl-l-alanine amidase in N. gonorrhoeae affects growth, peptidoglycan fragment release, and cell separation. Our finding that gonococci become hyperautolytic upon overexpression of amiC supports the purported function of AmiC as an autolysin. However, amiC mutants still undergo cell death and autolysis in stationary phase, indicating that other factors are also involved in these processes.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

All N. gonorrhoeae strains used in this study are derivatives of strain MS11 and are listed in Table 1. Piliated gonococcal variants were used for transformation, whereas nonpiliated variants were used for all assays where cell growth and viability were assessed by optical density readings. Gonococcal base liquid (GCBL) medium used for growth of N. gonorrhoeae consisted of 1.5% proteose peptone no. 3, 0.4% K₂HPO₄, 0.1% KH₂PO₄, 0.1% NaCl, pH 7.2, to which was added Kellogg's supplements and 0.042% NaHCO₃ (28, 37). N. gonorrhoeae was also grown on GCB agar plates (Difco) containing Kellogg's supplements in the presence of 5% CO₂ at 37°C. Luria broth or Luria agar plates were used for growth of E. coli (41). Antibiotics were used at the following concentrations for N. gonorrhoeae: streptomycin at 100 μg per ml, erythromycin at 10 μg per ml, and chloramphenicol at 10 μg per ml. For E. coli erythromycin at 500 μg per ml and chloramphenicol at 25 μg per ml were used.

TABLE 1.

Strains and plasmids used in this study

Plasmid or strain	Description	Reference or source
Plasmids
pIDN3	N. gonorrhoeae insertion-duplication plasmid (Ermr)	Hamilton et al. (16)
pKH37	Gonococcal complementation vector (Camr) derived from pGCC6	K. T. Hackett
pDG002	amiC insertion-duplication plasmid (Ermr); 463-bp SalI-ApoI fragment of amiC from pDG015 in EcoRI-SalI pIDN3	This work
pDG005	amiC in-frame deletion constructed in pDG015, contains 1,233-bp deletion internal to amiC (Ermr)	This work
pDG015	2,358-bp amiC PCR product (primers contain EcoRI and KpnI sites) in EcoRI-KpnI pIDN3 (Ermr)	This work
pDG016	amiC complementation; 1,864-bp fragment containing amiC with intrinsic promoter in pKH37 (Camr) with lac promoter	This work
pDG017	amiC complementation; 1,341-bp amiC in pKH37 (Camr); amiC under lac promoter control	This work
Strains
MS11	Wild-type N. gonorrhoeae (Strr)	Swanson et al. (47)
DG002	MS11 transformed with pDG002; amiC insertion-duplication mutant (Ermr)	This work
DG005	DG002 transformed with pDG005; amiC deletion mutant	This work
DG121	DG005 transformed with pDG017; amiC complemented (Camr)	This work
DG122	DG005 transformed with pDG016; amiC complemented (Camr); amiC gene containing intrinsic and lac promoters	This work

Assessment of N. gonorrhoeae growth and death.

Total protein concentrations were determined by the Bradford method using the Bio-Rad protein assay reagents as outlined in the manufacturer's instructions (5). For growth determinations, gonococci were centrifuged for 2 min at 9,300 × g, and the protein content of the cell pellet was determined. Viability of gonococcal strains was measured by using the Live/Dead BacLight bacterial viability kit (Molecular Probes) according to the method of Hamilton et al. (15). Cultures were diluted to an optical density at 540 nm (OD₅₄₀) of 0.2, and 1-ml aliquots were taken at 3.5 h and 22 h and washed in prewarmed 0.1 M morpholinepropanesulfonic acid, 1 mM MgCl₂ (pH 7.2). Triplicate 100-μl aliquots were added to a 96-well opaque black polystyrene plate. An equal volume of 2× Dye solution was added to each aliquot. Samples were excited at 485 nm, and fluorescence at 535 nm and 635 nm was recorded using an HTS 7000 Bio Assay reader (Perkin-Elmer). Percent viability was determined by comparing sample measurements to known standards of live/dead ratios of wild-type gonococci.

For assessment of differences in liquid culture turbidity between strains, gonococcal cultures were grown in supplemented GCBL medium with shaking at 37°C to an OD₅₄₀ of approximately 1.4, representative of late log phase. Cultures were then incubated at room temperature (25°C) for 3 h. Turbidity was determined optically and also measured by OD₅₄₀ spectrophotometer readings.

Plasmid construction.

Plasmids used in this study are listed in Table 1. Primers, which included restriction enzyme recognition sites (underlined), were designed based on the N. gonorrhoeae strain FA1090 genome sequence (GenBank accession no. AE004969). For amplification of amiC and the surrounding region, primers 5′-CGGAATTCGTCTGAAAGCAGGATGCGTAT-3′and 5′-ATGGTACCGGTGTAGTCTGTCCCGCTTAT-3′were used at an annealing temperature of 56°C with MS11 chromosomal DNA as the template. The amplicon contained the region from 527 bp upstream to 534 bp downstream of the amiC coding sequence. The PCR product was digested with EcoRI and KpnI, ligated into the plasmid pIDN3 (16) to produce pDG015, and transformed into competent E. coli TAM1 cells. Plasmids from the resulting erythromycin-resistant (Erm^r) transformants were screened for the predicted size and restriction digest patterns.

To make a deletion in amiC, the cloned amiC construct (pDG015) was amplified with a two-step PCR (initial and final annealing temperatures of 59.6°C and 63.2°C) with divergent primers containing BspHI restriction sites (underlined): 5′-ATATCATGAGGCCAGTTTGGTCATAGCGGCAG-3′and 5′-GCCTCATGATTGAAAAGGCGGTGTTTC-3′. The resulting PCR product was digested with BspHI, self-ligated, and transformed into E. coli. Plasmids from the resulting Erm^r transformants were screened for the predicted size and for the expected restriction digest patterns. One plasmid (pDG005) was confirmed to contain the expected deletion by DNA sequencing.

To construct a plasmid for insertion-duplication mutagenesis of amiC, a portion of the amiC gene sequence was excised from pDG015 using SalI and ApoI and ligated into pIDN3 linearized with EcoRI and SalI to produce pDG002. Erm^r transformants were screened for the presence of a plasmid of the predicted size and for the expected restriction digest patterns.

For complementation, the amiC gene containing its predicted intrinsic promoter was excised from pDG015 and ligated into pKH37 at the ClaI and PmeI restriction sites to generate pDG016. Transformants containing pDG016 were selected for chloramphenicol resistance. A complementation plasmid lacking the endogenous amiC promoter was also constructed in pKH37. Primers specific to the amiC coding region were designed: 5′-TTCTACTAGTTGCTGACCGCCCATACCGAA-3′(SpeI) and 5′-ACATCAAGCTTTGCCTGCCTTCATCCGACAACT-3′(HindIII). Amplification was performed by a two-step PCR (at initial and final annealing temperatures of 58°C and 64°C), and the resulting 1,362-bp product was ligated into pKH37 cut with SpeI and HindIII, forming pDG017. Correct plasmid construction was verified by PCR amplification and restriction digest mapping.

amiC mutants of N. gonorrhoeae.

N. gonorrhoeae strain MS11 was transformed with 2 μg pDG002. Transformants were selected for resistance to erythromycin and analyzed by PCR to verify insertion at the region of homology and duplication of the homologous fragment. The resulting insertion-duplication mutant (DG002) was transformed with NheI-linearized pDG005, and colonies were screened for the loss of resistance to erythromycin. Erm^s colonies were then screened for the absence of amiC by PCR using both intragenic and amiC-flanking primer pairs to identify an amiC deletion mutant (DG005). Complementation of gonococcal mutants was generated by transformation with pDG016 (containing amiC preceded by both endogenous and lac promoters) or pDG017 (containing amiC under the control of the lac promoter) to produce DG122 and DG121, respectively. Plasmids pDG016 and pDG017 contain portions of the gonococcal genes aspC and lctP and facilitate insertion of the complementation construct between these genes in the chromosome. Complemented amiC mutant strains DG122 and DG121 were screened by PCR for incorporation of the additional amiC allele and retention of the amiC deletion mutation.

Peptidoglycan turnover and released fragment characterization.

Gonococcal PG was characterized essentially as described by Cloud and Dillard (7). Gonococcal cultures were grown in GCBL medium containing 0.4% pyruvate and lacking glucose and containing 10 μCi/ml 6-[³H]glucosamine to label the PG. Cells were grown for 35 min, centrifuged for 5 min at 1,800 × g, washed, resuspended in an equal volume of GCBL without label, and grown for 2.5 h. The cultures were centrifuged for 5 min at 1,800 × g, and each supernatant was filter sterilized and stored at −20°C. Six ml of supernatant containing released labeled PG fragments was passed through a size exclusion column and eluted with 0.1 M LiCl. Three-ml fractions were collected, and 0.5 ml of each fraction was added to 3 ml of LS cocktail (Research Products International) and counted with a Packard Tri-Carb 2100TR liquid scintillation counter.

Lysis in buffer.

For measurement of autolysis in buffer, gonococci were grown in GCBL medium to log phase (OD₅₄₀, 0.6 to 1.4) from an initial OD₅₄₀ of 0.2. Cells were centrifuged for 5 min at 1,800 × g, washed, and resuspended in 1 ml of 50 mM Tris HCl (pH 6). After washing, ∼0.3 ml of cells was transferred to 6 ml of 50 mM Tris-HCl (pH 8) and monitored for a decrease in turbidity from an initial OD₅₄₀ of 0.3 at room temperature. Isopropyl-β-d-thiogalactopyranoside (IPTG) at 1 mM was added to gonococcal cultures during growth, 2.5 h prior to suspension in buffer.

RNA release assay.

To label RNA metabolically, nonpiliated gonococci were grown to log phase, centrifuged for 5 min at 1,800 × g, washed in GCBL medium, and resuspended at an OD₅₄₀ of 0.2 in defined medium (16) containing 2 μCi/ml [³H]adenine. After 35 min, cells were centrifuged for 5 min at 1,800 × g, washed, and resuspended in GCBL medium. At each time point during growth, 0.5-ml aliquots were removed and bacteria were harvested by centrifugation for 2 min at 9,300 × g. ³H contents in both the supernatant and cell pellet were determined by scintillation counting, and the amount of RNA released was calculated as a percentage of the amount present in the cell pellet at time zero.

Microscopy.

Thin-section electron micrographs of gonococcal strains were obtained as described by Mehr et al. (34). Confocal microscopy was used to examine the viability of gonococci using the BacLight Live/Dead staining kit (Molecular Probes) as described by Hamilton et al. (15).

Outer membrane preparations.

To determine expression of opacity proteins, outer membrane preparations incubated at 37°C and 100°C were compared by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Opacity proteins are heat modifiable, which results in a shift in migration of the Opa protein being expressed (49). Opaque and transparent colony variants were grown overnight on two agar plates per strain. Gonococci were swabbed from plates into 5 ml of 200 mM lithium acetate buffer (pH 6.0) plus 10 mM EDTA. Outer membrane blebs were prepared from gonococcal cells by passing the cell suspension 10 times through a 22-gauge needle. After removal of the cells by centrifugation at 12,000 × g for 10 min, outer membrane material in supernatants was harvested by centrifugation at 100,000 × g for 2 h and then resuspended in 50 μl of water. Samples were prepared for SDS-PAGE by adding SDS loading buffer to two samples from each variant, where one sample was subjected to 37°C for 1 h and the other sample to 100°C for 5 min. Samples from each variant were electrophoresed in a 10 to 20% polyacrylamide gel and stained with Coomassie brilliant blue.

Statistical analysis.

To assess statistical significance, we utilized the linear mixed-effects model for repeated measures (38). These statistical tests are more accurate for the analysis of repeated measures than standard t tests and linear regression models, which do not take into account the special correlation structure (i.e., measurements at each successive time point are not independent of one another). We included a fixed term in the model corresponding to a particular strain or variant, and the goal was to test for the significance of this term. Also included was a term to account for differences between individual experiments. For RNA release assay measurements (see Fig. 7, below), the relationship between time and response was found to be linear, while for the lysis in buffer assay (see Fig. 2, below), it was found to be quadratic. Random effects were kept in the model for the time coefficients as well as the intercept term. Likelihood ratio tests were used to determine which random effects should be included in the model. Significance of fixed parameters was determined with conditional F and t tests (38). Statistical significance for measurements taken at a single time point (see Fig. 4, below) was determined using Student's t test.

FIG. 7.

Cell lysis during growth. To determine the effects of amiC mutation on autolysis during growth, gonococcal RNA was pulse-labeled by growth in defined medium containing [³H]adenine. RNA released into the culture medium during the chase period was detected by scintillation counting. Initial counts per minute for the cell pellet were determined, and the amount of radioactivity in the culture supernatant at various times was divided by the initial counts per minute to determine the percentage of cells that had lysed. Values shown are the averages of at least three separate trials. Error bars represent the standard deviations. Linear fixed-effects statistical model P values for MS11 versus DG005 and DG122 plus IPTG strain comparisons are all <0.0001. The P value for MS11 versus DG122 is 0.6. P values for DG005 versus DG122 or DG122 plus IPTG and for DG122 versus DG122 plus IPTG are all <0.0006.

FIG. 2.

Autolysis under nongrowth conditions. Gonococcal cells grown in rich medium were washed and then suspended in Tris-HCl buffer, pH 8.0, to trigger autolysis. Turbidity of each sample was determined spectrophotometrically over time. Values shown are the averages from at least three separate trials. Error bars represent the standard deviations. P values for MS11 versus DG005, DG122, and DG122 plus IPTG are <0.0001, 0.8, and 0.003, respectively. P values for DG005 versus DG122 or DG122 plus IPTG and for DG122 versus DG122 plus IPTG are all <0.0001.

FIG. 4.

amiC cultures exhibit prolonged turbidity. Cultures were grown with aeration overnight and then left to sit at 25°C for 3 h. (A) The photograph of gonococcal cultures shows loss of turbidity and settling of cells (a white pellet formed at the bottom of the tube) apparent with the wild-type MS11 and complemented (DG122 and DG122 plus IPTG) cultures. (B) Spectrophotometric readings (OD₅₄₀) were taken at the time of photography to quantify visual observations. (C) OD readings for gonococci resuspended in culture supernatant (sup) from wild-type or amiC cultures. OD values represent averages of four separate trials. Error bars depict the standard deviations. ‡, Student's t test P value of <0.05 compared to all other measurements.

RESULTS

Identification of an N-acetylmuramyl-l-alanine amidase homologue in N. gonorrhoeae.

Biochemical investigations of autolysis performed by Hebeler and Young indicated that an N-acetylmuramyl-l-alanine amidase was predominately responsible for the cell lysis events observed in N. gonorrhoeae (17-19). To identify the putative autolysin, we searched the genome sequence of N. gonorrhoeae strain FA1090 (GenBank accession no. AE004969) and found two genes that encode putative N-acetylmuramyl-l-alanine amidases. One (NGO0237) is a homologue of the E. coli amidase AmpD that hydrolyzes muramyl peptide fragments in the cytoplasm for reuse in cell wall synthesis (25). The second (NGO1502) is a homologue of E. coli amidase AmiC, a periplasmic enzyme that degrades intact PG and contributes to cell separation (20). We designated the latter enzyme amidase C (AmiC). Gonococcal AmiC exhibits 48% identity and 66% similarity to E. coli AmiC. The predicted protein is 47 kDa and carries a characteristic twin-arginine secretion signal (51), suggesting that gonococcal AmiC transport is dependent on Tat, as is that of E. coli AmiA and AmiC (2).

Construction of amiC mutant and complemented strains.

To study the function of AmiC and its role in autolysis, a deletion mutation in amiC was constructed in the gonococcal chromosome via a two-step process. First, an insertion-duplication mutation was made in amiC in the gonococcal chromosome. To construct the insertion mutation, an internal region of amiC was cloned into pIDN3 (16). Since this plasmid is unable to replicate in N. gonorrhoeae, transformation of gonococcal strain MS11 with the plasmid and selection for Erm^r resulted in the isolation of transformants in which the plasmid had inserted in amiC, duplicating the cloned region. To construct the deletion, the region encoding amiC and approximately 500 bp of flanking DNA on each side was amplified from MS11 chromosomal DNA and inserted into pIDN3. An in-frame deletion of amiC was constructed in this plasmid, removing all but 15 bp of the coding sequence. The plasmid carrying the deletion was linearized, and the deletion was introduced into the gonococcal chromosome by transformation without selection. Potential transformants were screened for loss of Erm^r and for loss of amiC as detected by PCR.

For complementation of the amiC mutant, we created two strains carrying amiC complementation constructs incorporated at a distant location on the gonococcal chromosome. The first, DG121, carries the wild-type amiC gene under the control of the lac promoter/operator. The second, DG122, carries the amiC coding sequence and also includes its native promoter in addition to the IPTG-inducible lac promoter/operator. DG121 should not express significant amounts of amiC unless induced. DG122 is expected to give expression levels of amiC similar to wild type when not induced and increased expression of amiC when induced.

Mutation of amiC eliminates free disaccharide release.

To determine if AmiC functions as an amidase, we characterized PG fragments released by the wild-type, amiC, and amiC-complemented strains. N. gonorrhoeae releases several types of soluble peptidoglycan fragments that can be distinguished by size. These include PG multimers, PG monomers, and free disaccharide (40, 44). The release of free disaccharide is useful for gauging N-acetylmuramyl-l-alanine amidase action, since the liberation of disaccharide from PG requires the activity of both the amidase and a lytic transglycosylase. To determine the effects of amiC mutation on PG fragment release, the N. gonorrhoeae strains were pulse-labeled by growth in medium containing 6-[³H]glucosamine and lacking glucose. Supernatants containing radiolabeled PG released by the cells were collected after 2.5 h of growth. PG fragments in the supernatants were separated by size-exclusion chromatography and detected by scintillation counting. The resulting profile of the amiC mutant showed a loss of free disaccharide release (Fig. 1), thus confirming the role of AmiC as an amidase. Release of PG monomers was not affected. However, PG dimer release increased substantially. Small increases in released trimers and higher-order oligomers were also seen for the mutant. Wild-type strains also release a small amount of PG that is intermediate in size between monomers and dimers. This material has been shown to consist of a tetrasaccharide with only one of the muramic acid residues carrying a peptide (43, 44). Loss of this fragment in the amiC profile is also consistent with loss of amidase activity and suggests that glycosidically linked PG dimers are one substrate for AmiC. Complementation of amiC restored the release of free disaccharide to wild-type levels with a concomitant diminution of PG multimers (Fig. 1).

FIG. 1.

Mutation in gonococcal amidase (amiC) causes a loss of free disaccharide release and a gain in larger PG multimers. N. gonorrhoeae strains MS11 (wild type), DG005 (amiC), and DG122 (amiC + amiC⁺, complemented strain) were pulse-labeled by growth in medium containing 6-[³H]glucosamine and lacking glucose. Supernatants containing radiolabeled PG released by the cells were collected after 2.5 h of growth and separated over tandem size-exclusion columns. Labeled PG was detected by scintillation counting.

Autolysis under nongrowth conditions.

When suspended in buffer lacking nutrients and lacking divalent cations, gonococci do not grow but instead undergo rapid autolysis (17, 50). To determine if AmiC might act as an autolysin, we measured the decrease in optical density following transfer of the cells to buffer. Following growth in rich medium, gonococci were transferred to 50 mM Tris-HCl buffer (pH 8.0) at a starting OD₅₄₀ of 0.3. Under these conditions, the wild-type strain MS11 began to lyse immediately (Fig. 2). By contrast, the amiC mutant maintained significantly greater optical density, suggesting that the amiC mutant is deficient in autolysis. The complemented strain DG122 showed wild-type levels of optical density. However, when increased expression of amiC was induced with IPTG, DG122 showed much lower levels of optical density, presumably resulting from increased autolysis (Fig. 2). These results indicate that AmiC is capable of causing cell lysis.

Mutation of amiC results in altered growth and survival characteristics in vitro.

On agar plates, colonies of the amiC deletion mutant exhibited an opaque, granular appearance when viewed using a stereo dissecting microscope. When inoculated at equivalent optical densities and grown in liquid medium, the amiC mutant showed a slower increase in both the number of CFU per milliliter (Fig. 3A) and optical density (Fig. 3B) compared to the wild type. Cultures of the amiC mutant were approximately 10-fold lower in CFU at several points during log-phase growth. However, the amiC mutant exhibited a notable decline in CFU per ml while maintaining higher levels of optical density during the stationary and death phases (20 to 32 h) compared to the wild-type strain. Levels of total protein in the cells of each culture (Fig. 3C) showed that mutant and wild-type strains accumulated equivalent amounts of protein during growth, suggesting that differences in CFU per ml were not due to a growth deficiency in the amiC mutant. Complementation restored the wild-type growth characteristics (Fig. 3A and B) and transparent colony appearance on agar plates (data not shown).

FIG. 3.

An amiC mutant exhibits altered growth and survival characteristics in vitro. Aliquots of MS11 (wild type), DG005 (amiC), and DG122 (amiC + amiC⁺, complemented) cultures were taken, and CFU (A), optical density at 540 nm (B), and total protein concentration in the cell pellet (C) were measured. Panels A and B are representative of at least three separate trials. Panel C is an average of three separate trials. Error bars represent the standard deviations.

We also observed an unusual phenotype of the amiC mutant in liquid culture. When liquid cultures were grown to late stationary/death phase and then allowed to stand at room temperature, the wild-type cells settled out within 3 h, falling to the bottom of the culture tube. By contrast, the amiC mutant cells stayed in suspension, and the culture maintained higher optical density (Fig. 4A and B). This phenotype is similar to the autolysis-deficient gonococcal pldA (phospholipase A) mutant, which has been shown to maintain optical density in cell suspension after cell death (4). To further investigate the failure of the amiC cells to settle out of culture, we grew wild-type and amiC cultures to late stationary/death phase, centrifuged the cultures, and then resuspended the bacteria in either their own culture medium or the medium from the other culture. In the amiC culture medium the wild-type cells settled to the bottom of the culture tube slightly less than when resuspended in their own supernatant (Fig. 4C). Similarly, the amiC mutant cells settled less in their own supernatant than in supernatant from the wild-type culture. These results suggest that the amiC mutant cells release material into the medium that slows settling of the bacteria, possibly by increasing the viscosity of the medium. However, more of the amiC mutant cells remained suspended compared to the wild-type cells in either supernatant (Fig. 4C). This result suggests that the shape of the amiC mutant cells or cell aggregates may result in increased drag.

amiC mutant colony opacity is independent of opacity protein expression.

Our observation that amiC mutant colonies were always opaque on agar plates suggested a possible link with Opa protein expression. The opaque phenotype produced by opacity proteins can be identified using a stereo dissecting microscope and oblique substage lighting to differentiate these colonies from transparent (i.e., non-Opa-producing) colonies (27). Gonococci randomly switch opacity protein expression on and off by slipped-strand mispairing at an appreciable frequency (∼1/10⁴) (45). By using the complemented strain DG121, in which the amiC gene lacking its intrinsic promoter is put under the control of a lac promoter, we were able to examine the switch between opaque and transparent colony morphology. Colonies of DG121 appeared opaque on plates lacking IPTG, but upon transfer to IPTG-containing agar plates, the colonies appeared transparent. Controls DG005 and Opa⁺ MS11 showed no change in colony opacity when transferred to plates containing IPTG. With 100% of DG121 patches (128/128) converting to transparent colony morphology, this switch occurred at much too high a frequency to be due to Opa protein variation. Furthermore, Coomassie blue-stained SDS-PAGE outer membrane preparations of gonococcal strains show that Opa protein expression was not detected in the amiC mutant (Fig. 5), suggesting that the colony opacity phenotype of the amiC mutant does not require Opa protein expression.

FIG. 5.

amiC mutant colony opacity is independent of opacity protein expression. Coomassie blue-stained SDS-PAGE outer membrane preparations of gonococcal strains show that the wild-type opaque strain has an outer membrane protein band (arrows) that shifts in mobility following heating. No similar protein was detected in the amiC mutant (DG005), suggesting that mutant colony opacity is not linked with Opa protein expression.

Deletion of amiC diminishes cell separation.

Reasons for the aberrant growth characteristics of the amiC mutant could include decreased cell viability or a decrease in cell separation. A very similar growth phenotype was described by Cloud and Dillard (8) for N. gonorrhoeae mutants deficient in lytic transglycosylase C (LtgC) and was shown to be due to decreased cell separation. Similarly, thin-section transmission electron micrographs showed that, compared to wild-type strain MS11 (Fig. 6A and B), the amiC mutant is deficient in cell separation (Fig. 6C and D). Since gonococci divide in alternating planes (52), reduced separation following division results in aggregates. Counting the number of cells per cluster on a two-dimensional plane, we found that a significant number of the amiC cell clusters consisted of from 20 to over 40 cells per cluster. Also, relative to wild type, a small but notable increase in the number of lysed cells was observed. Complementation with a wild-type copy of amiC restored normal cell separation (Fig. 6E and F).

FIG. 6.

Deletion of amiC diminishes cell separation. Thin-section transmission electron micrographs at 8,800× (A, C, and E) and 15,000× (B, D, and F), respectively, show wild-type strain MS11 (A and B) as single cocci or diplococci; the amiC mutant (C and D) is deficient in cell separation. Since gonococci divide in alternating planes, reduced separation following division results in aggregates. Complementation with a wild-type copy of amiC (E and F) restored the wild-type phenotype.

Mutation of amiC increases cell lysis during growth.

Because mutation of amiC results in the formation of cell aggregates, CFU counts will greatly underestimate bacterial numbers, and spectrophotometric readings may not be directly comparable to those of wild-type cocci and diplococci. To measure cell lysis by a method that did not rely on turbidity or CFU determinations, we used an autolysis assay based on RNA release (P. L. Kohler and J. P. Dillard, unpublished). N. gonorrhoeae strains were grown in the presence of [³H]adenine to preferentially label the RNA (31), and the appearance of RNA in culture supernatants was used to reflect lysis. The wild-type and complemented strains showed a low degree of lysis (<10%) early in growth (3 h and 6 h), but lysis increased to ∼20% and 60% as the bacteria entered the stationary and death phases (9 h and 20 h, respectively) (Fig. 7). DG122 showed increased cell lysis compared to wild type at all time points when induced by IPTG (Fig. 7). The amiC mutant showed the greatest amount of RNA release; cell lysis was substantially increased relative to the wild type at all time points (Fig. 7).

Supporting evidence for increased lysis of the amiC mutant during growth came from use of a quantitative live-dead staining assay, where gonococcal cultures were grown in log phase (3.5 h) and cell viability was measured immediately after addition of the dyes. Microscopic examination of cells demonstrated immediate uptake of the dyes used in the live-dead staining assay. Thus, we could quickly assess the viability of the cultures by fluorometric readings, prior to autolysis that would occur due to suspension of the cells in buffer. The results of this assay showed the amiC mutant to be 39% (± 5% [standard deviation]) less viable than the wild type, whereas the complemented strain showed a wild-type level of viability (data not shown). The difference in viability between the amiC mutant and the wild-type strain was statistically significant at P < 0.001 by Student's t test.

Deletion of amiC increases outer membrane permeability.

Cell separation mutants of E. coli have been observed to display increased outer membrane permeability to various large and otherwise nonpermeable molecules, including vancomycin, lysozyme, and deoxycholate (21, 29). Similarly, N. meningitidis mltA mutants, which are defective in cell separation, show decreased membrane integrity and release outer membrane proteins into the medium (1). To determine whether the amiC mutation affected outer membrane permeability, we measured the susceptibility of each gonococcal strain to deoxycholate. We found that treatment of amiC mutant cultures with 0.01% deoxycholate, a concentration at which the wild-type strain is resistant, resulted in a 1,000-fold reduction in CFU/ml over 6 h (Fig. 8). These results suggest that cells lacking amiC form a unique cell cluster architecture that alters outer membrane integrity and increases permeability to macromolecules.

FIG. 8.

Deletion of amiC increases outer membrane permeability. Gonococcal cultures were grown in GCBL complete medium with deoxycholate (D) at 0.01% where indicated. The values depicted are the averages of four separate trials. Error bars represent the standard deviations. The linear fixed-effects statistical model P values were as follows: for MS11-treated versus DG005 plus D, 0.0004; DG005 untreated versus DG005 plus D, <0.0001. MS11-plus-D versus untreated cultures were not significantly different.

Characterization of PG fragments produced during autolysis.

Our results showing that the amiC mutant still underwent autolysis raised doubt about the nature of PG fragments released during lysis. Hebeler and Young found evidence of amidase activity but no evidence of glycan-splitting (glucosaminidase) activity (19). To further characterize the mechanism of autolysis in wild-type cells, we labeled gonococcal PG metabolically using 6-[³H]glucosamine and then transferred the cells to buffer. PG fragments released during lysis were analyzed by size-exclusion chromatography. A majority of the fragments were found in the void peak, consistent with the idea that autolysis releases large PG fragments (Fig. 9). Surprisingly, the second-most-abundant fragments were PG monomers. In growing cells, these fragments are known to be produced by lytic transglycosylases (7, 44). The amidase-specific fragment—free disaccharide—made up a minor peak. A peak of PG fragments smaller than disaccharide, likely monosaccharides, also made up a minor peak. These results suggest that lytic transglycosylases contribute to autolysis.

FIG. 9.

PG fragment release during autolysis under nongrowth conditions. Labeled gonococcal cells were allowed to lyse in Tris-HCl buffer at room temperature. PG fragments in the supernatants were filtered with a 0.45-μm filter prior to separation over tandem size-exclusion columns. PG was detected by scintillation counting.

DISCUSSION

We have shown that AmiC is necessary for cell separation in N. gonorrhoeae and is necessary for the production of free disaccharide from the bacterial cell wall. These data are consistent with the prediction that AmiC functions as a cell-wall-degrading N-acetylmuramyl-l-alanine amidase. Among the PG fragments released by the amiC mutant, the absence of glycosidically linked PG dimers with one peptide side chain also supports the idea that AmiC is an amidase. This result further suggests that glycosidically linked PG dimers may be one substrate for AmiC. The induced expression of AmiC resulted in decreased optical density for cells suspended in buffer (Fig. 2) and increased RNA release for growing cells (Fig. 7), suggesting that AmiC can function to break down the cell wall and cause cell lysis. N-Acetylmuramyl-l-alanine amidases are involved in autolysis in several other bacterial species. For example, in S. pneumoniae, an enzyme with this activity acts as a major autolysin and is necessary for cell separation following division (42, 48). The E. coli amidase, AmiC, is involved in antibiotic-induced autolysis (20).

Contrary to our expectations, the amiC deletion mutant was not deficient in autolysis but was instead more prone to lysis. Release of RNA was significantly greater in the amiC mutant compared to wild type, and an increased number of lysed cells were observed in thin-section electron micrographs of the amiC mutant cells. An important conclusion from these results is that amiC is not necessary for autolysis to occur. How then can it be that Hebeler and Young found the amidase-specific products to be the most prevalent PG fragments produced during autolysis? Hebeler and Young used cell extracts added to purified PG and then analyzed the products for an increase in free reducing ends and for an increase in N-terminal alanine (19). Their results showed a clear increase in N-terminal alanine over time without a concomitant increase in C-terminal alanine, clearly indicating the activity of an amidase. Only a small increase in free reducing ends was seen, leading the authors to conclude that glycan-degrading peptidoglycanases were not present. However, at the time those studies were done, lytic transglycosylases were just being described (23) and Hebeler and Young would not have known to look for PG fragments containing anhydro (nonreducing) ends. Our results suggest that lytic transglycosylases are active during autolysis (Fig. 9). Therefore, a lytic transglycosylase might serve as the autolysin instead of an amidase. Alternatively, the process of autolysis may involve multiple peptidoglycanases acting in the degradation of the cell wall.

It may be that AmiC is an important player in autolysis but that effects of the lack of cell separation in the amiC mutant prevent accurate measurement of this phenotype. The thin-section electron micrographs show that amiC mutant cells are found in clusters, often with 15 or more bacteria in the plane of the section. Nearly every cluster shows one or more dead (less electron dense) bacteria attached to the cluster. The cell wall material appears to be a single continuous cell wall contorted into the odd shape of the cell cluster. Similarly, the cluster likely has a single large outer membrane covering the entire cluster. More membrane blebs are seen surrounding the amiC mutant clusters than around the wild-type cells. This result, along with the increased membrane permeability to deoxycholate, suggests that the outer membrane stability is decreased in the amiC mutant. Thus, lysis occurring in our amiC mutant aggregates may be a direct effect of outer membrane instability resulting from the unusual cell architecture. It is known that autolysis can be triggered by membrane instability, and it was recently shown that N. gonorrhoeae mutants deficient in phospholipase A are decreased in autolysis (4). Alternatively, the absence of AmiC may allow another peptidoglycanase to act in an uncontrolled manner, leading to autolysis. This idea is consistent with the hypothesis that autolysis results from peptidoglycanases acting in an uncontrolled manner under nonideal conditions (24).

Altered growth characteristics similar to those of the amiC mutant have been observed before in our laboratory in gonococcal lytic transglycosylase C (ltgC) mutants (8). Interestingly, mutation of ltgC also abolished the production of free disaccharide, giving rise to a PG fragment release profile very similar to that of the amiC mutant (8). These results indicate that these two peptidoglycanases are involved in the same process in cell separation. One explanation for these results would be the formation of a multienzyme complex involved in septum synthesis and cell separation. Evidence for PG synthesis enzymes in association with PG degradation enzymes has been found in E. coli (39). Also, in N. meningitidis the LtgC homologue (MltA) was found to associate with PBP2, a PG-synthesizing enzyme similar to FtsI (26). Thus, the gonococcal AmiC may also assemble with LtgC and PBP2 into a complex involved in cell division and cell separation. Supporting this possibility is the observation that gonococcal cultures grown in a subinhibitory concentration of penicillin G produced a fragment release profile very similar to that of our amiC mutant, with the only difference being that the free disaccharide peak was notably diminished but not completely abolished (43). Penicillin G has been shown to bind to gonococcal PBP2 (11), and subinhibitory β-lactam-treated cells demonstrate a higher proportion of unseparated cells (30).

In conclusion, our data indicate that AmiC is an N-acetylmuramyl-l-alanine amidase that acts in cell separation. amiC mutants show growth irregularities, increased lysis, and increased cell permeability. These growth defects as well as the increased cell permeability suggest that cell separation may be a novel process to target for antibiotic therapy. Although AmiC is capable of lysing the cell, amiC deletion mutants still undergo autolysis. Therefore, it is likely that multiple enzymes are involved in this process, and further work will be required before the process of autolysis in N. gonorrhoeae is understood.