α-2,3-Sialyltransferase Enhances Neisseria gonorrhoeae Survival during Experimental Murine Genital Tract Infection

Abstract

The addition of host-derived sialic acid to Neisseria gonorrhoeae lipooligosaccharide is hypothesized to be an important mechanism by which gonococci evade host innate defenses. This hypothesis is based primarily on in vitro assays of complement-mediated and phagocytic killing. Here we report that a nonpolar α-2,3-sialyltransferase (lst) mutant of N. gonorrhoeae was significantly attenuated in its capacity to colonize the lower genital tract of 17-β estradiol-treated female BALB/c mice during competitive infection with the wild-type strain. Genetic complementation of the lst mutation restored recovery of the mutant to wild-type levels. Studies with B10.D2-HC^oH2^dH²-T18c/OSN (C5-deficient) mice showed that attenuation of the lst mutant was not due to increased sensitivity to complement-mediated bacteriolysis, a result that is consistent with recently reported host restrictions in the complement cascade. However, Lst-deficient gonococci were killed more rapidly than sialylated wild-type gonococci following intraperitoneal injection into normal mice, which is consistent with sialylation conferring protection against killing by polymorphonuclear leukocytes (PMNs). As reported for human PMNs, sialylated gonococci were more resistant to killing by murine PMNs, and sialylation led to reduced association with and induction of a weaker respiratory burst in PMNs from estradiol-treated mice. In summary, these studies suggest sialylation confers a survival advantage to N. gonorrhoeae in mice by increasing resistance to PMN killing. This report is the first direct demonstration that α-2,3-sialyltransferase contributes to N. gonorrhoeae pathogenesis in an in vivo model. This study also validates the use of experimental murine infection to study certain aspects of gonococcal pathogenesis.

Neisseria gonorrhoeae is responsible for a large number of uncomplicated infections of the lower urogenital tract and serious upper reproductive tract infections in the United States (7) and the developing world (14). The gonococcus exclusively infects humans and has no outside animal or environmental reservoir. Accordingly, this organism has evolved many elegant adaptation mechanisms to maintain its existence in the human population, despite numerous hostile physiological and immunological factors that constitute the host mucosal defense. Phagocytes and complement are two critical antimicrobial components of the innate immune system; the existence of a sophisticated interplay between N. gonorrhoeae and these two factors during genital tract infection is supported by the persistence of the organism amid an intense influx of polymorphonuclear leukocytes (PMNs) during acute infection (22) and the detection of complement deposition products on gonococci isolated from endocervical lavage (32).

The gonococcus has evolved several mechanisms to evade host complement and phagocytes, including the covalent transfer of sialic acid to the terminal galactose of the lacto-N-tetraose moiety of the lipooligosaccharide (LOS) α-chain. This modification, which is catalyzed by gonococcal α-2,3-sialyltransferase (29, 30), is the basis of the unstable serum resistance phenotype exhibited by gonococci within patient exudates (41, 42) in which resistance to the bactericidal activity of normal human serum (NHS) is rapidly lost upon subculture in the absence of cytidine monophospho-N-acetylneuraminic acid (CMP-NANA). Unlike N. meningitidis, this substrate must be supplied to N. gonorrhoeae by the host (31, 41, 59).

LOS sialylation is predicted to have a significant impact on gonococcal pathogenesis based on in vitro assays of complement and phagocytic function (reviewed in reference 52). In addition to protection from cross-reacting bactericidal antibodies present in serum from naïve hosts (16), LOS sialylation increases gonococcal resistance to opsonic (15, 28, 61) and nonopsonic (45) killing by human PMNs. LOS sialylation also protects N. gonorrhoeae from the bactericidal activity of porin and LOS-specific antibodies produced via immunization (10, 15, 40, 61), and therefore it may contribute to the ease with which N. gonorrhoeae causes repeated infections. At the molecular level, sialylated LOS down-regulates complement activation by binding factor H (fH), a negative regulatory protein of the alternative pathway of the complement cascade (44). Factor H inactivates the alternative pathway convertase (38, 39, 60) and acts as a cofactor in the cleavage of C3b to C3bi (39). The result of fH binding to the bacterial surface is reduced surface deposition of C3b; C4 deposition is also reduced, a step that is needed for activation of the classical pathway (15, 32). Fully sialylated gonococci also do not bind mannan-binding lectin, a key activator of the lectin complement pathway (17).

The apparent selection for the lacto-N-neotetraose LOS species during experimental urethral infection of volunteers (47) indirectly supports a role for LOS sialylation during genital tract infection, as does immunochemical evidence that a majority of gonococci with the lacto-N-neotetraose structure in urethral exudates is sialylated (1). However, direct testing of the importance of LOS sialylation in pathogenesis using an in vivo infection model has not yet been reported. As the concentration of CMP-NANA is limiting with respect to its effect on serum resistance and PMN killing in vitro (13, 45, 61), it is critical that the role of LOS sialylation in pathogenesis be tested in models with physiologically relevant levels of CMP-NANA and complement. Such models should also reproduce the O₂ tension of mucosal surfaces due to the reported influence of sialylation on the induction of the oxidative burst in PMNs (45); the importance of O₂ tension is also indicated by the demonstration that anaerobic growth increases CMP-NANA-dependent serum resistance (13) and reports that LOS biosynthesis genes may be oxygen regulated (5, 13). Although several host restrictions challenge animal modeling for gonorrhea, female mice in the proestrus stage of the estrous cycle can be transiently colonized with N. gonorrhoeae (9). Long-term infection can be accomplished by administering 17-β estradiol, which stabilizes the estrous cycle in an estrus-like state (23, 56). An inflammatory response occurs in 50 to 80% of infected mice (23, 24), which, together with similarities in O₂ tension, iron limitation, and pH between the human endocervix and the mouse lower genital tract, supports the use of female mice as a surrogate host for studying the role of gonococcal α-2,3-sialyltransferase in infection.

Here we tested the importance of LOS sialylation in N. gonorrhoeae adaptation to the genital tract by studying the survival of an isogenic nonpolar lst mutant during experimental murine infection. We also used in vitro assays of phagocytic function to investigate our in vivo results and to further define the potential of using estradiol-treated mice to study gonococcal evasion of innate host defenses.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

N. gonorrhoeae MS11 (PorB.1B, Sm^r) is a serum-resistant strain from a patient with uncomplicated endocervical infection. Strain MS11 has been tested extensively in male volunteers (47, 48, 55). N. gonorrhoeae was cultured on GC agar (Difco) with Kellogg's supplements (27) at 37°C under 7% CO₂ or in supplemented GC broth (GCB) with 5 mM NaHCO₃ in an air incubator with rotation as described previously (25). For experiments in which sialylated gonococci were tested, bacteria were cultured overnight in GCB containing saturating concentrations (50 μg/ml) of CMP-NANA. GC agar containing kanamycin (Km) (50 μg/ml) or erythromycin (Em) (3 μg/ml) was used to select lst mutant GP300 and the complemented mutant GP322, respectively. GC-vancomycin, colistin, nystatin, trimethoprim, streptomycin (GC-VCNTS) agar was used in single-strain mouse infection experiments as described previously (23). For competitive infections, GC agar containing streptomycin (Sm) (100 μg/ml) and Sm plus Km (50 μg/ml) was used. Escherichia coli TOP10 (Invitrogen) was the host strain for all recombinant plasmids and was cultured on Luria-Bertani (LB) agar or in LB broth at 37°C. Km (50 μg/ml), ampicillin (Amp) (200 μg/ml), tetracycline (Tet) (30 μg/ml), and Em (300 μg/ml) were used to maintain plasmids in E. coli when indicated. CMP-NANA and antibiotics were from Sigma Biochemical Corporation.

Construction of an lst-deficient mutant (GP300).

The lst gene and a 10-bp neisserial DNA uptake sequence (11) located 122 bp downstream of the translational stop site were amplified from chromosomal DNA of strain MS11 via PCR using primers F1-lst (5′ CGTCGCGAATGGAGTTTTTAGGATATGGG) and R1-lst (GCGTCGACCTGCCACGACAGTTTCCG). An NruI or SalI site (underlined) was introduced into each primer, respectively, to facilitate cloning. F1-lst corresponds to the 17 nucleotides directly upstream of the lst translational start codon, and R1-lst corresponds to a 20-bp region 181 nucleotides downstream of the lst open reading frame. The resultant 1.343-kb PCR product was cloned into PCR-Blunt vector (Invitrogen) to generate plasmid pHW501, and the insert was sequenced to confirm its identity. The lst gene and uptake sequence were subcloned from pHW501 into the SalI and NruI sites of pBR322 to create pHW501a, thereby changing the selectable marker from Km^r to Amp^r. A nonpolar aphA-3 (Km^r) cassette was obtained from plasmid pUC18Km on an 840-bp SmaI fragment (34) and inserted into a unique EagI site at nucleotide position 672 of the lst gene to create pHW501a-Km. Restriction endonuclease analysis and nucleotide sequencing were used to confirm that the transcriptional orientation of the aphA-3 cassette was the same as that of the lst gene. The inactivated lst gene was PCR amplified from pHW501a-Km using primers F1-lst and R1-lst, and the purified PCR fragment was introduced into MS11 by spot transformation (18). The occurrence of allelic exchange in a Km^r transformant (GP300) was verified by PCR using primers F45 (5′ACAGCCGGTATAAAGGGACCAC) and R46 (5′ACGCAGAAGGCAATGTCATACC) to detect the aphA-3 cassette and by PCR using primers F2-lst (5′AACCTGATACGGGAGAGCAGCT) and R2-lst (5′ATGCGTAAACGTGCACATTGG), which anneal within the lst gene to sequences on each side of the EagI site into which the aphA-3 gene was inserted. Primers F1-lst and R3 (5′GTTTGGTTCGGATGGTGCAA) were used to PCR amplify the mutated lst gene in mutant GP300, and the mutation was confirmed by nucleotide sequence analysis. To complement the lst mutation, we took advantage of the Neisserial Insertional Complementation System (NICS) (provided by H.S. Seifert, Northwestern University), which allows for genetic complementation by targeting a chromosomal intergenic region of no known function between the lactate permease (lctP) and aspartate transaminase (aspC) genes (33, 50). We were unable to successfully clone the wild-type lst gene with its own promoter. Therefore, the lst structural gene was obtained on a SalI, NruI fragment from pHW501 and cloned into the PmeI site of the NICS vector pGCC4, which has an inducible lac promoter to allow regulated expression of genes cloned downstream to create pGCC4.Lst. An 8-kb ClaI fragment that contained the lst gene, lac promoter, and an Em^r gene flanked by gonococcal DNA corresponding to the lctP aspC locus was purified from pGCC4.Lst, and it was transformed into mutant GP300 to create strain GP322 (Em^r, Km^r). Confirmation of the desired allelic exchange in strain GP322 was by PCR and sequence analysis. Isopropyl-β-d-thiogalactopyranoside (IPTG) (1 mM) (Sigma) was used to maximally induce lst transcription in strain GP322 in serum resistance studies; IPTG induction was not used when testing interactions between GP322 and PMNs (i.e., opsonophagocytic killing, PMN uptake, and chemiluminescence assays). Ligations, electroporation, and PCR were performed using standard methods. Restriction enzymes, DeepVent polymerase, and DNA ligase were from New England Biolabs. Plasmid isolation and purification of PCR products were performed with the QIAprep Spin miniprep and gel extraction kits (QIAGEN), respectively. Synthesis of oligonucleotide primers and automated nucleotide sequencing using the Big Dye Terminator V3.1 Cycle Sequencing kit (Applied Biosystems) were by the Biomedical Instrumentation Center of the Uniformed Services University of the Health Sciences (USUHS). Primers were designed based on sequence information from the N. gonorrhoeae FA1090 genome database (http://www.genome.ou.edu/gono.html).

Serum bactericidal assays.

Sensitivity of gonococci to NHS or heat-inactivated (HI) NHS was tested in a microtiter plate assay as described previously (23). NHS was obtained from venous blood collected from a healthy female volunteer under a protocol approved by the USUHS Institutional Review Board. HI-NHS was obtained by incubating serum at 56°C for 30 min. Bactericidal assays were performed using bacteria cultured in GCB in the presence or absence of CMP-NANA. Broth cultures were centrifuged at 10,000 rpm for 30 s, and the bacterial pellet was suspended in PBS containing 0.1% gelatin, 0.01% CaCl₂, and 0.01% MgCl₂ (PBS-GCM) to a concentration of 10⁵ CFU/ml. To directly test gonococci in mouse vaginal mucus, samples were collected on a sterile swab and suspended in 1 ml PBS-GCM. Bacterial aggregates were removed by centrifugation in a microcentrifuge at 3,000 rpm for 10 s. The supernatant was centrifuged at 10,000 rpm for 30 s, and the pellet was resuspended in PBS-GCM. In some experiments, bacteria (10⁶ CFU) were incubated with neuraminidase (NANase) (Clostridium perfringens type V) (Sigma) at a concentration of 8.3 U/ml for 1 h at 37°C or in a similar volume of distilled H₂O (control). The bactericidal₅₀ titer was defined as the percentage of NHS from which the number of CFU was 50% of that recovered from the assay medium without serum.

Opsonophagocytic killing assay.

Mouse PMNs were elicited via intraperitoneal (i.p.) injection of 2.5 ml of 3% thioglycollate broth (Difco) into estradiol-treated (3 to 14 days posttreatment) or untreated female BALB/c mice. Mice were humanely sacrificed 4 to 5 h postinjection, and PMNs were obtained by peritoneal lavage with 10 ml cold Hanks’ balanced salt solution (HBSS) and centrifuged for 5 min at setting 3 of an IEC Model CL centrifuge (International Equipment Co., Needham Heights, MA). The cell pellet was washed with an equal volume of HBSS and centrifuged as described above. The washed cells were resuspended in HBSS containing 10 mM glucose, 0.1% gelatin, 1 mM CaCl₂, and 1 mM MgCl₂ (HBSS II) and adjusted to 2 × 10⁷ cells per ml using a hemacytometer. Trypan blue exclusion staining was used to confirm that >95% of the PMNs were viable. Normal mouse serum (NMS) was used for opsonization and separated from blood obtained by cardiac puncture; serum was placed on ice and used within an hour after collection. The PMN killing assay was performed essentially as described previously (12). Briefly, 10⁵ to 10⁶ CFU of gonococci were suspended in 500 μl of HBSS II containing 10% NMS (opsonized) or 10% HI-NMS (control) for 10 to 15 min at 37°C. Mouse PMNs were added (10⁶ PMNs in a 50-μl volume), and the final suspension was incubated on a rotary shaker at 37°C. Samples (20 μl) were taken in quadruplicate at 45 and 90 min of incubation, serially diluted in GCB with 0.05% saponin, and cultured on GC agar with Sm for 24 to 36 h. The mean number of CFU recovered from tubes in which opsonized bacteria were tested was compared to that recovered from tubes in which gonococci were preincubated with HI-NMS (viability control) for each time point.

Acridine orange staining.

The number of PMNs with associated gonococci was assessed using acridine orange staining. Briefly, 5 × 10⁵ PMNs suspended in 250 μl HBSS II were inoculated onto coverslips (12 mm diameter) within a 24-well tissue culture dish. Gonococci were added at a ratio of 10 GC to 1 PMN. Coverslips were gently rinsed after 90 and 135 min of incubation, stained with acridine orange (2), and examined with an Olympus BX60 fluorescent microscope using an excitation filter of 488 nM. One-hundred PMNs were examined on each of three coverslips, and the number of PMNs associated with no gonococci or with 1 to 2, 3 to 10, or >10 gonococci was recorded.

Experimental murine infection.

Infection of estradiol-treated female BALB/c mice (National Cancer Institute, Bethesda, MD) or congenic female B10.D2-HC^oH2^dH²-T18c/OSN (C5⁻) and B10.D2-HC¹H2^dH²-T18c/NSN (C5⁺) mice (Jackson Laboratories, Bar Harbor, Maine) (8) was essentially as described previously (24). For single-strain infections, mice were inoculated intravaginally with 10⁴, 10⁵, or 10⁶ CFU of piliated MS11 or GP300 gonococci suspended in 20 μl of PBS-GCM (n = 8 mice per group). Vaginal mucus was cultured quantitatively for N. gonorrhoeae every other day for 14 days. Variants of wild-type and GP300 bacteria that expressed a single Opa protein of the same apparent molecular weight on sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels were used for all experiments. For competitive infections, groups of 8 to 10 mice were inoculated with mixed suspensions of approximately equal numbers of wild-type MS11 and GP300 or GP322 gonococci (total dose, 10⁶ CFU per 20-μl inoculum). Nonpiliated variants were used to minimize the transfer of the antibiotic resistance marker in vivo. Equal volumes of undiluted and diluted vaginal swab suspensions were cultured on GC agar with Sm (total) and Sm plus Km (GP300 or GP322) on days 2, 4, 6, and 8 postinoculation. The ratio of GP300 or GP322 CFU to the calculated number of wild-type CFU was determined for inoculum (input) and vaginal cultures (output), and the competitive indices (CI; output ratio over input ratio) were calculated as described previously (3). For i.p. challenge studies, overnight cultures of MS11 and GP300 bacteria grown in GCB with CMP-NANA were centrifuged at 10,000 rpm at room temperature. Bacterial pellets were resuspended in sterile PBS and adjusted to an A₆₀₀ of 0.08. Groups of 12 female BALB/c mice treated 4 days previously with 17-β estradiol per the infection protocol were injected i.p. with approximately 250 μl of the bacterial suspensions (2 × 10⁷ CFU), and three mice in each group were killed at 0 h, 3 h, 4 h, and 5 h postinjection. The peritoneal cavities were rinsed with 5 ml of PBS, and the lavage fluid was diluted at least 1:10 in 0.1% tryptic soy broth to lyse the phagocytes. The number of CFU recovered from 250 μl of lavage fluid was determined by quantitative culture for each time point. Animal experiments were conducted in the laboratory animal facility at USUHS, which is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care, under a protocol approved by the USUHS Institutional Animal Care and Use Committee.

Chemiluminescence assay.

The NADPH-oxidase activity in PMNs isolated by peritoneal lavage was measured using isoluminol- or luminol-dependent chemiluminescence (CL) assays that allow separate determinations of extracellularly released and intracellularly generated reactive oxygen species (ROS), respectively (6). CL activity was measured in an MLX microtiter plate luminometer (Dinex Technologies, Inc., Chantilly, VA) using white luminescence 96-well microtiter plates and a 350-μl reaction mixture containing 10⁶ PMNs isolated from mice 4 to 11 days after initiating estradiol treatment. For measuring extracellular ROS release, assays were conducted with 4 U horseradish peroxidase (a cell-impermeable peroxidase) and 3.75 × 10⁻⁵ M isoluminol (a cell-impermeable CL substrate). To measure intracellular ROS, assays contained 50 U superoxide dismutase (SOD; a cell-impermeable scavenger of O₂⁻), 2,000 U catalase (a cell-impermeable scavenger of H₂O₂), and 3.75 × 10⁻⁵ M luminol (a cell-permeable CL substrate). The plate was equilibrated for 15 min at 37°C, after which 10⁷ CFU of opsonized bacteria were added (ratio of bacteria to PMN, 10:1). Light emission was recorded continuously starting 5 min after the addition of bacteria. Experiments were performed on three separate occasions with similar results.

Data analysis.

An unpaired t test was used to evaluate differences in the average duration of recovery from mice, susceptibility to PMN killing, the number of viable gonococci recovered following i.p. challenge, and the number of PMNs with associated gonococci in the acridine orange assay. All statistical analyses were performed with SPSS software (Chicago, IL). All experiments were performed at least twice to test reproducibility.

RESULTS

In vitro characterization of an α-2,3-sialyltransferase-deficient mutant.

The lst mutant GP300 was constructed via allelic exchange with an insertionally inactivated lst gene (lst::aphA-3) as described in Materials and Methods. Mutant GP300 was functionally deficient for sialyltransferase activity, as evidenced indirectly by the absence of a serum-resistant phenotype following growth in the presence of CMP-NANA (Table 1). To confirm that the observed increase in serum sensitivity was due to an inactivated lst gene, we showed serum resistance was restored via complementation in trans using strain GP322, a derivative of mutant GP300, which carries an inducible copy of the lst gene in a nonessential region of the chromosome. Growth of GP322 in the presence of CMP-NANA and IPTG to maximally induce expression of the intact lst gene resulted in serum resistance at a level similar to that of wild-type strain MS11. A slightly lower level of serum resistance was displayed by GP322 when cultured in the presence of CMP-NANA but not IPTG. NANase treatment resulted in a fully serum-sensitive phenotype (data not shown); therefore, the serum resistance shown by GP322 in the absence of IPTG induction is likely due to leakage from the lac promoter in the complementation vector at levels sufficient for conferring serum resistance.

TABLE 1.

Sensitivity of wild-type (MS11), Lst-deficient (GP300), or complemented mutant (GP322) gonococci cultured in vitro with or without CMP-NANA or cultured in vivo

Culture type	% NHS required to kill 50% of gonococci of strain:
	MS11	GP300	GP322
GCB	3	3	3
GCB + CMP-NANA	>16	3	>16c
GCB + CMP-NANA + NANase	4	3	4
Vaginal swab suspensiona	>16	3	10
Vaginal swab suspension + NANase	3	4	4
Vaginal cultureb	3	3	2
Vaginal culture + CMP-NANA	>16	3	10d
Vaginal culture + CMP-NANA + NANase	3	2	2

^aGonococci within vaginal swab suspensions tested directly without culture.

^bGonococci tested after one in vitro passage from infected mice.

^cIPTG was used to induce the lst gene in GP322.

^dGP322 cultured in vitro in the absence of IPTG.

Gonococci in the mouse genital tract exhibit unstable serum resistance.

High concentrations of CMP-NANA (2 to 500 μg/ml) are often used to demonstrate the effect of sialylation on gonococcal resistance to bactericidal antibodies (13, 15, 61) and to PMN uptake and killing (28, 45). These concentrations are 100- to 1,000-fold higher than those in physiological conditions. To test if LOS sialylation occurs during murine infection at a level that is functionally similar to that exhibited by gonococci cultured under saturating amounts of CMP-NANA, we first determined if wild-type gonococci within vaginal swab suspensions from infected mice exhibited an unstable serum resistance phenotype. Mice were inoculated intravaginally with 10⁶ CFU of wild-type strain MS11 or lst mutant GP300 gonococci, and the serum sensitivity of gonococci within vaginal mucus (without in vitro passage) was determined over the course of infection. Wild-type gonococci within vaginal swab suspensions were highly resistant to NHS as early as 1 day after inoculation and on every day tested (through day 9). Incubation of vaginal swab suspensions with NANase resulted in loss of the serum-resistant phenotype, as did a single in vitro passage of the wild-type strain from infected mice in the absence of CMP-NANA (Fig. 1). In contrast, GP300 gonococci were sensitive to NHS throughout the course of the experiment. Vaginal washes from mice infected with the complemented mutant GP322 also contained serum-resistant gonococci, although the bactericidal₅₀ titer was slightly less than that of the wild-type strain. NANase treatment dropped the bactericidal₅₀ titer of in vivo-grown GP322 to the serum-sensitive range, however, which confirmed that the addition of sialic acid was responsible (Table 1). The slightly lower serum resistance of in vivo-grown GP322 gonococci compared to that of MS11 bacteria is most likely due to the absence of IPTG induction during infection.

FIG. 1.

Serum resistance of in vivo- and in vitro-grown wild-type N. gonorrhoeae. The number of viable gonococci recovered following incubation in increasing concentrations of NHS is shown for wild-type MS11 gonococci tested directly in vaginal swab suspensions or following in vitro subculture in the presence or absence of CMP-NANA. Of all conditions tested, only gonococci within vaginal swab suspensions and/or isolated from in vitro culture in the presence of CMP-NANA exhibited high levels of serum resistance. NANase was used to confirm the basis of the serum resistance.

We found it necessary to partially fractionate vaginal mucus from infected mice before we could detect serum resistance of gonococci within the suspensions, and therefore we hypothesize sialidases or inhibitors of sialyltransferase may be present in mouse vaginal fluid. A low-speed centrifugation was used to remove epithelial cells and debris; a higher speed was required to remove a more soluble inhibitor from the supernatant. These findings are consistent with a protocol used by others to process human urethral exudate samples before performing bactericidal assays (42, 59). Martin et al. (31) postulated that concentrations of CMP-NANA in biological samples are limiting, based on variability in the capacity of human seminal plasma, cervical scrapings, and vaginal exudates to confer serum resistance when added to growth media. Consistent with this proposal, we were not successful in inducing serum resistance by culturing wild-type gonococci in broth containing 20% vaginal wash material from mice.

Attenuation of the lst mutant in vivo.

We first tested the capacity of Lst-deficient gonococci to successfully colonize estradiol-treated BALB/c mice by comparing the infectivity of wild-type and Lst-deficient gonococci in mice inoculated with either MS11 or lst mutant GP300 gonococci. We detected no difference in the duration of colonization over time using doses of 10⁶, 10⁵, and 10⁴ CFU, which include and flank the 80% infectious dose (10⁵ CFU) for strain MS11 (data not shown). To further examine the importance of α-2,3-sialyltransferase in murine genital tract infection, we next utilized the more sensitive technique of competitive infection (3). For these experiments, mice were inoculated with a suspension containing wild-type gonococci mixed with similar numbers of the lst mutant GP300 or the complemented strain GP322 bacteria. A dramatic decrease in the recovery of the lst mutant relative to the wild-type strain occurred early in infection, as shown in Fig. 2. Based on a value of 1.0 as indicative of no change in the recovery of mutant relative to that of wild-type bacteria, the mean competitive index (CI) for mice inoculated with wild-type and GP300 gonococci was greater than 10-fold reduced within 2 days postinoculation; 10²- to 10⁵-fold decreases in the CI were detected in several mice. The mutant was no longer recovered after 2 to 6 days from 12 out of 15 mice (Fig. 2, open symbols), and the mean length of time that mutant and wild-type gonococci were detected in vaginal washes was 4.7 versus 9.5 days, respectively (two experiments combined; P = 0.001). In contrast, recovery of the complemented mutant GP322 was similar to that of the wild-type strain, and the mean CI was statistically different from that obtained from mice inoculated with GP300 and wild-type gonococci on days 2 and 4 postinoculation (P = 0.003 and 0.006, respectively). No significant difference between the mutant and complemented control groups was seen on day 6. However, unlike the GP300 mutant, which cleared in most mice by day 6, the complemented strain was recovered from all mice but one at the later time points, and there was no significant difference in the average length of time that GP322 and wild-type bacteria were recovered from mice (9.1 and 11.5 days, respectively). These in vivo complementation data rule out any potential difference in the in vivo growth rate between the mutant and wild-type strain that is unrelated to the lst gene; we also detected no difference in the growth rate of MS11, GP300, or GP322 gonococci when cultured separately in broth or in mixed broth cultures (data not shown). We conclude from these experiments that possession of a functional lst gene confers a detectable survival advantage in the lower genital tract of mice.

FIG. 2.

Recovery of the lst mutant (GP300) or complemented mutant (GP322) relative to the wild-type strain. Results are expressed as competitive indices (CI) from individual mice at each time point. The CI was calculated as described in Materials and Methods. The ratios of mutant to wild-type gonococci in the inoculum for all experiments were 0.47 to 0.50. Horizontal bars represent the geometric mean; a CI of <1.0 indicates a decrease in the ratio of mutant to wild-type gonococci with respect to that of the inoculum. Open symbols represent mice from which at least 100 wild-type CFU, but no GP300 or GP322 CFU, were recovered; the limit of detection (0.04 CFU per 100 μl of swab suspension) was used as the number of mutant CFU recovered in these cases.

Increased susceptibility of Lst-deficient gonococci to uptake and killing by mouse PMNs.

Based on the demonstration that LOS sialylation promotes resistance to opsonophagocytic killing by human PMNs (15, 61), we assessed the role of sialyltransferase in protecting N. gonorrhoeae from PMNs isolated from estradiol-treated mice. Human phagocytes can take up gonococci via the C3b receptor or, in the absence of opsonins, via the binding of certain gonococcal opacity (Opa) proteins to carcinoembyronic antigen cellular adhesin molecules on the PMN (21). The latter pathway is host restricted, and therefore we opsonized bacteria with NMS for these assays. Following preincubation with NMS, nonsialylated but not sialylated wild-type MS11 gonococci were significantly killed after 45 and 90 min of exposure to PMNs, based on comparison with the number of CFU recovered when HI-NMS was used. In contrast, opsonized, Lst-deficient GP300 gonococci were significantly killed by murine PMNs regardless of whether the bacteria were cultured in the presence or absence of CMP-NANA. CMP-NANA-dependent resistance to PMN killing was restored to wild-type levels through complementation of the lst mutation; IPTG induction of the complemented gene was not needed to restore resistance (Fig. 3). No appreciable decrease in viable counts occurred in the HI-NMS controls over the course of the assay; the lack of PMN killing when using HI-NMS supports the need for complement opsonization in the mouse system. Additionally, the killing capacity of PMNs from untreated or estradiol-treated mice against nonsialylated and sialylated bacteria was comparable (data not shown).

FIG. 3.

Susceptibility of wild-type (MS11), Lst-deficient (GP300), or complemented lst mutant (GP322) to opsonophagocytic killing by murine PMNs. Gonococci cultured in the presence or absence of CMP-NANA or with CMP-NANA followed by NANase treatment were preincubated with NMS (opsonized) or HI-NMS (control) and then incubated with murine PMNs. GP322 was cultured without IPTG for these experiments. Results are expressed as the mean number of CFU recovered after 90 min of incubation; standard error bars are based on samples cultured in quadruplicate. *, P < 0.05 compared to HI-NMS control; **, P < 0.005 compared to HI-NMS control.

The observed protection against opsonophagocytic killing that is afforded by sialylation may be explained by reduced bacterial adherence to and uptake by PMNs due to decreased deposition of complement opsonins. We therefore compared the degree to which sialylated versus nonsialylated gonococci associated with PMNs following opsonization. More PMNs with associated gonococci were detected when nonsialylated bacteria were tested compared to that observed for wild-type bacteria cultured in the presence of CMP-NANA (Fig. 4A). As expected, preincubation of wild-type gonococci in HI-NMS did not promote efficient association with PMNs, regardless of whether the bacteria were sialylated or nonsialylated. The number of gonococci associated with PMNs was also higher for samples that contained opsonized, nonsialylated wild-type or lst mutant gonococci compared to those with opsonized, sialylated wild-type bacteria. Complementation of the lst mutation restored association to wild-type levels (Fig. 4B).

FIG. 4.

Association of sialylated or nonsialylated gonococci with murine PMNs. Results were quantitated by examining 100 PMNs for each of three coverslips stained with acridine orange and determining the average number of PMNs associated with gonococci and the number of gonococci per PMN. A. Average number of PMNs associated with wild-type MS11 gonococci preincubated with NMS (opsonized) or HI-NMS. Standard error bars are shown. Significantly more PMNs were associated with opsonized gonococci (P < 0.05; unpaired t test) regardless of whether MS11 was cultured with or without CMP-NANA. B. Average number of PMNs associated with 0, 1 to 2, 3 to 10, or >10 MS11 or GP322 gonococci cultured in the presence or absence of CMP-NANA or GP300 gonococci cultured with CMP-NANA. All gonococci were preincubated in NMS. The average total number of PMNs associated with MS11 or GP322 gonococci cultured in CMP-NANA (and without IPTG) was statistically different from that detected for nonsialylated MS11, GP322, or GP300 (P < 0.05; unpaired t test).

Sialylation reduced but did not abolish opsonic interactions between N. gonorrhoeae and murine PMNs (Fig. 4); this result is similar to observations regarding nonopsonic (Opa-mediated) uptake of gonococci by human PMNS as reported by Rest and Frangipane (45). This study also showed nonopsonized, sialylated gonococci induce a weaker phagocytic respiratory burst than that induced by nonsialylated gonococci when incubated with human PMNs. We therefore tested the effect of sialylation on the capacity of N. gonorrhoeae to induce an oxidative burst in murine PMNs under opsonizing conditions. We found that sialylated wild-type gonococci induced a markedly reduced luminol-dependent chemiluminescence response in mouse PMNs compared to that induced by lst mutant GP300 cultured in the same concentration of CMP-NANA; this phenotype was restored by genetic complementation, as shown by the luminol-dependent chemiluminescence response induced by strain GP322 cultured in CMP-NANA. By measuring the extracellular and intracellular CL responses as described in Materials and Methods, we determined that the majority of the ROS were produced intracellularly following incubation with any of the three strains. The level of extracellular ROS produced was virtually undetectable (Fig. 5).

FIG. 5.

Oxidative response of mouse PMNs to sialylated and nonsialylated gonococci. Isoluminol- or luminol-enhanced CL assays were utilized to measure the total (top panel), intracellular (middle panel), and extracellular (bottom panel) release of ROS by PMNs from estradiol-treated mice following exposure to wild-type MS11, lst mutant GP300, and complemented strain GP322, as described in Materials and Methods. Bacteria were cultured in CMP-NANA; IPTG was not used to maximally induce the complementing lst gene in GP322.

Role of complement-mediated bacteriolysis in attenuation of the lst mutant.

The ability to test the effect of LOS sialylation on resistance to mouse serum is technically challenging, since mouse serum does not kill N. gonorrhoeae in vitro (23) due to the hyperlability of terminal components of the mouse complement cascade (Sanjay Ram, personal communication). We therefore performed competitive infections with the wild-type strain and lst mutant in C5-deficient mice to test the hypothesis that LOS sialylation protects gonococci from complement-mediated bacteriolysis in vivo. We predicted that if evasion of complement-mediated killing contributed to the observed attenuation of the lst mutant in normal mice, the mutant would be less attenuated in mice that were unable to form the membrane attack complex. Contrary to this predicted outcome, however, lst mutant GP300 was attenuated in C5⁻ mice at a level that was comparable to the attenuation observed in the C5⁺ congenic strain (Fig. 6). The decrease in CI was similar to that observed with BALB/c mice, and the mutant was cleared in most mice by day 4 (Fig. 6, open symbols). We conclude this arm of the complement system is not responsible for the observed attenuation of the lst mutant in the mouse model. This result may be due to recently reported host restrictions in the complement activation pathway (37), as is discussed further below.

FIG. 6.

Recovery of the lst mutant (GP300) from the C5-deficient and C5-sufficient mice relative to the wild-type strain. C5⁻ results are expressed as competitive indices (CI) for individual C5⁻ (n = 4) and C5⁺ (n = 6) mice as described in the Fig. 2 legend. Horizontal bars represent the mean CI. The ratio of mutant to wild-type gonococci in the inoculum was 0.5. Cultures from which no GP300 gonococci were recovered are represented by open symbols.

Increased survival of sialylated gonococci following i.p. challenge.

The data presented thus far suggest that the survival disadvantage of the lst mutant during experimental genital tract infection occurs at the level of phagocytic killing. To further test this hypothesis, we compared the survival of the lst mutant following i.p. injection of mice with that of sialylated wild-type gonococci. This approach was used by others to study the resistance of Yersinia enterocolitica to phagocytes (19), and it has the advantage of bypassing any known or undescribed effects that sialylation may have on the interaction of gonococci with nonprofessional phagocytic cells in the genital mucosa (i.e., epithelial cells). For these experiments, wild-type MS11 and lst mutant GP300 bacteria were cultured in the presence of CMP-NANA and injected into separate groups of estradiol-treated mice; the survival of bacteria over time was assessed by quantitative culture of peritoneal lavage fluid. Both sialylated wild-type and GP300 gonococci elicited a rapid and large influx of PMNs into the peritoneal cavity, and a decrease in the number of viable gonococci recovered occurred over time for both strains. However, significantly fewer lst mutant gonococci were recovered at 4 and 5 h postinoculation than were recovered from wild-type bacteria (P < 0.0001 and 0.02, respectively) (Fig. 7). The outcome of these experiments is consistent with our hypothesis that the lst mutant is more susceptible to phagocytic killing in vivo. We did not compare sialylated versus nonsialylated MS11 gonococci in this assay, since sialylation of the bacteria might occur within the peritoneal cavity. Such an event would impede our ability to perform a clean comparison and might lead to a false conclusion as to the effect of sialylation on survival in this body site.

FIG. 7.

Survival of sialylated and nonsialylated gonococci following i.p. injection of estradiol-treated mice. Wild-type MS11 and lst mutant GP300 gonococci cultured in CMP-NANA were used in these assays. The mutant was significantly killed compared to results for the wild-type strain at 4 and 5 h (P < 0.0001 and P = 0.02, respectively).

DISCUSSION

N. gonorrhoeae is a highly successful pathogen that is exquisitely well-adapted to evade host innate defenses. Persistence of this organism amid a vigorous PMN inflammatory response is likely due to the influence of several factors which have been hypothesized to protect N. gonorrhoeae phagocytic killing based on in vitro studies. One or more antioxidant factors may protect against toxic oxygen species produced by phagocytes (26, 51, 53, 57, 58); phase variation of Opa protein expression may protect against nonopsonic uptake by PMNs, and as discussed previously, sialylation of LOS may reduce nonopsonic and opsonophagocytic uptake (45) and killing by PMNs (15, 28, 45). Considerable debate as to the role of LOS sialylation in infection has stemmed from the fact that functional advantages conferred by sialylation are dose dependent, with both CMP-NANA and complement levels limiting (15, 46, 61). Additionally, the degree to which O₂-dependent mechanisms of phagocytic killing challenge the gonococcus in vivo is not known. Studies designed to test the resistance of gonococci to PMN killing in vitro are also usually performed under conditions that may falsely represent the microenvironment of mucosal sites inhabited by this pathogen.

Here we utilized a female mouse model of gonococcal genital tract infection to show that the lst gene conferred a strong survival advantage in vivo, as evidenced by the rapid loss of the lst mutant compared to the parent strain within 2 days after intravaginal inoculation. Recovery of the mutant from mice was restored through genetic complementation, which rules out polar effects of the mutation or the influence of unrelated mutations that may have occurred during mutant construction that might affect growth or survival in vivo. The lst mutant was also more susceptible to killing following i.p. inoculation; this result points to increased susceptibility to phagocytic killing being responsible for reduced survival of the mutant rather than differences in the interaction of Lst-sufficient and -deficient gonococci with the genital mucosa. Our results are consistent with other bacterial sialyltransferases; specifically, sialyltransferase mutants of Haemophilus influenzae were unable to cause acute otitis media and were less infectious in a chinchilla model (4).

In contrast to the sialyltransferase of H. influenzae, gonococcal sialyltransferase was not required in strain MS11 for experimental murine infection in that no difference in duration of infection or colonization load was detected in mice inoculated with wild-type or lst mutant gonococci at doses 10-fold lower than that required to infect 80% of mice. Competitive infection, which minimizes variability between animals and is the basis of signature-tagged mutagenesis (54), is a proven strategy for detecting the importance of virulence factors in pathogens that utilize functionally redundant colonization or survival mechanisms. The fact that we observed attenuation of the gonococcal lst mutant by competitive infection but not in comparisons of animals infected with either wild-type or mutant gonococci could be explained by the numerous mechanisms by which the gonococcus can evade complement-mediated and phagocytic killing. Mutating one of several functionally redundant factors may not cause a detectable effect unless doses lower than the 50% infectious dose are used; the number of mice required to generate a statistically significant difference in such an experiment is difficult to justify, however, when the more sensitive technique of competitive infection can be used. We also must consider the possibility that perhaps the wild-type strain causes a more productive infection than the Lst-deficient strain, which would result in a greater host challenge to the mutant than that which occurs during infection by the mutant alone. We currently have no evidence to support this latter theory, however, since we saw no difference in the degree of PMN influx or the presence of gonococci within PMNs upon examination of stained smears from mice infected with wild-type or lst mutant gonococci (data not shown).

Another explanation for the reduced recovery of the lst mutant relative to the wild-type strain during competitive infection is that lack of sialyltransferase, which is associated with the outer membrane (49), may make gonococci less viable. Functional advantages conferred by sialylation were abolished via treatment with NANase in vitro, which is consistent with the addition of sialic acid being important for increased survival and not the physical presence of the enzyme. However, we cannot rule out that absence of this protein in the membrane might alter survival in vivo. We do not favor this hypothesis, however, since sialyltransferase is a minor membrane component (49), and we saw no difference in the in vitro growth characteristics of the mutant cultured with the wild-type strain, even in the stationary phase. Another Lst-associated phenotype that might explain our results is the possibility that nonsialylated gonococci may retreat into an intracellular niche via binding of nonsialylated LOS to the murine homologue of the human asialoglycoprotein receptor (ASGP-R), which is expressed on primary urethral cells from men (20, 43). In this case, mutant gonococci might be less accessible for culture. Whether or not the rodent ASGP-R molecule can function as an adherence receptor for N. gonorrhoeae is not known, and we have not yet detected intracellular bacteria in mouse tissue (W. Song and A. E. Jerse, unpublished data). We also believe this explanation to be unlikely, since no difference in the colonization load was detected in mice inoculated with either wild-type or Lst-deficient gonococci.

Our investigation of the effect of sialylation on gonococcal interactions with mouse PMNs revealed several similarities with what has been reported for human PMNs. We demonstrated here that sialylation (i) dramatically increased resistance to killing by murine PMNs, (ii) dramatically reduced bacterial association with murine PMNs, and (iii) led to a reduced oxidative burst, all of which is consistent with studies using human PMNs (15, 45). We also found that phagocytes from estradiol-treated mice undergo a respiratory burst in response to N. gonorrhoeae that is primarily intracellular, as was reported with human PMNs (35). Together, these data support the usefulness of the mouse model as an experimental system for examining interactions between N. gonorrhoeae and PMNs during mucosal infection. As with all experimental models, however, differences from natural infection can confound the interpretation of data. As discussed, Opa-mediated (nonopsonic) uptake does not occur with mouse PMNs. The relative contribution of the opsonic and nonopsonic pathways during infection is not known, however, and certainly in the presence of inflammation, uptake of bacteria via the opsonic pathway occurs. Perhaps more importantly, during preparation of this work, Ram and colleagues reported that the binding of C4bp (36) and fH (37) to N. gonorrhoeae is human specific. Down-regulation of the alternative pathway via the binding of fH to sialic acid residues on gonococcal LOS leads to reduced deposition of opsonins as well as reduced bacteriolysis due to the amplification loop that links the alternative and classical pathways (44). The fact that the lst mutant was also attenuated in C5-deficient mice is consistent with a host restriction for fH. The impact of the fH host restriction on opsonic interactions with phagocytes is likely to be less pronounced, since relatively few opsonin molecules are needed for effective opsonization compared to the number of membrane attack complexes required for lysis. In this regard, the mouse model probably underdetects the impact of sialylation during genital tract infection. We are currently developing mice that are transgenic for human complement regulatory proteins in hopes that they will provide a better background for studying factors, such as sialyltransferase, which are known to play a role in evasion of complement-mediated bacteriolysis.

In summary, our study demonstrates a protective role for N. gonorrhoeae sialyltransferase early during infection, when serum factors and PMNs are introduced as part of the inflammatory response. We present in vitro and in vivo evidence to support the hypothesis that sialylation of gonococcal LOS confers a survival advantage in experimentally infected mice by increasing resistance to opsonophagocytic killing. This advantage was detectable despite newly recognized host restrictions in the complement cascade. The development of transgenic mice that express human fH and C4bp should allow us to more fully test the role of sialylation in vivo; similarly, the testing of mutants in mice that are defective in phagocytic function may help define the relative importance of different gonococcal defense mechanisms as proposed from in vitro studies.