A pleiotropic role of sialidase in the pathogenicity of Porphyromonas gingivalis

ABSTRACT

As one of the keystone pathogens of periodontitis, the oral bacterium Porphyromonas gingivalis produces an array of virulence factors, including a recently identified sialidase (PG0352). Our previous report involving loss-of-function studies indicated that PG0352 plays an important role in the pathophysiology of P. gingivalis. However, this report had not been corroborated by gain-of-function studies or substantiated in different P. gingivalis strains. To fill these gaps, herein we first confirm the role of PG0352 in cell surface structures (e.g., capsule) and serum resistance using P. gingivalis W83 strain through genetic complementation and then recapitulate these studies using P. gingivalis ATCC33277 strain. We further investigate the role of PG0352 and its counterpart (PGN1608) in ATCC33277 in cell growth, biofilm formation, neutrophil killing, cell invasion, and P. gingivalis-induced inflammation. Our results indicate that PG0352 and PGN1608 are implicated in P. gingivalis cell surface structures, hydrophobicity, biofilm formation, resistance to complement and neutrophil killing, and host immune responses. Possible molecular mechanisms involved are also discussed. In summary, this report underscores the importance of sialidases in the pathophysiology of P. gingivalis and opens an avenue to elucidate their underlying molecular mechanisms.

INTRODUCTION

Periodontitis is a chronic inflammatory disease that afflicts the gums and surrounding tissue (1). In the severe state of the disease, unresolved inflammation leads to the destruction of the gingiva, ultimately resulting in alveolar bone loss (2, 3). Periodontitis is associated with other diseases, such as rheumatoid arthritis, Alzheimer’s disease, diabetes, and oral squamous cell carcinoma (OSCC) (1, 4–7). The development of periodontitis is driven by an intricate interplay between the host immune system and oral microbiota (1, 2). Among those identified periodontal pathogens, the oral bacterium Porphyromonas gingivalis is considered a “keystone” pathogen due to its unique ability to subvert and co-opt the host inflammatory response; for example, it deploys a myriad of virulence factors to antagonize and cause dysregulation of host immune responses, particularly the complement system and neutrophil homeostasis, the two pillars of the innate immunity (2, 8, 9). Antagonizing the immune response also exerts protective effects on other members of the subgingival polymicrobial community where P. gingivalis resides (9, 10). While numerous virulence factors have been characterized in P. gingivalis, there are still knowledge gaps that must be filled to advance our understanding of how it contributes to periodontitis and other systemic ailments.

In our previous report, we identified and functionally characterized a putative sialidase (PG0352) in the P. gingivalis W83 strain; our results indicated that PG0352 is a novel virulence factor that affects surface polysaccharide biosynthesis and/or assembly, biofilm formation, and serum resistance (11). In addition, in vivo study using a murine abscess model further revealed that PG0352 is essential for P. gingivalis to cause systemic infection in mice (11). And yet, the observed phenotypes had not been rigorously confirmed through gain-of-function studies, that is, the deletion mutant of PG0352 constructed in that report had not been genetically complemented to restore its phenotypes. In addition, P. gingivalis strains are diverse, being divided into six serotypes (i.e., K1 to K6) based on their capsules (K-antigens) (12, 13) or six genotypes (i.e., I, Ib, II, III, IV, V) based on the fimA gene sequence variations (14). To date, at least 14 different P. gingivalis strains have been sequenced, among which W83 and ATCC33277 (hereafter referred to as 33277) are the two most commonly used laboratory strains. The genomic comparison revealed that their genomes encode 461 genes specific to 33277 and 415 genes specific to W83 despite a high similarity in their genome sizes and overall GC contents (15, 16). In addition, these two strains differ in their surface appendages (12, 13, 17). For instance, 33277 is non-capsulated but possesses fimbriae, including both major (FimA) and minor (Mfa1) types. By contrast, W83 is encapsulated but has no fimbriae. These differences bestow the two strains with different pathogenic traits (11, 18–20). For example, compared to 33277, W83 forms fewer biofilms and is less invasive to gingival epithelial cells, but is more toxic to mice, for example, W83 can induce spreading necrotic lesions and sepsis, ultimately resulting in animal death, whereas 33277 only induces localized abscesses with no impact on animal mortality. Like W83, the genome of 33277 also encodes a sialidase (PGN1608) (16) whose function has not been investigated. To fill these knowledge gaps, here we first corroborated the role of PG0352 in P. gingivalis growth, surface polysaccharides, and serum resistance through genetic complementation and then investigated the role of PGN1608 in P. gingivalis biofilm formation, serum resistance, neutrophils killing, cell invasion, and inflammation.

RESULTS

Complementation restores the sialidase activity in a PG0352 deletion mutant

In our previous report, a sigma⁷⁰-like promoter (P_σ70) was mapped to the upstream region of PG0352 (11). To generate the complementation vector, pG108-PG0352 (Fig. 1A), P_σ70, and the full-length PG0352 gene were PCR amplified and cloned into the shuttle vector pG108 (21). The resulting vector was electroporated into Δ0352, a mutant previously constructed in W83 strain via in-frame replacement of PG0352 with the full-length ermF/AM cassette (11), and plated for selection of positive transformants with both clindamycin and tetracycline. Positive colonies were isolated, grown in liquid media, and subjected to PCR to screen for the presence of PG0352, ermF, and tetQ genes. Numerous positive clones were identified and one of which (CΔ0352) was selected and subjected to RT-PCR to detect expression of PG0352 (Fig. 1B) and to filter paper spot assays to determine sialidase activity (Fig. 1B). Recombinant PG0352 protein (rPG0352) was included in the sialidase activity assay as a positive control. As expected, the expression of PG0352 and sialidase activity were abolished in Δ0352 and restored in its isogenic complemented CΔ0352 strain (Fig. 1B). These results demonstrate that the trans-complementation successfully restores the expression and sialidase activity of PG0352 in Δ0352.

Fig 1.

Constructions and validations of P. gingivalis sialidase-deficient mutants and their isogenic complemented strains. (A) Complementation of Δ0352, a previously constructed PG0352 deletion mutant in W83, using pG108 shuttle vector. Primers P1/P2 were used to amplify PG0352 and its upstream promoter sequence (P); Primers P5/P6 were used to detect the tetQ cassette in pG108. (B) RT-PCR (top) and filter paper spot test (bottom) showed that the expression of PG0352 and sialidase activity were abolished in the Δ0352 mutant and restored in its isogenic complemented strain CΔ0352. (C) Schematic illustration of replacing PGN1608 with ermF in 33277 strain. PGN1608::ermF was constructed by two-step PCR with three pairs of primers (P9 to P14) as labeled, which was electroporated into 33277 to in-frame replace PGN1608 with ermF via DNA allelic exchange, generating the Δ1608 mutant. (D) RT-PCR and (E) filter paper spot test showed that the expression of PGN1608 and sialidase activity were abolished in Δ1608 and restored in its isogenic complemented strain CΔ1608 which was constructed using pG108 shuttle vector. Recombinant PG0352 (rPG0352) and substrate alone were used as positive and negative controls, respectively.

Deletion of PGN1608 abolishes the sialidase activity in the 33277 strain

W83 and 33277 are the two most commonly studied P. gingivalis strains, and both genomes were sequenced (16, 22). Like W83, the genome of 33277 also encodes a sialidase (PGN1608) that shares 97% sequence identity with PG0352. To determine whether PGN1608 plays a similar role to PG0352, we constructed a deletion mutant of PGN1608 (Δ1608) using ermF, a short version of the ermF/AM cassette that still allows for selection using clindamycin (Fig. 1C), and its isogenic complemented strain CΔ1608 using the shuttle vector pG108. RT-PCR analysis showed that expression of PGN1608 was abolished in Δ1608 and restored in CΔ1608 (Fig. 1D and E). Consistent with the RT-PCR result, filter spot paper assays showed that sialidase activity was abolished in the mutant and successfully restored in the complemented strain (Fig. 1D).

PG0352 and PGN1608 are not required for P. gingivalis planktonic growth

Bacteria often use sialidases to scavenge host sialic acids as carbon and nitrogen sources for growth (23, 24). To determine whether this is the case in P. gingivalis, we measured the growth rates of W83, Δ0352, and CΔ0352 under three conditions. We first measured their growth in TSB, a normal growth medium for P. gingivalis and found all three strains had a similar growth rate (Fig. 2A). We further substantiated this result using a chemically defined medium (CDM) supplemented with or without N-acetylneuraminic acid (Neu5Ac, 30 µg/mL), the predominant sialic acid found in humans (24), and found that neither deletion of PG0352 nor the addition of Neu5Ac had impact on P. gingivalis growth (Fig. 2B). Interestingly, CΔ0352 grew slightly slower than W83 and Δ0352 in CDM, possibly due to the presence of pG108, an ectopic plasmid (21). Moradali et al. reported that P. gingivalis exhibits metabolic plasticity in the subgingival microenvironment through subsisting on human serum, which contains a high level of sialic acids (25). The subgingival sulcus is constantly bathed in gingival crevicular fluid (GCF), a serum-rich inflammatory exudate that contains various sialoglyco-conjugates (26, 27). To mimic this growth condition, we grew P. gingivalis in 50% heat-inactivated rabbit serum (HI-RS), as previously reported (28) and found that all three P. gingivalis strains grew at similar rates (Fig. 2C). We also monitored the growth rates of 33277, Δ1608, and CΔ1608 in TSB medium, CDM with or without Neu5Ac, and HI-RS. Similar to W83, we found that these strains had similar growth rates under all three conditions (Fig. 2D through F). Collectively, these results indicate that sialic acids and sialidases are not required for the planktonic growth of P. gingivalis.

Fig 2.

Deletion of PG0352 and PGN1608 has no impact on P. gingivalis planktonic growth. (A–C) Measuring the growth rates of W83, Δ0352, and CΔ0352 in TSB medium, a chemically defined medium (CDM) with or without Neu5Ac (30 µg/mL), and 50% heat-inactivated rabbit serum (HI-RS). (D–F) Measuring the growth rates of 33277, Δ1608, and CΔ1608 in TSB, CDM, and HI-RS. P. gingivalis growth rates were recorded as the averages of three independent experiments ± standard errors of means (SEM). Statistical analysis was assessed using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05.

Deletion of PGN1608 attenuates P. gingivalis biofilm formation

In addition to planktonic growth, P. gingivalis can form biofilms (20). Our previous study revealed that the deletion of PG0352 impaired biofilm formation in P. gingivalis (11). To determine whether PGN1608 is also implicated in biofilm formation, we first measured the biofilm formation of 33277, Δ1608, and CΔ1608 in static 96-well microtiter plates (20). Compared to 33277 and CΔ1608 strains, the Δ1608 mutant formed less biofilm (Fig. 3A and B). To confirm this result, we microscopically examined P. gingivalis biofilms grown on coverslips prepared in static 12-well microtiter plates (20). Microscopic examination showed that both 33277 and CΔ1608 developed robust biofilms evident by formation of dense microcolonies on the coverslips. By contrast, Δ1608 poorly formed biofilms. Representative top-down views and x-z planes (i.e., horizontal side views) of biofilms formed by the three P. gingivalis strains are shown in Fig. 3C through E. Collectively, these results indicate that sialidases contribute to P. gingivalis sessile growth and biofilm formation.

Fig 3.

Deletion of PGN1608 impairs P. gingivalis biofilm formation. (A) A representative image showing P. gingivalis biofilms in static 96-well microtiter plates. This assay was performed as previously described (11). (B) Quantification of biofilms formed by 33277, Δ1608, and CΔ1608 via measuring the absorbance at OD₅₇₀. Statistical significance was assessed using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05. ***<I>P < 0.001. (C–E) Microscopic examination of biofilms formed by 33277, Δ1608, and CΔ1608. For this study, P. gingivalis biofilms were grown on sterile glass coverslips in 12-well microtiter plates for 3 days at 37°C under anaerobic conditions, washed with PBS, fixed with 4% paraformaldehyde, and finally stained with DAPI (29). Images were obtained with KEYENCE BZ-X Series All-in-One fluorescence microscope. Top panels: top-down views of the biofilms; lower panels: x–z plane side views of the biofilms.

Deletion of PG0352 and PGN1608 alters P. gingivalis cell surface structures

W83 is encapsulated and our previous study revealed that the deletion of PG0352 caused defects in the outer surface layer, presumably to be polysaccharides in nature (i.e., capsule polysaccharides) (11). To establish that this defect is due to the deletion of PG0352 rather than off-target effects (e.g., spontaneous mutations), we examined our newly constructed CΔ0352 strain using cryo-electron tomography (cryo-ET) and compared its surface structure to that of W83 and Δ0352. Like our previous report, cryo-ET analysis showed that ∆0352 has a defective layer of surface polysaccharides (SP), which was restored to the wild-type level in C∆0352 complemented strain (Fig. 4A through C). We measured the thickness of SP and found that the SP layer in ∆0352 (16.86 nm, n = 10 cells) is ~2-fold thinner than that in W83 (31.89 nm, n = 10 cells) and CΔ0352 (30.75 nm, n = 10 cells) (Fig. 4G). Unlike W83, 33277 is non-capsulated but produces fimbriae. As expected, hair-like fimbriae were visualized by cryo-ET at the cell surface of 33277, ∆1608, and C∆1608. In addition to fimbriae, cryo-ET also detected an extra electron density surface layer (EDSL) at the surface of 33277 cells (Fig. 4D through F), which is very similar to what was reported by Song et al. (30). The EDSL in ∆1608 was notably thinner and less dense compared to that in 33277 and was restored in the complemented strain C∆1608 (Fig. 4D through F). Using nano gold particles as a scale, we measured the diameters of EDSL in the three P. gingivalis strains and found that the average thickness of EDSL in Δ1608 (21.75 nm, n = 10 cells) was significantly (P < 0.05) less than that of 33277 (33.63 nm, n = 10 cells) and CΔ1608 (27.35 nm, n = 10 cells). P. gingivalis strains such as 33277 and W83 possess a type IX secretion system (T9SS), also known as the Por secretion system (PorSS) (30–32). Song et al. reported that T9SS-deficient mutants, such as ΔporK and ΔporN, do not produce EDSL (30). Therefore, it is possible that deletion of PGN1608 affects the assembly of T9SS or production of surface proteins secreted through T9SS, in turn altering P. gingivalis surface properties as revealed by cryo-ET. Taken together, these results suggest that PG0352 and PGN1608 are involved in production and/or modifications of P. gingivalis surface macromolecules (e.g., surface polysaccharides).

Fig 4.

Cryo-ET analysis revealed that the deletion of PG0352 and PGN1608 alters the surface structures of P. gingivalis cells. Top panel: W83 (A), ∆0352 (B), and C∆0352 (C); middle panel: 33277 (D), ∆1608 (E), and C∆1608 (F); and bottom panel (G, H): measuring the thickness of capsule polysaccharide or an electron-dense surface layer. ***P < 0.001. The prominent structural features include the following: the outer membrane (OM), inner membrane (IM), peptidoglycan layer (PG), fimbriae, capsular polysaccharide (CPS), and electron-dense surface layer (EDSL). The black dots are nano golden particles (diameter, 10 nm) which were used to calibrate the measurements.

Loss of PG0352 and PGN1608 has no impact on P. gingivalis anionic polysaccharide production

P. gingivalis strains produce various surface polysaccharides such as lipopolysaccharide (LPS) and capsule polysaccharides. In addition, some P. gingivalis isolates produce a unique anionic polysaccharide (APS) that is composed of a phosphorylated branched mannan (33). APS is also referred to as A-LPS because it is attached to lipid A (34). Interestingly, APS shares a common epitope with posttranslational additions (presumably glycans) to Arg-gingipains (e.g., RgpA) that can be recognized by MAb1B5, a monoclonal antibody raised against RgpA (33, 35, 36). APS contributes to P. gingivalis serum resistance and EDSL, for example, an APS-deficient mutant was found to be susceptible to serum killing and had a defective EDSL (34), which is reminiscent of what we observed in ∆0352 (11) and Δ1608 (Fig. 4). Therefore, we speculated that PG0352 and PGN1608 may contribute to APS production and/or modifications. To determine whether this is the case, we examined our mutants for their cross-reactivity to MAb1B5 using immunoblots as previously described (33, 34). We found that both ∆0352 and Δ1608 mutants exhibited similar cross-reactivity to their parental strains and isogenic complemented strains (Fig. 5A). This result was further substantiated by silver staining (Fig. 5B). Based on these results, we conclude that PG0352 and PGN1608 have no impact on APS. However, these experiments cannot rule out the possibility that APS is modified by sialic acids but cannot be recognized by MAb1B5 or visualized by silver staining.

Fig 5.

Measuring P. gingivalis cell surface hydrophobicity. This assay was performed using n-hexadecane as previously described with some modifications (38, 39). In brief, 200, 400, or 800 µL of n-hexadecane was added into 3.0 mL of P. gingivalis cell suspensions, vortexed for 60 seconds, and then set aside for 15 minutes to allow the layers to partition. Samples of each strain were taken before and after the addition of n-hexadecane and subjected to the Varioskan LUX multimode plate reader for measuring absorbance at OD₅₅₀. These assays were performed three times in triplicate; the final readouts were expressed as means ± standard errors of means (SEM). The percentage of bacteria adhering to n-hexadecane was calculated using the formula described in the Materials and Methods. (A) W83, ∆0352, and C∆0352; (B) 33277, 1608, and C∆1608. Statistical analysis was performed using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05. ***P < 0.001, **P < 0.01, *P < 0.05, ns, no significant (P > 0.05).

Assessing the impact of PG0352 and PGN1608 on P. gingivalis cell surface hydrophobicity

A previous report showed that loss of surface structures (e.g., polysaccharides) reduced P. gingivalis cell surface hydrophobicity (CSH) (37). In addition, P. gingivalis strains exhibiting high hydrophobic surfaces form more robust biofilms (19). Our results indicate that loss of PG0352/PGN1608 has a significant impact on P. gingivalis biofilm formation (Fig. 3) and cell surface structures (Fig. 4). Therefore, we reasoned that loss of PG0352/PGN1608 may alter P. gingivalis CSH. To test this hypothesis, we measured P. gingivalis CSH using n-hexadecane as previously described (38, 39). As shown in Fig. 6, the capacity of ∆0352 and Δ1608 partitioning into the n-hexadecane layer was significantly attenuated compared to that of their parental strains and two isogenic complemented strains. Interestingly, such an impact appeared to be more evident in Δ1608. For example, with the addition of 200 µL n-hexadecane, while there was no significant difference between ∆0352 and W83, the capacity of Δ1608 adhering to n-hexadecane was decreased nearly 10-fold relative to that of 33277 and its isogenic complemented strain (Fig. 6A). These results suggest that sialidases play a role on modulating P. gingivalis cell surface hydrophobicity and such a role may vary among different P. gingivalis strains.

Fig 6.

Assessing the role of PG0352 and PGN1608 on P. gingivalis polysaccharides. (A) Immunoblotting analysis of proteinase K-treated P. gingivalis whole cell lysates probed against Mab1B5 (1:100), a mouse monoclonal antibody that recognizes the epitope of P. gingivalis APS/A-LPS (33, 34). (B) SDS-PAGE analysis of proteinase K-treated P. gingivalis whole cell lysates followed by silver staining. P. gingivalis W50 was used as a positive control.

Both PG0352 and PGN1608 contribute to P. gingivalis serum resistance

P. gingivalis is resistant to complement killing (also known as serum killing) (9, 10, 35). Our previous study indicated that PG0352 contributes to P. gingivalis serum resistance (11). To substantiate this study, we first repeated serum killing assays by including our newly constructed C∆0352 complemented strain (Fig. 7A) and found that W83 was resistant to serum killing, with ~91% survival rates after 1 hour of incubation with 25% human serum, whereas ∆0352 became susceptible to serum killing, with a survival rate of 29.67%. The serum resistance was restored in C∆0352, with a survival rate of ~90%. We then performed serum killing assays using 33277, ∆1608, and C∆1608 (Fig. 7B) and observed a trend similar to what we found in W83 and its isogenic mutants. Taken together, these results indicate that both PG0352 and PGN1608 contribute to P. gingivalis resistance to serum killing but are not absolutely required because the mutants still retain some degree of resistance to serum killing.

Fig 7.

Serum killing assays. (A) W83 and its two derivative mutants and (B) 33277 and its two derivative mutants were grown to mid-log phase and then incubated with 25% human serum or 25% heat-inactivated serum anaerobically for 1 hour at 37°C. Following the incubation, the samples were serially diluted and plated. Colonies were enumerated after 6 days. Survival rates were recorded as follows: the total number of colonies in the samples treated with human serum divided by the number of colonies in the samples treated with heat-inactivated serum. Statistical analysis was determined using one-way ANOVA followed by Tukey’s multiple comparison, P < 0.05. ***P < 0.001, **P < 0.01, *P < 0.05.

PG0352 and PGN1608 protect P. gingivalis from neutrophil killing

Neutrophils (also known as polymorphonuclear leukocytes, PMNs) are the predominant innate immune cells in the oral cavity (8, 40). As the keystone pathogen of periodontitis, P. gingivalis has evolved complex mechanisms to subvert neutrophil functions, for example, impairment of neutrophil recruitment, chemotaxis, and activation (8, 9). However, bacterial factors underpinning these functions remain largely unknown. Surface polysaccharides such as capsules often protect bacteria from neutrophil phagocytosis (41, 42). Since the deletion mutant of PG0352 and PGN1608 has a defective layer of surface polysaccharides (Fig. 4), we speculated that this defect may impact P. gingivalis resistance to neutrophil killing. To test this speculation, we conducted neutrophil killing assays using HL-60 cells, which can be differentiated into neutrophil-like cells with DMSO. For this experiment, P. gingivalis strains, including W83, Δ0352, and CΔ0352, were incubated with PMA-activated HL-60 cells for 30 min at an MOI of 1. Under this condition, W83 and CΔ0352 had survival rates of 28% and 25%, respectively, whereas Δ0352 had a survival rate below 10% (Fig. 8A). We also performed neutrophil killing assays using 33277, Δ1608, and CΔ1608 and found a similar pattern (Fig. 8B): the survival rate of Δ1608 (~14.33%) was significantly lower than that of 33277 (~29.09%) and CΔ1608 (26.75%).

Fig 8.

Deletion of PG0352 and PGN1608 compromises P. gingivalis’s ability to resist neutrophil killing. (A) and (B) Neutrophil killing assays. For this study, DMSO-differentiated HL60 cells were first stimulated with PMA for 30 minutes and then incubated with six different P. gingivalis strains (W83, ∆0352, C∆0352, 33277, ∆1608, and C∆1608) at an MOI of 1 for 30 minutes. Following the incubation, treated HL-60 cells were lysed, serially diluted, and then plated on blood agar plates for measuring CFU. P. gingivalis survival rates were recorded as follows: the CFU in PMA-HL60 divided by that in non-PMA stimulated HL-60. (C) Measuring the intracellular survival rates of P. gingivalis using antibiotic protection assays. For this experiment, PMA-stimulated HL-60 cells were incubated with six different P. gingivalis strains at an MOI of 10 for 1 hour and then treated with metronidazole and gentamicin to kill extracellular bacteria. The resulting samples were washed, lysed, serially diluted, and then plated on blood agar plates for measuring CFU. Survival rates were calculated as follows: CFU in PMA-HL60 divided by that in non-PMA stimulated HL-60. Statistical analysis was determined using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05. ***P < 0.001, **P < 0.01, *P < 0.05.

Neutrophils eliminate pathogens through dedicated mechanisms, including phagocytosis, antimicrobial peptides, and release of neutrophil extracellular traps (NETs) (40, 43, 44). The above experiment measures the total survival rates of P. gingivalis in neutrophils. In addition to this study, we also explicitly measured the intracellular survival rate of P. gingivalis in HL-60 cells. For this experiment, we used an MOI of 10 and increased coincubation time from 30 minutes to 1 hour (45). After incubation, the remaining extracellular P. gingivalis cells were eliminated by treatment with metronidazole and gentamicin. Under this condition, we found that the mean intracellular survival rate of W83 and CΔ0352 was approximately 7%–11%, which decreased to 0.6% in Δ0352 (Fig. 8C). These results demonstrate that both PG0352 and PGN1608 are implicated in P. gingivalis resistance to neutrophil killing and phagocytosis.

PG0352 and PGN1608 have no evident role on P. gingivalis invasion

Sialidases can reveal hidden epitopes via desialylation and thus promote bacterial invasion (23, 46). For example, the NanA sialidase of Streptococcus pneumoniae promotes invasion into brain endothelial cells and is critical in causing meningitis (47). To determine whether a similar scenario occurs in P. gingivalis, we measured the impact of PG0352 on P. gingivalis invasion using telomerase-immortalized gingival keratinocytes (TIGKs), as previously described (48, 49). Deletion of PG0352 slightly increased P. gingivalis invasion but the difference between W83 and Δ0352 was not statistically significant (P > 0.05) (Fig. 9A). The same experiment was performed using 33277, Δ1608, and CΔ1608 and the result showed that the three strains had similar invasion rates (Fig. 9B). Collectively, these results indicate that PG0352 and PGN1608 are not implicated in P. gingivalis cell invasion, at least under the examined condition.

Fig 9.

Assessing the impact of PG0352 and PGN1608 on P. gingivalis invasion. A standard antibiotic protection assay was carried out to measure P. gingivalis invasion to TIGK cells, as previously described (48, 49). TIGKs were infected with six different P. gingivalis strains (W83, ∆0352, C∆0352, 33277, ∆1608, and C∆1608) at an MOI of 10. Following the infection and antibiotics treatment, the resulting TIGKs were lysed, serially diluted, and plated for measuring CFU. The percentage of invasion was determined as follows: CFU recovered after antibiotic treatment divided by CFU of total P. gingivalis cells used. Statistical analysis was determined using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05. N.S, not significant (P > 0.05).

Deletion of PG0352 and PGN1608 alters P. gingivalis-induced proinflammatory cytokines

A previous report using Epi4, an SV40 T-antigen-immortalized gingival epithelial cell line, revealed that deletion of PG0352 reduces the production of pro-inflammatory cytokines (50). To corroborate this report, we measured and compared cytokine production among W83, 33277, and their isogenic sialidase-deficient mutants when exposed to human monocytes. For this study, freshly isolated human peripheral blood monocytes were challenged with P. gingivalis cells at an MOI of 10 for 8 hours. Cell-free supernatants were then collected and subjected to ELISA to measure specific proinflammatory cytokines, including TNF-α, IL-1β, IL-6, and IL-12 (Fig. 10). As in the previous report, we found that the proinflammatory cytokines induced by W83 and 33277 were significantly (P < 0.05) higher than those induced by Δ0352 and Δ1608. Interestingly, the level of IL-12 induced by W83 is much higher than that induced by 33277. We repeated this experiment three times and observed a similar pattern. These results suggest that PG0352 and PGN1608 are implicated, either directly or indirectly, in P. gingivalis-induced inflammation.

Fig 10.

Assessing the role of PG0352 and PGN1608 on P. gingivalis-induced proinflammatory cytokine production. For this study, freshly isolated human monocytes were challenged with different P. gingivalis strains as labeled; the resulting samples were subjected to ELISAs for measuring TNF-α (A), IL-1β (B), IL-6 (C), and IL-12 p40 (D). Unchallenged monocytes were used as a control. ELISAs were performed in triplicate and recorded as the averages of each experiment. Statistical analysis was performed by Student’s t-test at P < 0.05. **P < 0.01, *P < 0.05.

DISCUSSION

We initially attempted to complement the Δ0352 mutant using pTCOW, a widely used shuttle vector for P. gingivalis (51). However, these attempts were not successful. As an alternative approach, we used pG108, a shuttle vector derived from a pYHBA1 plasmid and subsequently modified with tetracycline (tetQ) and erythromycin resistance (ermB) markers for selection in P. gingivalis and Escherichia coli, respectively (21). Using this vector, we successfully complemented Δ0352 and restored expression of PG0352 and sialidase activity (Fig. 1). We also examined the expression of PG0352 by immunoblotting with a specific antibody against PG0352 (αPG0352) which was raised in rats using recombinant PG0352 proteins (rPG0352) purified in E. coli on a fee-for-service base (52). Interestingly,αPG0352 strongly reacted to rPG0352 but somehow failed to detect its native form in P. gingivalis whole cell lysates. We tried to overcome this issue by preparing P. gingivalis whole cell lysates under different conditions, for example, harvesting P. gingivalis cells at different growth phases and adding protease inhibitors to the whole cell lysates to inhibit gingipain activity (53). We also tried to detect PG0352 in the supernatant of P. gingivalis cultures using immunoblotting. Unfortunately, these attempts were unsuccessful. The N-terminus of PG0352 contains a putative lipoprotein signal peptide. Thus, it is possible that PG0352 is lipidated which, in turn, prevents the antibody from recognizing this protein in P. gingivalis cells. We are currently planning to raise PG0352 antibody using recombinant proteins prepared from P. gingivalis. If successful, this antibody will allow us to determine whether PG0352 is secreted and/or exposed on the P. gingivalis cell surface using immunoblotting and immunofluorescence microscopy.

Our previous study revealed that the deletion of PG0352 or the addition of exogenous Neu5Ac has no impact on the planktonic growth of P. gingivalis cells (11), but this study was performed only in TSB, a rich growth medium that contains sialic acids in the form of Neu5Ac, Neu5Gc, and 9-O-acetyl-n-acetylneuraminic acid (54). To consolidate our previous report, we repeated this experiment using a chemically defined medium (CDM) (55). We confirmed that this medium contains no detectable sialic acids using thiobarbituric acid assay (56, 57). Using this medium, we reassessed the impact of sialic acids and PG0352 on P. gingivalis growth and found that both WT and Δ0352 have the same growth rates, further supporting that sialic acids and sialidases are not required for P. gingivalis planktonic growth, which is well in agreement with its asaccharolytic growth nature (58). Moradali et al. demonstrated that P. gingivalis exhibits a degree of metabolic plasticity as an adaptation to limited nutrients in the subgingival sulcus using various human serum proteins for its growth (25). Most serum proteins are glycoproteins with varying amounts of sialic acids (24, 59). Thus, it is possible that P. gingivalis employs sialidases to free sialic acids from those serum glycoproteins for its growth in vivo. To rule out this possibility, we grew P. gingivalis strains in 50% heat-inactivated rabbit serum, which mimics GCF serum-rich conditions, and found that deletion of PG0352 and PGN1608 has no impact on P. gingivalis growth. Taken together, we conclude that sialic acids and sialidases are not required for P. gingivalis in vitro planktonic growth.

In our previous report, cryo-ET analysis revealed that the Δ0352 mutant has a defective layer of surface polysaccharides (11). However, it remained possible that this defect was caused by other factors, for example, polar effects and/or spontaneous mutations induced by genetic manipulations. To rule out this possibility, we repeated cryo-ET analysis with CΔ0352, an isogenic complemented strain of Δ0352. As in our previous report, Δ0352 had a defective layer of surface polysaccharide that was fully restored in the complemented strain (Fig. 4). Collectively, both loss-of-function and gain-of-function studies demonstrate that PG0352 is implicated in P. gingivalis surface polysaccharide biosynthesis and/or assembly though its underpinning molecular mechanism remains elusive. P. gingivalis strains are diverse in capsule type and structure and can be divided into six capsule serotypes (K1 to K6) (60, 61). W83 strain belongs to K1. Some bacterial pathogens, for example, pathogenic E. coli K1 strain (62), synthesize polysialic acid (PSA) capsules. Thus, W83 may employ PG0352 to scavenge host sialic acids for the synthesis of PSA-like capsules.

Unlike W83, 33277 has no capsule (12, 15); however, Chen et al. demonstrated that this strain produces an electron-dense surface layer (EDSL) at its cell surface (63), which is very similar to what we observed here (Fig. 4). The chemical nature and biological role of EDSL remain elusive. Some P. gingivalis strains, such as W50 and 33277, produce APS/A-LPS which contributes to EDSL, for example, APS/A-LPS-deficient mutants often have a defective EDSL (34, 63, 64). Using MAb1B5, a monoclonal antibody that cross-reacts to both RgpA and APS/A-LPS (34, 65), we demonstrated that deletion of PG0352 and PGN1608 has no evident impact on APS/A-LPS (Fig. 5). Chen et al. proposed that EDSL may be composed of surface proteins secreted through T9SS (63). Some of these surface proteins are glycosylated and modified by sialic acids (32). For example, RgpA, a gingipain secreted via T9SS, contains a high level of sialic acid that was attenuated in the sialidase-deficient mutant of P. gingivalis (36, 66). Thus, we reasoned that loss of PGN1608 may impair P. gingivalis acquisition of sialic acids and ability to expose their underlying sugar moieties for glycosylation, which, in turn, impacts anchoring of surface proteins such as gingipains to P. gingivalis cell surfaces and, thus, reduces the thickness of EDSL, as we observed in the Δ1608 mutant. In addition to surface proteins and APS/A-LPS, P. gingivalis also produces other surface macromolecules such as LPS and sphingolipids (15, 67). Zaric et al. showed that P. gingivalis LPS are sialylated; Moye et al. reported that the synthesis of sphingolipids impacts the presentation of surface polysaccharides (67, 68). Thus, it is also possible that deletion of PGN1608 affects LPS and lipids, in turn altering the surface properties of P. gingivalis as revealed by cryo-ET. In addition, our cell surface hydrophobicity assays introduced another avenue by which sialidase impacts the surface of P. gingivalis. Deletion of both PG0352 and PGN1608 resulted in a profound decrease in cell surface hydrophobicity in both W83 and 33277 despite their distinct different cell surface macromolecules (e.g., capsule and fimbriae). Davey et al. reported that P. gingivalis strains with increased cell surface hydrophobicity have enhanced biofilm formation (19). Our results are in accordance with this report as our sialidase mutants with more hydrophilic surfaces had poor biofilm formation when compared to their parental and isogenic complemented strains.

P. gingivalis is resistant to complement killing (35). In this report, we found that deletion of PG0352 and PGN1608 reduces its resistance to serum killing. Although the exact mechanism underlying this phenotype remains unknown, we speculated that the following factors may contribute to the role of sialidases in serum resistance. First, PG0352 and PGN1608 may render P. gingivalis serum resistant by modifying surface molecules such as surface polysaccharides (e.g., LPS) with sialic acids; for example, Neisseria gonorrhoeae evades complement killing by modifying its surface lipooligosaccharides (LOS) with sialic acid to capture host factor H (fH), a negative regulator of the complement system (54, 69). A similar scenario may exist in P. gingivalis given that its LPS are sialylated. Second, P. gingivalis serum resistance has been linked to gingipains (9, 27, 35). Aruni et al. reported that deletion of PG0352 diminishes gingipain activity (70). Thus, it is also possible that PG0352 and PGN1608 contribute to serum resistance by controlling the activity of gingipains. Lastly, the complement system harbors over 30 proteins, including plasma proteins, regulatory factors, receptors, and ligands, some of which are modified by sialoglycans that contribute to protein function, stability, protein-protein interactions, and self-recognition (59, 71). For instance, sialic acids are believed to minimize non-specific interactions of the C1 complex by stabilizing C1q (59). Desialylation of fH can alter its function and lead to pathogenic effects (72). In line with these reports, we recently found that PG0352 removes sialic acids from human serum and several complement factors such as C1q, C4, and fH and thus likely protects P. gingivalis from serum killing by disarming key complement factors (e.g., C1q and C4) via desialylation (73).

Most P. gingivalis isolates are resistant to neutrophil killing through different mechanisms (8, 74, 75). For instance, SerB, a serine phosphatase secreted by P. gingivalis, suppresses IL-8 production that would otherwise stimulate neutrophil migration (8, 74). Gingipains can degrade MyD88, a TLR2 signaling adapter protein required for clearance of P. gingivalis. Degradation of MyD88 triggers an alternative TLR2-PI3K signaling pathway that, in turn, inhibits neutrophil phagocytosis (9). P. gingivalis also produces ruberythrin (Rbr) and superoxide dismutase (Sod), which confer its resistance to neutrophils via nullifying reactive oxygen species (76). In this report, we found that both PG0352 and PGN1608 are implicated in P. gingivalis resistance to neutrophil killing. It is well known that surface polysaccharides such as capsules protect bacterial pathogens from phagocytosis of neutrophils and macrophages (77, 78). The deletion mutant of PG0352 has a defective layer of surface polysaccharides; thus, PG0352 may impair neutrophil function through surface polysaccharides. Several complement factors (e.g., C1q, C4, and C5) and the receptors involved in neutrophil chemotaxis and activation are sialylated (59). Our recent report showed that PG0352 can disarm key complement factors via desialylation (73). Since the complement system and neutrophils are interconnected, PG0352 and PGN1608 may modulate neutrophil functions by desialylating those sialylated complement factors. There are also reports that sialic acids and sialidases affect neutrophil activation via Siglec (sialic acid-binding immunoglobulin-type of lectins) receptors and chemotaxis via E-selectin (46, 79). For example, Pseudomonas aeruginosa uses sialic acids to modify its cell surface to inhibit neutrophil activation via Siglec-9 (80). During infection, neutrophils use rolling adhesion to interact with the endothelial cell adhesion molecule E-selectin via a sialylated Lewis X Ag (sLe^x) (81). Treatment of neutrophils with sialidases can abolish neutrophil binding to these receptors and thus inhibit neutrophil chemotaxis induced by fMLP (N-formyl-methionyl-leucyl-phenylalanine) (82). We are currently exploring these possible mechanisms.

As a keystone pathogen of periodontitis, P. gingivalis produces an array of virulence factors that induce inflammation and disrupt host immune homeostasis (1, 2). Previous studies have shown that P. gingivalis induces the production of pro-inflammatory cytokines from oral epithelial cells, gingival fibroblast cells, and macrophages (83, 84). These pro-inflammatory cytokines, such as TNF-α, IL-1β, IL-6, and IL-12, have been well characterized for their major contributions to inducing bone resorption and tissue damage (85–87). For example, TNF-α induces osteoclast differentiation mediated by RANK ligand stimulation, and IL-1β can synergize with TNF-α to induce bone resorption (85). Not surprisingly, the major virulence factors of P. gingivalis, such as gingipains, LPS, and fimbriae, can induce the production of these pro-inflammatory cytokines (83, 84, 88, 89). Their concerted activities can stimulate host inflammatory responses to P. gingivalis infection and disarm host immune systems to foster its survival. To determine whether sialidases contribute to P. gingivalis-induced inflammation, we measured the level of those pro-inflammatory cytokines using freshly isolated human monocytes and found that Δ0352 and Δ1608 stimulated much less cytokine production than their parental WT strains (Fig. 10). At this time, we can only speculate on the molecular mechanism involved. Most TLR receptors are glycosylated and some of which are sialylated (90–92). For instance, TLR4 is modified by nine N-linked glycans, some of which contain terminal sialic acids (91, 92). Removal of terminal sialic acids with bacterial sialidases can stimulate TLR4-mediated immune responses to LPS such as production of pro-inflammatory cytokines (92, 93). A similar scenario may stand for P. gingivalis, for example, PG0352 and PGN1608 may enhance P. gingivalis-induced pro-inflammation via desialylation of TLR4. Thus, the absence of these two enzymes might have an opposite effect on P. gingivalis-induced pro-inflammation, which is in line with what we observed in Fig. 10. However, other mechanisms may also exist, which require further investigation. P. gingivalis fimbriae (i.e., Mfa1) can stimulate cytokine (e.g., IL-1β, IL-6, and TNF-α) production in murine macrophages (84). As Mfa1 is glycosylated (94), we speculated that the loss of PGN1608 somehow impacts Mfa1 glycosylation and its ability to stimulate cytokine production. Likewise, the RgpA gingipain contains a significant level of sialic acid (36). Loss of PG0352 or PGN1608 may influence gingipain activity, which, in turn, alters the production of pro-inflammatory cytokines. P. gingivalis LPS is a major agonist to both TLR2 and TLR4 and stimulate the production of pro-inflammatory cytokines (89). Interestingly, the LPS isolated from both W83 and 33277 contains sialic acids, with W83 containing higher concentrations (68). Therefore, it is also conceivable that sialidases affect P. gingivalis-induced pro-inflammation via LPS sialylation.

MATERIALS AND METHODS

Bacterial strains and growth conditions

P. gingivalis W83 (22) and ATCC33277 (16) (hereafter designated as 33277), two wild-type (WT) strains, and their derived isogenic mutants were maintained on trypticase soy agar (TSA) plates supplemented with yeast extract (5.0 µg/mL), hemin (5.0 µg/mL), L-cysteine hydrochloride (0.5 mg/mL), vitamin K (1.0 µg/ml), and 5% defibrinated sheep’s blood (Colorado Serum Company, CO), as previously described (11). For liquid growth, single P. gingivalis colonies were inoculated into trypticase soy broth (TSB) supplemented with yeast extract (5.0 µg/ mL), hemin (5.0 µg/ml), L-cysteine hydrochloride (0.5 mg/mL), vitamin K (1.0 µg/ml), and appropriate antibiotics, including clindamycin (1.0 µg/mL) and tetracycline (1.0 µg/mL). P. gingivalis strains were grown in an AS-500 anaerobic chamber (Anaerobic System, CA) at 37°C with an atmosphere of 90% N₂, 5% H₂, and 5% CO₂, as previously described (11). E. coli NEB5α strain was used for DNA cloning; BL21-Star (DE3) was used for the preparation of recombinant proteins. These E. coli strains were grown in Luria-Bertani (LB) medium supplemented with appropriate antibiotics, including ampicillin (100 µg/mL), kanamycin (50 µg/mL), and erythromycin (300 µg/mL for plates; 50 µg/mL for liquid culture).

Construction of a PG0352 complemented strain

A DNA fragment containing the full-length PG0352 gene and its upstream promoter region was PCR amplified from W83 genomic DNA using P1/P2 primers (Table 1). The resulting ~2.0 kb amplicon was cloned into the pJET1.2 cloning vector (Thermo Fisher, IL) and verified by DNA sequencing (Genewiz, NJ). The cloned amplicon was then released from pJET1.2 using BamHI and SphI (New England Biolabs, MA) and subcloned into a pG108 shuttle vector (a gift from Dr. Janina Lewis, VCU), generating pG108-PG0352 (Fig. 1A). This plasmid was transformed into NEB5α cells and then purified using the Takara DNA purification kit. Approximately ~5.0 µg of pG108-PG0352 was electroporated into Δ0352, a previously constructed PG0352 deletion mutant (11), using a MicroPulser electroporator (Bio-Rad, CA). After electroporation, P. gingivalis cells were recovered in fresh pre-warmed TSB medium and grown at 37°C in the anaerobic chamber. After 2 days, P. gingivalis cells were plated onto TSA blood agar plates containing clindamycin (1.0 µg/mL) and tetracycline (1.0 µg/mL) (95). Antibiotic-resistant colonies were screened for the presence of PG0352 gene and tetracycline resistance gene (tetQ) using PCR with specific primers listed in Table 1. One complemented clone, CΔ0352, was selected for further characterizations.

TABLE 1.

PCR primers used in this study

Primer	Sequence (5′-3′)	Description
P1	GGATCCCTGGTTAGTTTTTGGTTTGTG	PG0352 complementation, Fa
P2	GCATGCTCATTGCCGGACATC	PG0352 complementation, Ra
P3	CATCAAGCAGGGTACCCCCGATAGCTTC	Confirmation of ermF insertion, F
P4	GGCGTTGCCCTCTTTTACGTTTCCGCTCC	Confirmation of ermF insertion, R
P5	AGAACGATATTTGGCGGATAGCGAAATTT	Confirmation of tetQ, F
P6	CCCAATAAGGGTTGGGCGGCACTTCGAT	Confirmation of tetQ, R
P7	GATAGCTTCCCGATCTCAAAGG	Confirmation of PG0352, PGN1608, F
P8	CTGTCGGCTCTCCTGCCGTC	Confirmation of PG0352, PGN1608, R
P9	AACTGGTTAGTTTTTGGTTTGTG	PGN1608 upstream region, F
P10	CGGGGGTACCTGAAAACTATTTTATACCATTTTGG	PGN1608 upstream region, R
P11	ATAGTTTTCAGGTACCCCCGATAGCTTC	ermF, F
P12	AGCACTATTCTTCCGCTCCATCGCCAATTTG	ermF, R
P13	TGGAGCGGAAGAATAGTGCTTTTTTTATCGAGTTTTTC	PGN1608 downstream region, F
P14	ACCGTACAGGCCCTAAAGGTTATCC	PGN1608 downstream region, R
P15	GCATGCCTGGTTAGTTTTTGGTTTGTGAAAAAT	PGN1608 complementation primer, F
P16	CTGCAGTCATTGCCGGGCATCGAAGAGATCGT	PGN1608 complementation primer, R
P17	GATAGCTTCCCGATCTCAAAGG	PG0352 and PGN1608 RT-PCR, F
P18	CCCTATATCCGCAGCCAAATAA	PG0352 and PGN1608 RT-PCR, R
P19	CCATGCAGCACCTACATAGAA	16S RT-PCR, F
P20	GATGATACGCGAGGAACCTTAC	16S RT-PCR, R

^a F: forward; R: reverse.

Construction of PGN1608 deletion mutant and its isogenic complemented strain in 33277

PGN1608 in 33277 is a homolog of PG0352. This gene was in frame deleted and replaced with an erythromycin resistance cassette (ermF) using two-step PCR as described before (11). In brief, to delete this gene, ~426 bp of the upstream (US’) and ~376 bp of the downstream (DS′) flanking regions of PGN1608 were PCR amplified from 33277 genomic DNA using primers P9/P10 and P13/P14, respectively, as illustrated in Fig. 1C. The ermF gene was amplified from the ermF/AM cassette using primers P11/P12 (11). The resulting products (US′, ermF, and DS′) were PCR ligated together. The resulting ~2.0 kb amplicon was cloned into the pJET1.2 vector and transformed into E. coli NEB5α cells. The knockout construct (PGN1608::ermF) was verified by DNA sequencing (GENEWIZ, NJ) and PCR-amplified. The yielded PCR product was purified and ~1.0 µg/µL DNA was electroporated into 33,277 cells. The same protocol described above was used for electroporation and plating. Clindamycin (1.0 µg/mL)-resistant colonies were screened by PCR for the absence of PGN1608. One mutant clone (Δ1608) was selected and further confirmed by PCR, DNA sequencing, and sialidase activity. The same method described above was used to complement the Δ1608 mutant using the shuttle vector pG108 and obtained isogenic complemented strain (CΔ1608) was confirmed by PCR and sialidase activity. The primers used here are listed in Table 1.

Production of PG0352 recombinant proteins and antibodies

The production of PG0352 recombinant proteins in Escherichia coli was described in our previous reports (11, 73). In brief, the nucleotide sequence encoding PG0352 without the N-terminal signal peptide (1–30 amino acids) was synthesized and codon optimized for expression in E. coli (GenScript). The synthesized gene was cloned into the pQE80L expression vector (Qiagen) and transformed into E. coli BL21-DE3 strain. PG0352 recombinant proteins expressed in BL21-DE3 were purified using a HisTrap HP Ni-NTA column (GE) under native conditions on an NGC FPLC System (Bio-Rad). Protein concentration was determined via a Pierce BCA Protein Assay Kit per manufacturer protocol (Thermo Fisher, Rockford, IL). The purified proteins were used to raise PG0352 antibodies in rats on a fee-for-service base (General Bioscience, CA), as previously described (52).

Detection of sialidase activity

Sialidase activity was detected using a filter spot paper assay as described before (11, 28). For this assay, sialidases cleave the substrate 4-MU-NANA (4-methyl-lumbelliferyl-α-D-N-acetylneuraminic acid) between 4-MU and NANA, liberating the fluorescent 4-MU moiety. This assay was adopted and modified to measure sialidase activity using either P. gingivalis whole-cell lysates or PG0352 recombinant protein (rPG0352), as previously described (11). In brief, P. gingivalis cells were harvested from 1 mL log-phase cultures by centrifugation; the resultant cell pellets were washed and resuspended in 50 µL of water or PBS and then spotted onto filter paper discs soaked in 4-MU-NANA. These discs were then incubated for 15 to 30 minutes at 37°C. Sialidase activity was detected by measuring fluorescence (λ_ex = 350 nM; λ_em = 450 nM) using a ChemiDoc Imaging System (Bio-Rad). For this assay, purified rPG0352 was included as a positive control, and 4-MU-NANA alone as a negative control.

RNA isolation and RT-PCR

P. gingivalis strains were grown to late-log and then spun down for RNA isolation using a QIAGEN RNeasy Mini Kit (Qiagen) following the manufacturer’s protocol. Following RNA extraction, residual DNA was further removed using a DNA-free DNase Treatment & Removal kit (Thermo Fisher) according to the manufacturer’s protocol. cDNA was synthesized using a SuperScript IV Vilo Master Mix kit (Thermo Fisher) followed by PCR using iQ SYBR green supermix and QuantStudio™ 3 System (Applied Biosystems, Waltham, MA). P. gingivalis 16S rRNA was used as an internal control. Primers for RT-PCR are listed in Table 1.

Measuring P. gingivalis growth rates

To determine whether deletion of PG0352 or PGN1608 impairs P. gingivalis growth, we measured their growth rates using three different growth media, including TSB, a chemically defined medium (CDM) (55), and 50% heat-inactivated rabbit serum (HI-RS) that mimics the growth condition in the gingival crevicular fluid (GCF). For TSB, individual P. gingivalis colonies were picked from blood agar plates, inoculated into TSB, and grown overnight. The next morning, these cultures were diluted into fresh TSB to an optical density (OD) at 600 nm (OD₆₀₀) of 0.1. Growth rates were monitored at 0, 4, 8, 24, 48, and 72 hours under anaerobic conditions at 37°C using a Genesys spectrophotometer (Thermo Fisher, IL). CDM was prepared and supplemented with 10 mM α-ketoglutarate, 3% BSA (bovine serum albumin), hemin (5 µg/mL), vitamin K (1 µg/mL), and adjusted to pH 7.4, as previously documented (55). To assess the impact of sialic acids on P. gingivalis growth, CDM was supplemented with Neu5Ac (30 µg/mL). For growth in HI-RS, rabbit serum was first heat-inactivated at 56°C for 30 minutes, then diluted to 50% (vol/vol) with sterile PBS, and supplemented with hemin and vitamin K (28). The resulting medium was equilibrated in the anaerobic chamber overnight and used for measuring P. gingivalis growth rates. P. gingivalis growth rates were represented as the averages of three independent experiments ± standard errors of means (SEM). Statistical analysis was assessed using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05.

Measuring P. gingivalis biofilm formation

Biofilm formation was assessed using polystyrene plates and sterile glass coverslips for microscopic examination, as previously described (11, 19, 29). In brief, for biofilm formation on polystyrene plates, overnight P. gingivalis cultures were diluted to OD₆₀₀ of ~0.1. Approximately 200 µL of these cultures were added to individual wells in a 96-well flat-bottom polystyrene tissue culture plate (Corning, NY). Biofilms were allowed to form for 3 days at 37°C under anaerobic conditions. After 3 days, the plates were removed from the anaerobic chamber and the growth medium was discarded. The plates were gently washed with water and then stained with 1% crystal violet for 30 minutes at room temperature (RT). Afterward, the plates were gently washed and then air-dried. Approximately 150 µL of 95% ethanol was added to each well and then incubated with gentle shaking at RT. Absorbance at OD₅₇₀ was measured using a Varioskan LUX multimode microplate reader (Thermo Fisher) for quantification. Statistical significance was assessed using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05. For microscopic examination, overnight P. gingivalis cultures were prepared and diluted as stated above. Approximately 1 mL of culture was added to a 12-well polystyrene plate containing sterile glass coverslips. The biofilms were allowed to mature for 3 days at 37°C under anaerobic conditions. And then, the growth medium was removed and the coverslips were gently washed with PBS, fixed with 4% paraformaldehyde, and then stained with 4′,6-diamidino-2-phenylindole (DAPI; 50 µg/mL)(29). The biofilms were observed on a KEYENCE BZ-X Series All-in-One fluorescence microscope (Keyence Corporation, Japan) on a 10× objective lens.

Cryo-ET data collection and tomogram reconstruction

Cryo-ET data collection on frozen-hydrated P. gingivalis cells was carried out with a similar protocol as previously described (11). Briefly, P. gingivalis cells (OD_600nm = ~0.1) in PBS were mixed with 10 nm gold fiducial particles, deposited onto a freshly glow-discharged Holey carbon grid (200 mesh, R2/1, Quantifoil), blotted, and rapidly frozen in liquid ethane with a home-made gravity-driven plunger apparatus. The frozen-hydrated specimens were transferred to a 300-kV Titan Krios electron microscope (ThermoFisher) equipped with a field emission gun, a post-Gatan imaging filter (GIF), and a K3 camera (Gatan). The tilt series was collected at a magnification of ×42,000 (pixel size of 2.15 Å) with SerialEM using the FastTOMO script with defocus values set as −6 µm. A total dose of ~70 e−/Ų was distributed over 33 tilt images covering angles from −48° to +48° with an angular increment of 3°. The recorded images were motion corrected with MotionCorr2 (96) and aligned using IMOD (97). In total, 60 tomographic reconstructions were generated by TOMO3D (98). IMOD was used to take snapshots from tomograms.

Measuring P. gingivalis CSH)

This experiment was performed using n-hexadecane according to the method described by Rosenberg with some minor modifications (38, 39). Briefly, individual P. gingivalis strains were grown in TSB overnight, harvested by centrifugations, washed with PBS, and then resuspended in PUM buffer (150 mM potassium phosphate, 30 mM urea, and 0.8 mM magnesium sulfate at pH 7.1). After adjusting cell numbers, 3.0 mL of P. gingivalis cell suspensions was placed into a clear test tube and then mixed with 200, 400, or 800 µL of n-hexadecane (Sigma-Aldrich, MO) (38). The tubes were vortexed for 60 seconds and then set aside for 15 minutes at room temperature to allow the layers to partition. A sample of the top layer was carefully removed and then transferred to a 96-well plate. In addition, a sample of the aqueous phase was taken before and after the addition of n-hexadecane to calculate the percentage of bacteria adhered to hexadecane using the following formula: [(OD_Before − OD_After) / OD_Before] × 100, as previously reported (99). The optical density (OD₅₅₀) was measured using a Varioskan LUX multimode plate reader (Thermo Fisher). PUM buffer with hexadecane was used as a reference. These assays were performed in triplicate; the final readouts were expressed as means ± standard errors of means (SEM). Statistical analysis was performed using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05.

Detection of P. gingivalis APS/A-LPS by immunoblotting and silver staining

P. gingivalis whole cell lysates were prepared as previously described (33, 63). P. gingivalis cultures (~5 mL) were grown in TSB to the stationary phase, harvested, washed with PBS, and then resuspended in PBS. The obtained samples were first treated with 10 mM Nα-tosyl-L-lysine chloromethyl ketone hydrochloride (TLCK, Sigma-Aldrich) for 10 minutes on ice. After treatment, SDS-PAGE loading buffer was added and then boiled for 5 minutes. After boiling, P. gingivalis cell lysates were allowed to cool to room temperature and then incubated with Proteinase K (0.1 mg/mL) overnight at 50°C (33). The obtained samples were subjected to 12% SDS-PAGE gel, followed by either immunoblotting or silver staining. For immunoblots, the samples were transferred to PVDF membranes, blocked with 5% non-fat skim milk in PBS-T [0.05% Tween-20 (vol/vol) in PBS], and then probed against MAb1B5 (1:100 dilution in 2% non-fat milk in PBS-T), a mouse monoclonal antibody that recognizes P. gingivalis APS/A-LPS (36). The blots were subsequently probed with anti-mouse horse peroxidase-conjugated IgG (1:5,000 dilution in 2% non-fat milk in PBS-T). Immunoblotting signals were detected by an ECL luminol assay (Bio-Rad Laboratories) and visualized with a ChemiDoc MP Imaging System (Bio-Rad Laboratories). Silver staining was performed using the Pierce Silver Stain Kit (Thermo Fisher) per the manufacturer’s protocol.

Serum killing assays

These assays were conducted using fresh human serum as previously described (11, 28). In brief, overnight P. gingivalis cultures derived from single colonies were diluted in TSB medium and grown to mid-log phase (OD₆₀₀ 0.5–0.7). After adjusting cell numbers, 75 µL of P. gingivalis cultures were mixed with 25 µL of either human serum (HS) or heat-inactivated HS (HI-HS) to a final concentration of 25% serum (vol/vol) and then incubated at RT for 1 hour in the anaerobic chamber. After the incubation, the samples were serially diluted, plated onto TSA blood agar plates, and incubated at 37°C for 6 days. And then, P. gingivalis colonies were enumerated to calculate survival rates, which were recorded as follows: the number of P. gingivalis colonies in HS relative to that of colonies in HI-HS. Data are represented as the averages of three independent experiments ± SEM. Statistical analysis was determined using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05.

Human cell lines and growth conditions

Human telomerase immortalized gingival keratinocyte cell line (TIGK) was isolated from a healthy human adult and immortalized by way of a bmi1/hTert combination to inhibit telomerase shortening and senescence (48). TIGK cells were cultivated and maintained in serum-free keratinocyte growth medium (KGM; Lifeline Cell Technology, MD) in T75 tissue culture flasks in an incubator at 5% CO₂ at 37°C. The KGM medium for the TIGKs was replenished every 2 days. TIGK cells were grown to confluency before usage in subsequent experiments or before splitting/passaging. TIGK cells were used to assess P. gingivalis invasion, as previously described (100, 101). HL-60 cell line was purchased from the American Type Culture Association (ATCC, VA). This cell line was originally isolated from the peripheral blood of a female with acute promyelocytic leukemia and is frequently used to study neutrophil functions (102). HL-60 cells were cultivated and maintained in Iscove’s Modified Dulbecco Medium (IMDM; Thermo Fisher) supplemented with 20% (vol/vol) fetal bovine serum per the ATCC protocol. For neutrophil killing studies, HL-60 cells were differentiated into neutrophils by supplementing the IMDM with DMSO 5 days prior to use. Cell differentiation was assessed microscopically and using immunoblotting analysis for CD11b, a biomarker for neutrophils (data not shown). Prior to neutrophil killing assays, DMSO-differentiated HL-60 cells were stimulated with 5 µM phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich, MO) for 30 minutes at 37°C.

Neutrophil killing assays

Neutrophil killing assays were performed using PMA-stimulated HL-60 cells as previously described (45). In brief, individual P. gingivalis colonies were picked from TSA blood agar plates and grown overnight in a TSB medium. The next morning, P. gingivalis cultures were diluted and grown to mid-log phase (OD₆₀₀ 0.5–0.7), spun down, and then washed 2× with pre-balanced Hanks-Balanced Salt Solution (HBSS; Thermo Fisher). HBSS was equilibrated in the anaerobic chamber overnight. DMSO-differentiated cells were seeded at 1 × 10⁵ cells per well in a sterile 24-well non-treated polystyrene plate (Corning), stimulated with 5 µM PMA for 30 minutes at 37°C, and then co-incubated with P. gingivalis cells at a multiplicity of infection (MOI) of 1 at 37°C in 5% CO₂ for 30 minutes. Non-stimulated HL-60 cells were used as a control. Following the incubation, the plates were brought into the anaerobic chamber and lysed with 1% filtered saponin prepared in TSB media for 20 minutes. The cell lysates were then serially diluted and plated onto TSA blood agar plates. The plates were incubated at 37°C for 6 days in the anaerobic chamber before P. gingivalis colonies were enumerated. P. gingivalis survival rates were calculated as follows: the colony forming unit (CFU) in PMA-HL60 divided by that in non-PMA stimulated HL-60. For the intracellular survival assays, DMSO-differentiated HL-60 cells were seeded at 1 × 10⁵ cells/well in 24-well non-treated polystyrene plates, stimulated with PMA, and then infected with different P. gingivalis strains at an MOI of 10 at 37°C in 5% CO₂ for 1 hour (45). The resulting samples were then treated with gentamicin (300 µg/mL) and metronidazole (300 µg/mL) for 1 hour to eliminate extracellular P. gingivalis cells. Cell suspensions were transferred to microcentrifuge tubes and carefully washed with HBSS to remove residual antibiotics. The cells were then lysed with 1% filtered saponin in TSB medium, serially diluted, and plated onto TSA blood agar plates for colony counting. Non-stimulated DMSO differentiated HL-60 cells were included as a control. Survival rates were calculated as follows: CFU in PMA-HL60 divided by that in non-PMA stimulated HL-60. Statistical analysis was determined using one-way ANOVA followed by Tukey’s multiple comparisons at P < 0.05.

Cell invasion assays

A standard antibiotic protection assay was modified for assessing P. gingivalis invasion in TIGK cells (48, 49). In brief, TIGK cells were seeded at 10⁵ cells/well on 12-well tissue culture polystyrene plates (Corning) and grown to confluency at 37°C in 5% CO₂. On the day of infection, overnight P. gingivalis cultures were diluted in fresh TSB and grown to mid-log phase (OD₆₀₀ 0.5–0.7). P. gingivalis cells were then spun down and washed 2× with an equilibrated KGM medium. Prior to infection, 12-well plates containing TIGK cells were equilibrated in the anaerobic chamber for at least 30 minutes at 37°C, and then infected with P. gingivalis cells at an MOI of 10 for 1 hour at 37°C under anaerobic conditions. The infected TIGKs were washed with sterile PBS and then incubated with KGM supplemented with gentamicin (300 µg/mL) and metronidazole (400 µg/mL) for 1 hour to kill extracellular bacteria. After the antibiotic treatment, infected TIGK cells were washed with sterile PBS to remove residual traces of antibiotics and then lysed with 1% filtered saponin. Cell lysates were serially diluted and plated on TSA blood agar plates. The plates were incubated at 37°C for 6 days under anaerobic conditions for measuring CFU. Percent invasion was calculated as follows: CFU recovered after antibiotic treatment divided by CFU of total P. gingivalis cells used.

Detection of P. gingivalis-induced proinflammatory cytokines

Peripheral blood mononuclear cells were prepared as previously described (103). Human blood was collected from healthy donors with written informed consent. After venipuncture, blood was collected into sterile tubes containing EDTA to prevent coagulation. Monocytes were isolated using a human monocyte isolation kit II (Miltenyi Biotec, MD). Purity was determined by flow cytometry with a FITC-labeled anti-CD14 antibody. Enzyme-linked immunosorbent assays (ELISAs) were performed to detect cytokines released by human monocytes. For this assay, 2 × 10⁵ monocytes were plated per well on a 96-well microplate in RPMI 1640 medium and then challenged with late log/early stationary P. gingivalis cells at an MOI of 10 for 8 hours and then cell-free supernatants were collected for measuring the production of IL-6, TNF-α, IL-12 p40, and IL-1β using ELISA kits (human IL-6 and TNFα, Invitrogen; human IL-1β and IL-12/IL-23 (p40), Biolegend) according to the manufacturers’ protocols. Approximately 20 µL of cell-free supernatants was diluted 1:10 for the IL-6 assay; 100 µL of cell-free supernatants was diluted 1:4 for TNF-α, followed by twofold serial dilutions until the dilution reached 1:80 and 1:32, respectively. IL-12 p40 and IL-1β were diluted from 1:3, followed by twofold serial dilutions three times. The dilution buffer (provided in the kit) was used as a blank control; the supernatants from unstimulated cells were used as sham controls. A standard curve was generated in each assay using recombinant standard proteins provided in the kit. SpectraMaxR iD3 (Molecular Devices) was used for OD measurements at OD₄₅₀ and OD₅₄₀ in triplicate. The optical imperfections in the plate were corrected by subtracting OD₅₄₀ from readings at OD₄₅₀. Concentrations were calculated using GraphPad Prism with interpolation of a linear standard curve. ELISAs were performed in triplicate and the averages were used for quantifications.

AFTER EPUB

[This article was published on 20 February 2024 with the Fig. 5 and 6 legends reversed. The error was corrected in the current version, posted on 12 March 2024.]