Loss of Neutrophil Homing to the Periodontal Tissues Modulates the Composition and Disease Potential of the Oral Microbiota

ABSTRACT

Periodontal disease is considered to arise from an imbalance in the interplay between the host and its commensal microbiota, characterized by inflammation, destructive periodontal bone loss, and a dysbiotic oral microbial community. The neutrophil is a key component of defense of the periodontium: defects in their number or efficacy of function predisposes individuals to development of periodontal disease. Paradoxically, neutrophil activity, as part of a deregulated inflammatory response, is considered an important element in the destructive disease process. In this investigation, we examined the role the neutrophil plays in the regulation of the oral microbiota by analysis of the microbiome composition in mice lacking the CXCR2 neutrophil receptor required for recruitment to the periodontal tissues. A breeding protocol was employed that ensured that only the oral microbiota of wild-type (CXCR2^+/+) mice was transferred to subsequent generations of wild-type, heterozygote, and homozygote littermates. In the absence of neutrophils, the microbiome undergoes a significant shift in total load and composition compared to when normal levels of neutrophil recruitment into the gingival tissues occur, and this is accompanied by a significant increase in periodontal bone pathology. However, transfer of the oral microbiome of CXCR2^−/− mice into germfree CXCR2^+/+ mice led to restoration of the microbiome to the wild-type CXCR2^+/+ composition and the absence of pathology. These data demonstrate that the composition of the oral microbiome is inherently flexible and is governed to a significant extent by the genetics and resultant phenotype of the host organism.

INTRODUCTION

As is the case in many chronic inflammatory conditions, periodontal disease is considered to arise from an imbalance in the interplay between the host and its commensal microbiota (1, 2). In this instance, the disease is characterized by a deregulated and injurious inflammatory response in the periodontal tissues with consequent irreversible destruction of bone and major changes to the total load and community structure of the subgingival microbiota, frequently referred to as dysbiosis (3, 4). Unsurprisingly, risk factors for periodontal disease include defects, or reductions in efficacy, in the repertoire of antimicrobial host defense measures required for protection of the periodontal tissues (5, 6).

The role that the dysbiotic microbiota plays in the disease process is controversial. On the one hand, it is evident that the changes in the hierarchy of the predominant organisms in the subgingival domain are driven by the altered environmental conditions, including the intensity of the inflammatory response and the availability of nutrients for bacterial growth. These pressures exert selective effects on the microbial community organization, resulting in enhanced proliferation of inflammophilic organisms and reductions in those bacteria routinely associated with low levels of inflammation (7). Furthermore, in some circumstances, alleviation of the stressed environmental conditions through interventionist approaches designed to reduce the inflammatory response in periodontal patients has been shown to lead to a reversal of dysbiosis and restoration of the community structure associated with health (8, 9).

Contrary to this inflammocentric view of a bystander role for the microbiota in periodontal disease, it is clear that organisms that predominate in subgingival dysbiosis have properties consistent with a more direct influence in driving the destructive disease process. The secretion of a broad range of hydrolytic enzymes and cytotoxic agents and employment of mechanisms which both subvert and deregulate the inflammatory and immune response all provide some justification for considering these disease-associated organisms as causative in the pathology of this condition (10). Understanding the contribution of the microbial challenge to the development and progression of periodontal disease has implications particularly in relation to the principles of treatment of this condition, targeted to the control of the dysbiotic bacterial community or the inflammatory response.

In studies using the murine model of periodontal bone loss, we previously demonstrated that introduction of Porphyromonas gingivalis strain W50, a bacterium frequently associated with human disease, into the oral cavity led to the development of dysbiosis of the murine commensal oral microbiota and concomitant periodontal bone loss (1, 11). Critically, in the context of dysbiosis as a cause or consequence of the disease, we showed that once the commensal microbiota was transformed into a dysbiotic mode with elevated load and altered community organization, the structure of the community was able to transfer into healthy animals and recapitulated the disease experience of the donor mice (11). These data suggested that, in this particular instance of experimentally induced dysbiosis, there is an inherent resilience to perturbation of the altered community, a reversal in the environmental conditions to those found in health, in this instance vertical or horizontal transfer into nondiseased mice, that did not lead to restoration of the health-associated microbiota or abrogation of the pathological response. In addition, we have also demonstrated in these studies and others the inherently low diversity of the laboratory mouse oral microflora, which therefore makes laboratory culturing a convenient and effective way to represent the microbial population in these mice (1, 11, 12).

In the present work, we aimed to extend these findings to an alternative murine model that does not rely upon the introduction of a human periodontal organism to provoke the disease process in mice. We performed these experiments in the CXCR2^−/− mouse, which is compromised in its ability to recruit neutrophils into the periodontal tissues because of the absence of the receptor, which guides these cells along chemokine (CXCL1 and CXCL2) gradients produced by the junctional epithelium at the subgingival tooth surface. The neutrophil is regarded as a critical element of the defense of the periodontium in humans, and defects in either the number or efficacy of function of neutrophils predisposes to the development of periodontal disease (13–15). Accordingly, we have previously demonstrated that the absence of neutrophils in the periodontal tissues of the CXCR2^−/− mouse is correlated with a higher oral microbial load and more periodontal bone loss than for wild-type animals (16). In this work, we aimed to shed further light on the relationship of oral microbial dysbiosis to the development of disease by analysis of the composition of the oral microbiota in CXCR2^−/− and wild-type BALB/c mice and the potential stability of the CXCR2^−/− microbiome and associated disease phenotypes to transfer into a wild-type background.

RESULTS

A modified host genotype (CXCR2^−/−) prevents neutrophil transmigration in oral tissues and leads to the development of a dysbiotic oral microbiome.

Immunohistochemical analysis of the F2 mouse groups enabled examination of the influence of genotype on the transmigration of neutrophils and expression of chemokine signals (CXCL2) in the periodontal tissues. All of the second-generation mice demonstrated localized expression of CXCL2 in the junctional epithelium, indicating that the absence of the CXCR2 gene had no effect on expression of this chemokine (Fig. 1A). Neutrophil staining showed the expected high localization in the junctional epithelium of wild-type (WT) mice and lower levels in the oral epithelium (Fig. 1B). Neutrophil staining was not detected in the junctional epithelium of homozygote CXCR2^−/− animals (HM), consistent with the absence of the neutrophil receptor for junctional epithelium (JE)-derived chemokine gradients in these mice. Neutrophils were present in the junctional epithelium of heterozygote mice (HT) but at a lower level than their WT littermates (Fig. 1B). These visual observations were then confirmed by neutrophil quantitation. Neutrophils were counted and expressed as percentage of the total number in the entire tissue section in each of the JE, oral connective tissue (OT), oral epithelium (OE), and blood vessels (BV) (Fig. 1C). In WT mice, 25% of neutrophils were present in the JE, 5.1% in the OE, 13.7% in the OT, and the remainder (65%) within the vasculature (BV). In contrast, only 0.06% of neutrophils were located in the JE of HM mice, and over 95% of cells were confined to the BV. In the HT mice, 7% of the neutrophils were found to be in the JE and 86% in the blood vessels (Fig. 1C). Hence, in the absence of CXCR2, neutrophil transmigration to the gingival crevice is abolished, and the oral microbiome in these animals develops in the absence of this element of the innate host defense.

FIG 1.

Detection of CXCL2 and neutrophils in oral tissues of BALB/c CXCR2^+/+, CXCR2^+/−, and CXCR2^−/− mice. (A) CXCL2 staining in the F2 generation wild-type (WT), homozygous (HM), and heterozygous (HT) mice (×20 magnification). CXCL2 was detected in blood vessels and JE of all the three genotypes. (B) Neutrophil staining of F2 generation WT, HM, and HT mice (×20 magnification). Most neutrophils were localized to the JE of WT mice with reduced staining in the JE of HT mice and minimal detection in HM. (C) Quantification of neutrophils in the junctional epithelium, oral connective tissue, oral epithelium, and blood vessels was performed by counting their numbers in tissue sections of F2 genotypes and were expressed as percentages of the overall cell numbers present. Data are means ± SD. (WT, n = 5; HT, n = 4; HM, n = 2).

We next examined the influence of mouse genotype on the oral commensal microbiota from 16-week-old parents (BALB/c WT and CXCR2^−/−). The predominant organisms in the oral microbial population of all mouse groups belonged to the genus Streptococcus, with Gemella species 2 and low levels of Actinobacillus muris and Enterobacteriaceae members in BALB/c WT mice. The oral microbiota of the CXCR2^−/− mice, on the other hand, was substantially different, where, apart from the dominant Streptococcus, the remaining community was composed of Actinobacillus muris and members of the Enterobacteriaceae family. Gemella species 2 MOT43 was recently identified as a unique, yet-to-be named murine-specific species of the genus, found to be predominant in the oral microbiome of certain specific-pathogen-free (SPF) laboratory mouse backgrounds (12). Comparative permutational analysis of variance (PERMANOVA) analyses showed a significant difference between the oral microbial compositions of the BALB/c WT and CXCR2^−/− mice (P = 0.026; R² = 0.73) (Fig. 2A).

FIG 2.

Effect of the mouse genotype on the oral microbiota and total microbial counts of F0 parents and F2 offsprings. (A) Composition of the cultivable oral microbiome of 16-week-old BALB/c WT mothers and the CXCR2^−/− fathers (F0 generation) expressed as relative abundance in percentages (y axis). The different-colored segments represent bacterial species based on the mean CFU of each microorganism in each group. Statistical significance in the differences between the microbial communities of the two mouse groups was determined by PERMANOVA and has been indicated. (B) Total oral bacterial counts in 16-week-old BALB/c WT mothers and homozygous CXCR2^−/− fathers (F0 generation) and second-generation wild-type CXCR2^+/+ (F2 WT), homozygous CXCR2^−/− (F2 HM), and heterozygous CXCR2^+/− (F2 HT) mice, expressed as log₁₀ of the total number of CFU. Each data point represents an individual mouse (**, P < 0.05; ***, P < 0.005; ****, P < 0.0005). (C) Linear regression analysis of oral bacterial counts against percentage of neutrophils in the junctional epithelium in the various F2 mouse genotype groups.

In addition to a different community structure, the CXCR2^−/− mice also showed elevated total microbial counts compared to the BALB/c WT mice (mean, 8.47 × 10⁷ versus 1.76 × 10⁷ CFU; P < 0.05) (Fig. 2B). Similarly elevated total microbial counts have been observed in previous studies of oral microbial dysbiosis in both mouse and human periodontal investigations (1, 11).

Vertical transmission of the maternal CXCR2^+/+ oral microbiome into second-generation CXCR2^−/− mice demonstrates the direct influence of genotype on the composition of the oral microbiota.

To verify a direct influence of mouse genotype on the oral microbiota, we next determined the quantitative composition of the oral microbiome of the F2 wild-type CXCR2^+/+, heterozygous CXCR2^−/+, and homozygous CXCR2^−/− mice (Fig. 2B). Importantly, as the CXCR2^−/− mice in the F0 breeding were removed prior to the birth of the F1 generation, the F1 generation and subsequent F2 generations were only exposed to those bacteria present in the maternal CXCR2^+/+ (BALB/c WT) oral microbiome. The pattern of total microbial counts seen in the F2 generation mice mirrored that of the original F0 breeders. The total microbial counts of the F2 HM mice were significantly higher (mean, 4.4 × 10⁷ CFU) than those of both the F2 WT mice (mean, 4.07 × 10⁶ CFU; P < 0.0005) and the F2 HT mice (6.47 × 10⁶ CFU; P < 0.0005). Thus, there is an inverse correlation between dosage of the cxcr2 gene and total oral microbial counts in these mice.

A strong negative correlation (R² = 0.7859) was also observed between the percentage of neutrophils in the junctional epithelium and total oral microbial counts: BALB/c WT mice with the highest proportion of JE associated neutrophils had the lowest bacterial counts, and the homozygote CXCR2^−/− mice with a very low proportion of neutrophils in this epithelium had the highest microbial burden (Fig. 2C).

These experiments therefore demonstrate the flexibility of the oral microbiome of the CXCR2^+/+ mice (BALB/c WT) in response to the genetic background of the host, in this case resulting in the presence or absence of neutrophils in the periodontal tissues. In summary, the oral microbiome of the BALB/c WT mice is inherently flexible, is strongly governed by the genetic background of the host organism, and will form reproducible symbiotic and dysbiotic signatures upon vertical transfer into either wild-type or CXCR2^−/− offspring, respectively.

Relationship between host genotypes on alveolar bone levels.

The influence of the genotype on the development of periodontal disease was assessed by measuring the alveolar bone loss relative to the bone height of the F0 CXCR2^+/+ (BALB/c WT) animals. (Fig. 3). The F0 CXCR2^−/− mice displayed significantly elevated bone loss (mean, −0.11 ± 0.03 mm; P < 0.05) compared to BALB/c WT mice. This level of bone loss is similar to that observed in previous studies following oral challenge with Porphyromonas gingivalis (1, 11). Thus, it has been demonstrated that the deletion of the CXCR2 gene predisposes these animals to periodontal bone destruction. The second-generation CXCR2^−/− mice (F2 HM) also demonstrated reduced bone levels similar to those of F0 CXCR2^−/− mice (mean, −0.11 ± 0.04 mm; P< 0.05). The F2 WT mice also showed higher levels of bone loss (mean, −0.067 ± 0.02; P < 0.05) than the F0 WT animals. However, no significant difference was observed in the bone levels compared with their F2 HM littermates (mean, −0.06 ± 0.02 mm). This may reflect the continuous cohousing of these littermates of different genotypes in excess of 8 weeks: we have previously suggested that cocaging of mice with markedly different oral microbiomes and bone loss phenotypes can lead to cage-normalization effects wherein a dysbiotic oral microbiome in one group of animals may have an influence on the microbiome and disease phenotype of cocaged animals with normal health-associated oral microbiome (11).

FIG 3.

Effect of the mouse genotype on alveolar bone levels in F0 parents and F2 generation offsprings. Alveolar bone levels in 16-week-old BALB/c WT mothers and CXCR2^−/− fathers (F0 generation), homozygous CXCR2^−/− (F2 HM), heterozygous CXCR2^+/− (F2 HT), and wild-type CXCR2^+/+ (F2 WT) is shown. Bone loss was expressed as negative values relative to the BALB/c WT mothers, the donors of the maternal microbiome (baseline). Each point represents the mean bone level for an individual mouse, with horizontal lines representing the mean bone levels per group ± standard deviations (**, P < 0.05).

The dysbiotic microbiome from CXCR2^−/− mice is not stable upon transfer to wild-type mice and does not lead to transfer of the bone loss phenotype into recipients.

In previous work, we demonstrated the efficient transfer of the dysbiotic microbiome, generated following oral gavage of SPF mice (C3H/Orl) with Porphyromonas gingivalis, into germfree (GF) mice (C3H/Orl) by cocaging donor and recipient mice at a 1:2 ratio for 6 weeks. Transfer and establishment of the dysbiotic microbiota was accompanied by the development of periodontal bone loss in the formerly GF mice, equivalent to that observed in the original challenged mice (11). Here, we used the same cocaging protocol to transfer the oral health-associated microbiome from BALB/c WT mice and the dysbiotic microbiome from CXCR2^−/− mice into C3H/Orl GF recipients. The oral microbiota of donor and recipient mice was examined at 1, 3, and 6 weeks of cocaging.

The commensal oral microbiome from BALB/c WT mice transferred to GF mice, with no significant differences to the qualitative or quantitative composition of the microbiota assessed by laboratory culture at all three time points (Fig. 4A and B). At the end of 6 weeks, the donor BALB/c WT mice and recipient GF-BALB/c mice had a microbiota predominantly comprised of the Streptococcus and Gemella species, both with mean total counts of 1.13 × 10⁷ CFU.

FIG 4.

Transfer of the CXCR2^−/− oral microbiota to germfree mice is not to the same total microbial load and does not induce bone loss. (A) Bacterial composition of the cultivable oral microbiome of BALB/c WT and CXCR2^−/− mice and after transmission into initially germfree C3H/Orl mice (GF-BALB/c and GF-CXCR2^−/−) sampled at 1, 3, and 6 weeks. The sizes of the pie charts are indicative of the variations in the total oral bacterial counts in the different groups. The graphs have been plotted using the observed number of CFU of each microorganism in each group. Statistical significance in the differences between the microbial communities of groups was determined by PERMANOVA (**, P < 0.05; ***, P < 0.005; ****, P < 0.0005). (B) Total oral bacterial counts in the different groups of mice at weeks 1, 3, and 6 expressed as log₁₀ of total CFU. Each data point represents an individual mouse (**, P < 0.05; ***, P < 0.005). (C) Alveolar bone levels in BALB/c WT and CXCR2^−/− mice and conventionalized germfree C3H/Orl mice (GF-BALB/c and GF-CXCR2^−/−) after 6 weeks. Bone loss was expressed as negative values relative to the BALB/c WT donors. Each point represents the mean bone level for an individual mouse, with horizontal lines representing the mean bone levels per group ± standard deviations (**, P < 0.05; ***, P < 0.005).

In contrast, a different pattern of microbial transfer was observed using the CXCR2^−/− mice as donors into GF animals. This was particularly marked in the case of the total microbial counts. After 1 week of cocaging, the mean oral bacterial count of the recipient animals (GF-CXCR2^−/−) was 5.14 × 10⁷ CFU, which was not significantly different from that of the donor CXCR2^−/− mice at the same time point (9.6 × 10⁷ CFU; P = 0.05). However, by 3 weeks and until the end of the experiment, the mean total oral microbial counts of the recipient mice fell to significantly below the level of the CXCR2^−/− donors (3 weeks: mean of 1.02 × 10⁷ CFU versus donor mean of 6.24 × 10⁷ CFU, P = 0.0002; 6 weeks: mean of 1.23 × 10⁷ CFU versus donor mean of 9.7 × 10⁷ CFU, P = 0.0002). At both 3 weeks and 6 weeks, the mean total oral counts of the GF-CXCR2^−/− microbiota were equivalent to the mean counts of BALB/c WT mice and the GF-BALB/c recipient mice (Fig. 4B). The microbial composition of the GF-CXCR2^−/− recipient mice also reverted to represent the GF-BALB/c mice, comprised of Streptococcus, Gemella, and Enterobacteriaceae species. Actinobacillus muris, which was one of the major components of the donor CXCR2^−/− mice, was conspicuously absent from the recipients after 6 weeks (Fig. 4A). The interindividual variation in these mice groups has also been presented in Fig. S2 in the supplemental material in the form of box-and-whisker plots.

We next examined the levels of periodontal bone loss in donor and recipient mice. The conventionalized recipient GF-BALB/c of the BALB/c WT mouse microbiome did not show any differences in the alveolar bone levels compared to the donors (Fig. 4C). As anticipated, CXCR2^−/− donor mice showed significantly more bone loss than the BALB/c WT mice of the same age (−0.122 mm, P ≤ 0.005). However, the recipient GF-CXCR2^−/− mice did not demonstrate increased alveolar bone loss compared to GF-BALB/c mice after 6 weeks of cohousing (Fig. 4C). Hence, these transfer experiments demonstrated that the oral microbiota of CXCR2^−/− mice is not stable to transfer into a wild-type mouse background, and cocaging does not lead to transfer of the disease phenotype of these mice.

DISCUSSION

In this study, we aimed to determine the influence of reduced effectiveness of the innate defenses operative within the periodontal tissues, in this instance abrogation of neutrophil transmigration, on the nature and stability of the oral microbiota. We have demonstrated how the absence of the murine CXCR2 gene led to an almost complete loss of neutrophil homing to the periodontal tissues, which reinforced the crucial role played by this receptor and its two murine ligands, CXCL1 and CXCL2, in recruitment of this cell type into the periodontal tissues of the mouse. It also provided the opportunity to determine the impact of a significant loss of periodontal defense function on the oral microbiota in this model.

The human leukocyte adhesion deficiency (LAD) group of inherited disorders presents a similar pattern of reduced neutrophil extravasation and their recruitment to sites of infection/inflammation and the periodontium. LAD patients have defects in the expression or function of the leukocyte-restricted β2 integrins or other adhesion molecules. Consequently, circulating neutrophils of LAD patients do not adhere to vascular endothelial cells and, hence, are unable to leave the blood vessels and enter the tissues (17). In all the different variations of the condition, LAD patients are found to be highly susceptible to periodontal disease (18). However, because of the rarity of these monogenic diseases, there have been few comprehensive studies on the microbiology of these patients, although LAD-1 patients are reported to harbor more bacterial biomass than control subjects (19), and Gram staining of extracted teeth and surrounding tissues from LAD-I patients has demonstrated very significant microbial colonization of tooth surfaces, although not of the underlying diseased gingival tissue (20).

Mouse models of reduced neutrophil trafficking to the oral tissues provide an alternative means to address the question of the influence of defects in the innate host defenses on the oral microbiota (21). However, there are few such comparative studies of the oral microbiology of wild-type compared to appropriate homozygous gene knockout mice that have properly controlled for animal husbandry/cage effects. When mice of different genotypes are sourced and reared separately, it becomes problematic to differentiate the effects of variations to the microbiota caused by differences in mouse genotype from variations caused by different environmental exposures during rearing. In the present study, we developed a breeding protocol (see Fig. S1 in the supplemental material) in which second-generation F2 WT, HT, and HM mice were only exposed to a microbiome which had originated from the original maternal BALB/c WT F0 generation. Differences in the oral microbiota of the F2 mice can therefore be ascribed reasonably to the influence of genotype.

We have previously demonstrated that the oral microbiota of laboratory (and wild) mice exhibits low diversity compared to human studies and is frequently dominated by a small number of microbial taxa and a very low proportion of uncultivable species (11, 12). Culture therefore provides an appropriate means of analysis of the community structure in addition to providing a direct readout of total microbial load, with congruence being observed with the 16S rRNA gene amplicon-based population analysis as well (11, 12). As seen in previous mouse studies, similarly low-diversity oral microbiomes were observed in the current study (Fig. 2A). Analysis of the oral microbiota of F2 littermates revealed significant differences in the total oral bacterial counts inversely proportional to the dosage of the CXCR2 allele, CXCR2^−/− > CXCR2^+/− > CXCR2^+/+, and the elevated bacterial numbers present in the F2 HM mice were similar to the counts in the original F0 CXCR2^−/− founder mice. The total bacterial counts of the F2 WT mice were higher than those in the F0 wild-type mice, which we speculate is a consequence of the prolonged cocaging of the former with their F2 HT and HM littermates (Fig. 2B).

These findings accord with accumulating evidence on the modifying effects of host genetics on the human oral microbiome, particularly that derived from studies on mono- and dizygotic twins (22, 23). An analysis of over 750 twin pairs demonstrated that, independent of cohabitation status, the beta-diversity of monozygotic twins is significantly lower than that for dizygotic or unrelated individuals. In this population, a number of microbiome phenotypes were more than 50% heritable, consistent with the hypothesis that human genes influence oral microbial communities (24). Similarly, in a study of 485 monozygotic and dizygotic twins (25), the similarity of the oral microbiome increased with shared genotype, and although most of the variation in the microbiome was determined by environmental factors, highly hereditable oral taxa were also identified. More recently, however, Mukherjee and colleagues (26) determined the influence of host genetics on acquisition of the oral microbiome by comparing the oral microbiota of 55 biological versus 50 adoptive mother-child dyads. Only children adopted immediately at birth and unrelated to the adoptive family were included to minimize transmission of bacteria from the biological mother. They discerned no difference in how closely oral bacterial community profiles matched for adoptive versus biological mother-child pairs, and, from this, concluded that there is little, if any, effect of host genetics on the fidelity of transmission. The apparent contradictory finding of a significant effect of genotype on the acquisition of the oral microbiome that we report in the current study using the CXCR2^−/− mouse is likely to reflect the very different phenotypes of the wild-type versus homozygote knockout animals with respect to neutrophil extravasation into the periodontal tissues.

Following earlier studies in this laboratory, which indicated that a dysbiotic oral microbiome generated by oral gavage with the human periodontal organism P. gingivalis is stable to transfer both horizontally and vertically and transmits the disease phenotype into recipient animals, we examined whether the CXCR2^−/− dysbiotic oral microbiome shared similar properties. When CXCR2^−/− mice were cohoused with GF wild-type mice (Fig. 4A and B), we observed transmission of the dysbiotic oral microbiota to a similarly raised bacterial count after 1 week. However, by week 3 and until the end of the experiment, the level of bacterial colonization was reduced to the level of the total counts routinely observed in wild-type mice. The temporary nature of the initially high counts and their subsequent reduction may reflect temporal adaptation of the GF mice to microbial exposure. It is well established that the response to microbes of mice reared under germfree conditions differs from conventionally reared animals and can lead to a delay in the inflammatory response (27). Specifically in relation to the oral cavity, Fukuhara and colleagues (28) reported that following administration of lipopolysaccharide from P. gingivalis to the gingiva, GF mice show decreased numbers of CD4⁺ cells in the periodontal tissues and lower expression of tumor necrosis factor alpha and fork head box protein p3 than SPF mice of similar age. Hence, in the current work, the initial exposure to the high levels of bacteria in the CXCR2^−/− mice may have led to immediate colonization at the same level in the recipients (GF-CXCR2^−/−), but, following a period of maturation of the immune system in these mice, colonization levels reduced to the level seen in wild-type animals. In this regard, it is perhaps noteworthy that the levels of bacteria in GF-BALB/c mice housed with BALB/c WT animals were also at their highest, albeit not significantly, at week 1. The overall community composition of the microbiota of the recipients was significantly different from the CXCR2^−/− donor mice at all time points. Conversely, no difference in either total microbial load or composition of microbiota was observed between wild-type donor and recipient mice.

As might be predicted based on these microbiological findings, recipients of the CXCR2^−/− microbiota experienced similar levels of bone loss as wild-type animals and far lower levels than the donor mice (Fig. 4C). These findings contrast with many reports regarding the transfer of dysbiotic microbiomes into GF recipients and associated transmission of the host phenotype. For example, several experimentally generated inflammatory conditions, such as obesity, diabetes, heart disease, autoimmune disorders, and cancer, have been shown to be transmissible from diseased donors into GF mice through either cohousing or direct transfer of the associated gut dysbiotic microbiota (29, 30). Mutations in the genes for leptin production have been targeted to develop the ob-/ob- or obese mouse genotypes that have often been used in such microbiome studies. Using a similar experimental protocol described in the our current report, transfer of the gut microbiota of ob-/ob- mice into wild-type GF animals led to the development of obesity in the recipients, with the resilience of the dysbiotic ob-/ob- microbiota appearing to be sufficient to enable transfer into a +/+ lean genetic background (31). In the current work, transfer of the dysbiotic microbiome from CXCR2^−/− donor mice into mice with a properly functioning neutrophil recruitment into the periodontal tissue is accompanied by transformation of the microbiota to that routinely present in wild-type animals. Thus, the neutrophil function affected by the mouse host genotype appears to be the deciding factor in this case for the development of dysbiosis and disease.

In summary, these data suggest that the genetic background of the host can have an impact on the composition, amount, and disease potential of the murine oral microbiome. The oral microbiota thereby demonstrates significant flexibility in composition and expression of virulence potential. Such flexibility is a significant factor in the bidirectional relationship between the microbiota and the host response (32), which is a distinguishing characteristic of current concepts of the susceptibility to and progression of periodontal disease.

MATERIAL AND METHODS

Animal experiments and ethics.

Animal experiments were conducted in accredited facilities in accordance with the UK Animals (Scientific Procedures) Act 1986 (Home Office license number 7006844). Specific-pathogen-free (SPF) CXCR2^−/− mice (strain C.129S2-B6) deficient in the receptor for the murine homologues of human interleukin-8 (CXCL1 and CXCL2) were derived from SPF BALB/c strain mice and purchased from The Jackson Laboratory (Bar Harbor, ME, USA). SPF BALB/c wild-type (BALB/c WT) mice were purchased from Charles River Laboratories, UK. Mice were maintained in individually ventilated cages (IVCs) at the animal care facilities of Queen Mary University of London (QMUL). Germfree (GF) C3H/Orl mice were bred and maintained under axenic conditions at the Royal Veterinary College, University of London, using founder GF C3H/Orl mating pairs purchased from Charles River Laboratories, UK, as previously described (1). The GF mice were bred and maintained in a separate location from the SPF colonies due to the highly specialized nature of their housing requirements.

Breeding and genotyping.

To unambiguously determine the influence of genotype on the oral microbiome and disease phenotype, adult CXCR2 ^−/− males deficient in the receptor for the murine homologues of interleukin-8 (CXCL1/CXCL2) were bred with SPF BALB/c WT females to produce the first-generation (F1) heterozygotes. After pregnancy was confirmed by daily examination of the females, the male CXCR2^−/− mice were separated from the pregnant females’ cages such that the resultant heterozygous CXCR2^+/− litters (F1) were only exposed to the maternal wild-type microbiome. Thereafter, the F1 mice were interbred (6 pairs) to produce segregating second generation (F2) litters of wild-type CXCR2^+/+ (F2 WT), heterozygous CXCR2^+/− (F2 HT), and homozygous CXCR2^−/− (F2 HM) (see Fig. S1 in the supplemental material).

The F2 mice were genotyped after weaning (3 weeks) by ear notches. Briefly, genomic DNA was isolated from each ear notch and analyzed by amplification using the wild-type and mutant primers oIMR0453 (GGTCGTACTGCGTATCCTGCCTCAG) (wild type F), oIMR0454 (TAGCCATGATCTTGAGAAGTCCATG) (wild type R) (33), oIMR6916 (CTTGGGTGGAGAGGCTATTC) (mutant F), and oIMR6917 (AGGTGAGATGACAGGAGATC) (mutant R) (The Jackson Laboratory). The conditions for thermocycling were the following: step 1, 94°C for 4 min; step 2, 94°C for 20 s, 64°C for 30 s, and 72°C for 35 s, for 12 cycles; step 3, 94°C for 20 s, 58°C for 30 s, and 72°C for 35 s, for 25 cycles; and step 4, 72°C for 2 min. Diagnostic mutant and wild-type bands were 750 bp and 900 bp in size, respectively, on 2.0% agarose gel electrophoresis (Fig. S1).

Microbiota transfer into germfree mice.

Eight-week-old GF mice were transferred under sterile conditions from the Royal Veterinary College, University of London, to the QMUL animal facility, and cocaging was set up immediately on arrival into IVCs. Oral swabs were collected from the GF mice before cocaging to confirm sterility and from the BALB/c WT and CXCR2^−/− mice to determine the microbiology at baseline. Two female donors and 4 recipient GF mice were cohoused in each cage at a ratio of 1:2 for 6 weeks. A total of 16 SPF and 16 GF mice were used in these experiments. Oral swabs were collected from all mice at weeks 1, 3, and 6 of cocaging. At the end of the experiment, all mice were euthanized by the CO₂ asphyxiation method prior to assessment of periodontal bone loss.

Cultural microbiology analyses.

The murine oral cavity was sampled for 30 s using sterile fine-tip rayon swabs (VWR) and placed in a tube containing 100 μl reduced John’s transport medium as previously described (12). Serial dilutions of the suspension were spread onto blood agar plates for aerobic and anaerobic (80% nitrogen, 10% hydrogen, and 10% carbon dioxide) growth for 48 h at 37°C. The predominant CFU of cultivatable bacteria were counted, purified by subculture, and identified by full-length 16S rRNA gene sequencing. On average, 4 to 6 different colony types were identified on each blood agar plate. Genomic DNA for each isolated bacterial strain was extracted by using a GenElute bacterial DNA kit (Sigma-Aldrich) by following the Gram-positive protocol according to the manufacturer’s instructions. The extracted DNA was used as a template for PCR, followed by full-length Sanger sequencing of the 16S rRNA gene and species-level identification based on the murine oral microbiome database, as described previously (12).

Immunohistochemistry of neutrophils and IL-8.

All mice were sacrificed at the age of 16 weeks by carbon dioxide asphyxiation. Three to 5 mice were selected in each genotype group, and their mandibles and maxillae were dissected and fixed in 4% (wt/vol) paraformaldehyde solution at 4°C for 3 days. Tissues were processed according to the standard histological procedures and embedded in paraffin. Each molar was sectioned in frontal buccolingual orientation using a microtome (5 mm) for approximately 100 serial sections and mounted on charged glass slides. Every tenth section was stained by hematoxylin and eosin, such that 2 to 3 sections were available for each mouse, and in each section the neutrophils were counted across the slide. The mean total neutrophil count per section was then calculated for each genotype. Immunohistochemistry (IHC) was carried out by following previously published methodology (16). Briefly, neutrophils were detected and counted using an anti-neutrophil elastase primary antibody (Abcam) and anti-rabbit secondary antibody (Abcam). CXCL2 (murine homologue of human IL-8) was detected using anti-CXCL2 antibody (Abcam) with biotinylated secondary anti-rabbit antibody (Abcam). The slides were developed using a peroxidase substrate DAB kit (Vector Laboratories). The microscopy images were analyzed using the ImageJ software (NIH, USA), and scale bars were set to the global calibration setting.

Periodontal bone loss determination.

Mice were euthanized as described above, and mandible and maxillae were dissected, defleshed, and assessed under a Stemi SVII dissecting microscope (Zeiss), and the captured images were analyzed by ImageJ software (NIH, USA). The distance from the cemantoenamel junction (CEJ) to the alveolar bone crest (ABC) was measured on 14 predetermined points on the buccal surface of the maxillary molars. To calculate bone loss for the 14 sites, total CEJ-ABC distance from each mouse was subtracted from the mean CEJ-ABC distance of the control mice (34). The results were expressed in millimeters, and negative values indicate bone loss relative to the controls.

Statistical analyses.

Bone levels between different experimental groups were compared using one-way analysis of variance (ANOVA) and unpaired Student's t test between 2 comparison groups. Significance was expressed at a P value of <0.05. The statistical analyses and graphical visualization of the bone levels were performed using GraphPad Prism 7 (GraphPad Software Inc.). The differences in the composition of the overall microbiome between the treatment and control groups at each time point in each experiment were assessed by PERMANOVA using the Adonis function in the Vegan package in R (35). Differences in the number of CFU of individual bacterial genera between treatment and control groups were assessed by Welch’s t test using GraphPad Prism 7.