Nasal Delivery of a Commensal Pasteurellaceae Species Inhibits Nontypeable Haemophilus influenzae Colonization and Delays Onset of Otitis Media in Mice

Nasopharyngeal colonization with nontypeable Haemophilus influenzae (NTHi) is a prerequisite for developing NTHi-associated infections, including otitis media. Therapies that block NTHi colonization may prevent disease development. We previously demonstrated that Haemophilus haemolyticus, a closely related human commensal, can inhibit NTHi colonization and infection of human respiratory epithelium in vitro. We have now assessed whether Muribacter muris (a rodent commensal from the same family) can prevent NTHi colonization and disease in vivo using a murine NTHi otitis media model.

common, especially in pediatric populations (4). NTHi can also cause a range of respiratory-related diseases, including middle ear infections (otitis media), sinusitis, conjunctivitis, and pharyngitis, as well as acute exacerbations in patients with chronic lung diseases and invasive diseases, such as meningitis and bacteremia (4,5). There are currently no licensed preventative therapies that specifically target NTHi colonization and/or disease (6).
The burden of NTHi-associated otitis media is high. Studies from North America, Europe, and Oceania have shown that NTHi accounts for 45 to 61% of all recorded otitis media cases (7)(8)(9)(10), and a systematic review of otitis media etiology from 1970 to 2014 found NTHi to be the predominant otopathogen (11). With over 700 million annual cases of acute otitis media throughout the world, ϳ31 million chronic infections, and ϳ21,000 deaths from otitis media complications every year (12), preventing NTHi otitis media would have a significant impact on reducing global morbidity. In addition, prevention of NTHi otitis media would have a significant impact on reducing antibiotic use, with otitis media being the main reason for antibiotic prescriptions in children (13).
Colonization of the nasopharynx with NTHi is a prerequisite for developing and transmitting disease, with early life colonization and increased NTHi density in the nasopharynx associated with the onset of otitis media and other respiratory infections (14)(15)(16)(17). Thus, preventing or eradicating NTHi colonization of the nasopharynx is an attractive target to stop progression to disease and minimize host-to-host transmission.
Microbial interference offers a potential solution for inhibiting NTHi colonization and preventing development of disease. This approach involves the use of commensal bacteria to compete with pathobionts for binding sites, nutrients, and space in order to beneficially alter the host microflora. Microbial interference is currently being investigated to combat a range of bacterial infections, including pneumococcal otitis media in children using a nasal alpha-hemolytic streptococcal probiotic spray (18), experimental meningococcal meningitis in mice using intranasal delivery of the closely related commensal Neisseria lactamica (19), and pneumococcal pneumonia in mice using intranasal delivery of Streptococcus mitis (20). From these and other studies (21,22), it appears that microbial interference requires the commensal and pathogenic species to belong to the same family of bacteria and be able to colonize the same niche.
We have previously demonstrated that the human respiratory tract commensal H. haemolyticus can be used to prevent NTHi infection of epithelial cells in vitro (23), reducing both NTHi attachment and invasion, indicating that microbial interference may occur between these two species. Some H. haemolyticus isolates have been found to produce a bacteriocin-like substance that specifically inhibits NTHi growth (24), further supporting a role for H. haemolyticus as a bacterial therapy to prevent NTHi disease. H. haemolyticus does not colonize mice (our unpublished data); therefore, we sought alternatives to further investigate microbial interference of NTHi in vivo. In this study, we have used Muribacter muris, a closely related rodent equivalent of H. haemolyticus from the Pasteurellaceae family (25), in a murine model of NTHi acute otitis media. We have assessed whether intranasal pretreatment of mice with M. muris can be used to prevent NTHi colonization and development of disease.

RESULTS
Intranasal treatment of mice with M. muris can temporarily reduce NTHi colonization and prevent development of NTHi otitis media. Intranasal administration of 5 ϫ 10 7 CFU of M. muris to mice prior to challenge with influenza A/Memphis/ 1/71 H3N2 virus (IAV) and NTHi (M. muris ϩ IAV ϩ NTHi group) reduced the NTHi density recovered from the nose of mice on day 5 from a median log 4.94 CFU/ml (95% confidence interval [CI] of median 4.36 to 5.50) to a median log 3.97 (95% CI, 3.88 to 4.08) when compared with that of no pretreatment (IAV ϩ NTHi group; P Ͻ 0.001) (Fig.  1A). M. muris pretreatment also prevented development of NTHi otitis media by day 5, with only 1 out of 12 (8%) M. muris-treated mice developing NTHi otitis media compared with 53% (8/15) of mice given no M. muris pretreatment (P ϭ 0.019) (Fig. 1B).  1C). There was no difference in the proportion of mice that had otitis media by day 7, with 4 out of 12 (33%) M. muris-treated mice developing NTHi otitis media compared to 8 out of 21 (38%) mice with no M. muris pretreatment (P Ͼ 0.999) (Fig. 1D). The median log density of NTHi recovered from the middle ear on day 7 was the same for each group at log 1.92 (M. muris treated ϭ 1.92 [95% CI, 1.92 to 4.35] and untreated ϭ 1.92 [95% CI, 1.92 to 3.00]; P ϭ 0.981), which is at the limit of quantification.
Intranasal pretreatment with M. muris temporarily reduces inflammatory mediator levels in the middle ear tissue of mice in the NTHi otitis media model. On day 5, IL-6 was elevated in the middle ears of mice receiving NTHi, either alone or in combination with IAV (IAV ϩ NTHi) and M. muris (M. muris ϩ IAV ϩ NTHi) (Fig. 2B). There was no difference between IL-6 levels in the ears of mice pretreated with M. muris (M. muris ϩ IAV ϩ NTHi) compared to untreated mice (IAV ϩ NTHi). KC levels were also elevated in the ears of mice that received NTHi on day 5; however, mice that were pretreated with M. muris (M. muris ϩ IAV ϩ NTHi) had significantly reduced KC titers compared with those of mice that received NTHi challenge alone (P ϭ 0.0008) or IAV ϩ NTHi challenge (P ϭ 0.0004) (Fig. 2B). By day 7, the KC titer returned to baseline value in all of the groups. Pretreatment of mice with M. muris (M. muris ϩ IAV ϩ NTHi) prevented the elevated IL-1␤ response observed in the ears of mice in the NTHi otitis media group (IAV ϩ NTHi) on day 5 (P ϭ 0.0427); however, this was reversed by day 7 where mice receiving M. muris pretreatment had higher median IL-1␤ titers in their ears than untreated mice in the otitis media model (P ϭ 0.0125) (Fig. 2B). IFN-␥ and IL-10 levels were either very low or not detected in the middle ear tissue, with no difference between median titers for any groups.
Mice that were pretreated with M. muris had better clinical outcomes in the NTHi otitis media model. Administration of a single intranasal dose of M. muris reduced disease symptoms over the 7-day time frame in the NTHi otitis media model, with lower clinical scores (Fig. 3A) and less weight loss (Fig. 3B) than mice with no M. muris pretreatment (IAV ϩ NTHi). On day 6, the clinical score of mice in the M. muris-pretreated group (M. muris ϩ IAV ϩ NTHi) was significantly lower than those that did not receive M. muris pretreatment (IAV ϩ NTHi) (P Ͻ 0.05) (Fig. 3A). M. muris administration alone had no impact on the condition of the mice, with a mean clinical score of 0 to 1 over the 7 days (see Fig. S1A in the supplemental material). M. muris only treatment also had no significant impact on weight loss (see Fig. S1B). On the day after IAV challenge (day 1), the M. muris-treated group (M. muris ϩ IAV ϩ NTHi) lost 5% of body weight compared with that of the IAV-treated group (IAV ϩ NTHi) (P Ͻ 0.01) (Fig.  3B). Similar weight loss also occurred following IAV challenge in the groups that did not receive M. muris pretreatment, but this was observed on day 2 after IAV challenge rather than on day 1 (see Fig. 3B for IAV ϩ NTHi; see also Fig. S1B for IAV only and IAV ϩ NTHi). However, upon NTHi challenge, the mice that were pretreated with M. muris (M. muris ϩ IAV ϩ NTHi) had significantly less weight loss than those that had no M. muris pretreatment (IAV ϩ NTHi), and this was sustained from day 5 until day 6 (P Ͻ 0.01) (Fig. 3B).

M. muris challenge induced a short-lived inflammatory local response.
Intranasal challenge with M. muris alone induced an inflammatory response in the upper respiratory tract on the day after challenge (day 0, as M. muris was administered on day Ϫ1), with elevated IL-6 and KC in the nasal washes (P Ͼ 0.0001 and P ϭ 0.0020) and middle ear tissue (P ϭ 0.0027 and P ϭ 0.0007) compared with those of naive mice (Fig.  4). The IL-6 and KC titers returned to baseline levels by day 3 (4 days after M. muris challenge).
To estimate the duration of M. muris colonization following M. muris treatment (and whether M. muris entered the middle ear), all colonies that appeared M. muris-like on nonselective chocolate agar plates were counted in nasal washes and middle ear tissue (Fig. 5). It is important to note that asymptomatic colonization with M. muris as part of the normal microbiome was present, as indicated by the M. muris-like counts in specimens collected from naive mice (Fig. 5). While the median M. muris-like counts

DISCUSSION
A noninvasive bacterial therapy to prevent NTHi colonization of the upper respiratory tract is a particularly attractive approach to reduce the burden of otitis media and resulting antibiotic use. Prevention of NTHi otitis media could reduce the global burden of otitis media by approximately 50% (26), resulting in an estimated 350 million fewer cases of acute otitis media episodes per year. This would have a major impact on preventing chronic and recurrent otitis media to improve health and educational outcomes for children throughout the world. In this study, we have demonstrated that a commensal Pasteurellaceae species can be applied intranasally to reduce NTHi colonization and delay onset of NTHi otitis media in mice. This builds upon our previous findings that H. haemolyticus can prevent NTHi infection of human respiratory epithelial cells in vitro (23). Nasal probiotic therapies are being investigated to protect against pneumococcal otitis media using commensal alpha-hemolytic streptococci (18,(27)(28)(29)(30)(31). In a prospective, randomized, double-blind, placebo-controlled study, 100 Italian children aged 1 to 5 years with histories of recurrent acute otitis media (AOM) were randomized 1:1 to intranasally receive Streptococcus salivarius 24SMB or saline placebo twice daily for 5 days each month for 3 consecutive months (27). Children that were successfully colonized with S. salivarius 24SMB were protected from developing otitis media (13.6% versus 42.8%; P ϭ 0.03). In our study, only one application of M. muris was administered, and the beneficial effects were transient. It is likely that a nasal commensal therapy for NTHi otitis media will also require multiple applications to enhance duration of protection from NTHi colonization and disease. We chose not to decolonize mice with antibiotics prior to this study for removal of endogenous Pasteurellaceae species, as the intended application of this bacterial therapy is that it is suitable for all children regardless of the profile of their nasal microbiota.
A human infection study has been conducted with NTHi, where 15 healthy adult volunteers were intranasally challenged with a single ascending dose of NTHi to establish colonization (32). All participants experienced mild clinical symptoms, including rhinorrhea, pharyngitis, and/or headaches. We propose that pretreatment with an intranasal application of a commensal Pasteurellaceae species, e.g., H. haemolyticus (the human equivalent of M. muris), could prevent development of the NTHi-associated symptoms in this human challenge model. The potential impact of a therapy that prevents NTHi colonization is not limited to otitis media, as prevention of NTHi colonization could also prevent other NTHi-associated diseases such as pneumonia, the major global killer of children under 5 years of age (33), and chronic obstructive pulmonary disease, which is the 3rd most common cause of adult mortality (34).
Inflammation in the middle ear is a major feature of otitis media (35,36). Our observation that mice pretreated with a single dose of M. muris have lower levels of inflammatory mediators in their upper respiratory tract than mice with no M. muris pretreatment in the otitis media model further demonstrates the potential use of commensal Pasteurellaceae species to prevent NTHi disease. The inflammatory response of the control groups, particularly that seen on initial delivery of M. muris, may offer insight into the mechanism behind M. muris prevention of NTHi colonization and otitis media. The initial increases in IL-6 and KC on day 0 (24 h after M. muris treatment) in both the middle ear and nasal washes of the mice suggest that despite not seeing an increase in the commensal density, an innate immune response was elicited. While inflammation was short-lived, it indicates that immune modulation, rather than physical competition with NTHi, may play a role in microbial interference with M. muris protection against NTHi colonization and otitis media. This is further supported by the observation that direct bacterial-bacterial interference was not apparent in vitro.
Intranasal influenza virus challenge is essential for ensuring NTHi colonization and driving development of NTHi otitis media in the murine ascension model (37), presumably from NTHi taking advantage of the inflamed respiratory tract of influenza viruschallenged mice. Pretreatment with M. muris significantly reduced the nasal inflammatory response to influenza virus, suggesting that M. muris alters the host response to viral infection. It is likely that M. muris dampening of the innate inflammatory response to influenza virus is associated with reducing the ability of NTHi to colonize and go on to cause otitis media. The improved clinical scores of M. muris-pretreated mice are also likely to be a reflection of the lower amount of NTHi that was able to colonize the respiratory tract and, thus, ascend into the middle ear to cause infection and inflammation. A potential explanation for the earlier weight loss response to IAV exposure in M. muris-pretreated mice than in mice that were not pretreated with M. muris may be the accumulative stress of repeat anesthesia in the M. muris ϩ IAV group compared with that of the IAV groups. Of interest is the dampened nasal IL-6 and KC response and middle ear IL-1␤ response to challenge with NTHi alone, often at or below controls including naive mice. This is a phenomenon that we have previously observed in cell culture experiments where NTHi "flies under the radar" of the host innate immune response and appears to suppress IL-6 and IL-8 production rather than elicit it (23).
In summary, we have demonstrated that intranasal treatment with the commensal Pasteurellaceae species M. muris can reduce NTHi colonization and prevent development of NTHi otitis media in vivo. This work supports further investigation into the potential use of a commensal Pasteurellaceae species to prevent NTHi colonization and disease in humans.

MATERIALS AND METHODS
The sources of the microorganisms used in this study are detailed in Table 1. Bacterial inoculum. Standard inoculum of mid-log phase NTHi 2866 Spec r was prepared in 1-ml aliquots and stored as previously described (38), with the exception that 0.1 mg/ml spectinomycin was added to the culture medium. M. muris was isolated from the respiratory tract of a mouse in our animal facility by plating a nasal wash onto chocolate agar plates and selecting Haemophilus-like colonies. Species identity of a selected isolate (M. muris TKI) was confirmed by sequencing of the 16S gene at the Australian Genome Research Facility in Melbourne, Australia. Standard inoculum of M. muris was prepared by picking 3 colonies from an overnight chocolate agar plate and seeding into 15 ml culture medium (heart infusion broth supplemented with 44 ml/liter glycerol, 30 mg/liter hemin, and 10 mg/liter NAD). The culture was incubated at 37°C with shaking to mid-log phase (optical density at 600 nm [OD 600 ] was between 0.55 and 0.65), then 20% heat-inactivated fetal calf serum was added, and 1-ml single-use aliquots of M. muris TKI were prepared and stored in cryovials at -80°C (as for the NTHi inoculum). The number of CFU for each inoculum was determined after at least 24 h storage at -80°C as previously described (38). Viability of the frozen inoculum was assessed over the study period and found to remain stable for both species for at least 12 months.
Animals. All animal experiments were approved by the Telethon Kids Institute Animal Ethics Committee, Perth, Australia (number A302). Female specific-pathogen-free BALB/c mice were obtained from the Animal Resources Centre (Perth, Australia). Experiments were conducted in sets of 12 to 15 mice, ensuring representation from each group at each time point.
NTHi otitis media model. Viral coinfection is required for reliable development of NTHi otitis media infection in mice using the ascension model (37). Briefly, 6-to 8-week-old female BALB/c mice were inoculated intranasally with 1 ϫ 10 4.5 PFU IAV in a volume of 10 l. At 72 h after IAV challenge, mice were intranasally administered 5 ϫ 10 7 CFU of NTHi R2866 Spec r in 10 l of phosphate-buffered saline (PBS). For the groups pretreated with M. muris, mice received intranasal inoculation of 5 ϫ 10 7 CFU of M. muris TKI on day Ϫ1. Mice were monitored, weighed, and clinically assessed each day. Clinical disease scores were assessed as previously described (44) using a scale ranging from 0 to 20 according to the following criteria: score 0 ϭ normal appearance, healthy, and active; score 1 to 5 ϭ barely ruffled fur, mildly/ intermittent hunched appearance, and otherwise healthy; score 6 to 10 ϭ moderately ruffled fur,  Table 2. Specimen collection and processing. Nasal washes and middle ear bullae were collected immediately postmortem and stored on ice. The nasal washes were conducted by lavaging the nares with 0.1 ml PBS. Middle ear tissue (combined from both ears of a mouse) was mechanically homogenized in 0.5 ml PBS using hand-held sterile plastic pestles (Interpath) until all tissue was disrupted. Nasal washes and middle ear tissue homogenates were serially diluted in PBS and spotted onto chocolate agar plates with and without an overlay of 200 l of 10 mg/ml spectinomycin (to select for the NTHi Spec r strain). Remaining middle ear homogenate and nasal washes were centrifuged at 13,000 rpm for 10 min at 4°C to remove cell debris. The supernatants were filtered using 0.2-m syringe filters and stored in aliquots at -80°C for subsequent measurement of inflammatory mediators.
Measurement of inflammatory mediators in nasal washes and middle ear tissue. Stored supernatants from nasal washes and middle ear tissue homogenates were tested using a Bio-Rad express assay 5-plex murine cytokine/chemokine magnetic bioplex kit to measure IFN-␥, IL-1␤, IL-6, KC, and IL-10 on the BioPlex 2000 (Bio-Rad) according to the manufacturer's instructions. Where cytokine titers were below the limit of detection (LOD), half of the value of the lowest standard was assigned to permit statistical analysis. The LOD of each cytokine was as follows: IFN-␥ ϭ 0.94 pg/ml, IL-1␤ ϭ 1.79 pg/ml, IL-6 ϭ 0.68 pg/ml, KC ϭ 1.51 pg/ml, and IL-10 ϭ 4.62 pg/ml. Assessment of bacterial interference. The following three in vitro methods were used to assess bacterial-bacterial interference between M. muris and NTHi using previously described methods: direct coculture in broth (45), spot agar test (45), and a well-diffusion assay (24).
Statistical analysis. Mann-Whitney U tests were applied to nonparametric data (bacterial counts, CFU/ml; cytokine levels, pg/ml), with a P value of Ͻ0.05 considered significant. Fisher's exact testing was used for categorical analyses (development of otitis media). Mean clinical scores and percent weight loss were compared by Student's t test.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.2 MB.