Edinburgh Research Explorer A new model for NTHi middle ear infection in the Junbo mutant mouse

Acute otitis media, inflammation of the middle ear, is the most common bacterial infection in children and as a consequence is the most common reason antimicrobial prescription to this age group. There is currently no effective vaccine for the principal pathogen involved, non-typeable Haemophilus influenzae (NTHi). The most frequently used and widely accepted experimental animal model of middle ear infection is in chinchillas, but mice and gerbils have also been used. We have established a robust model of middle ear infection by NTHi in the Junbo mouse, a mutant mouse line that spontaneously develops chronic middle ear inflammation under specific pathogen free conditions. The heterozygote Junbo mouse ( Jbo/+ ) bears a mutation in a gene (Evi1, also known as Mecom) that plays a role in host innate immune regulation; pre-existing middle ear inflammation promotes NTHi middle ear infection. A single intranasal inoculation with NTHi produces high rates (up to 90%) of middle ear infection and bacterial titers (10 4 to 10 5 CFU/  l) in bulla fluids. Bacteria are cleared from the majority of middle ears between day 21 and 35 post-inoculation but remain in approximately 20% of middle ears at least up to day 56 post-infection. The expression of TLR-dependent response cytokine genes is elevated in the middle ear of the Jbo /+ mouse following NTHi infection. The translational potential of the Junbo model for studying antimicrobial intervention regimens was shown using a 3-day course of Azithromycin to clear NTHi infection, and its potential use in vaccine development studies by demonstrating protection in mice immunized with killed homologous, but not heterologous, NTHi bacteria.

The chinchilla otitis media model was first developed to study pneumococcus infection (Giebink et al. 1976). It has subsequently become the most frequently used model for NTHi infection studies, vaccine development, and is considered to be a robust, reproducible model for polymicrobial infections (Bakaletz, 2009). Following direct inoculation of the chinchilla bulla, NTHi can form a biofilm that promotes bacterial survival against the host response and treatment (Jurcisek and Bakaletz, 2007;Brockson et al., 2014).
In addition to the chinchilla, there is increasing use of inbred strains of mice and mouse mutants in OM research (Ryan et al., 2006;Metzger, 2008, Hernandez et al., 2015). In chinchillas and mice, direct injection of bacteria into the middle ear (ME) bulla is efficient and allows the dose to be controlled, but intranasal (IN) inoculation that mimics natural ascending Eustachian tube (ET) infection produces only sporadic ME infection in mice (Ryan et al., 2006) and no ME infection in chinchillas (Bakaletz, 2009). However, IN inoculation of the chinchilla produces sustained infection of the ME when it is used in conjunction with barotrauma (lowering pressure in the ME) (Giebink et al., 1979) or following virus infection (Giebink, 1981). In mice, IN  Mice with mutations in TGF- signaling pathway genes that modulate proinflammatory responses or mutations that lead to ET malformation are predisposed to develop OM spontaneously without the need for bacterial challenge and a number of these genes are also implicated in modulating susceptibility to human OM (review by Rye et al., 2010).
In this study we focus on NTHi induced ME infection in the Junbo mouse, a mutant mouse line that spontaneously develops chronic ME inflammation under specific pathogen free (SPF) conditions. The heterozygote Junbo mouse (Jbo/+) bears an Asn763Ile mutation in the gene encoding the transcription factor Evi1, also known as Mecom (Parkinson et al., 2006). One mechanism that may underlie the predisposition to OM in Jbo/+ mice is that Evi1 is a negative regulator of NFkB and the loss of function Evi1 Junbo mutation exacerbates NTHi induced inflammation in the lung (Xu et al., 2011).
We hypothesized that the ME inflammation in Jbo/+ mice could provide a niche in which, after IN challenge, bacteria would establish infection following contiguous spread along the ET. As a proof of this concept and its utility as a validated animal model of otitis media, we used the human commensal pathogen NTHi to establish ME infection. We have characterized the dynamics and host responses to NP colonization and OM, using multiple, genetically distinct NTHi strains. Our data demonstrate the utility of the Jbo/+ model for testing immunization and antibiotics strategies aimed at the prevention or treatment of NTHi infection of the ME.

NTHi inoculated intranasally infect the ME of Jbo/+ mice for at least 56 days
We have successfully established a robust model of ME infection using a The ME infection rates (Fig. 1B) and titers (Fig. 1D) 2C). ME titers peaked at day 10 and remained at ~10 4 -10 5 CFU/l up to day 56 (Fig. 2E), the last period sampled.
There was a strong positive association (P=2. 62E-16) between NTHi infection in the ME and NP (Table S1) and ME/NP co-infection declines with time (Fig. S1). The recovery of NTHi by NP washing is only semi-quantitative and the counts were generally low, 10 1.1 CFU (10 0.9 -10 1.1 95% CI, n=86 in 50 l sample of 200 l wash volume).
The histology of the NTHi infected middle ear was examined in 12-week-old Jbo/+ mice at day 7 post IN inoculation. To maintain the anatomical integrity of the bulla contents the tympanic membrane (TM) was not opened. A necrotic caseous core of neutrophils is surrounded by viable and apoptotic neutrophils (cleaved caspase 3 positive) and an outer, variably thick, band of foamy macrophages (F 4/80 positive) ( Fig. 3A-E). There were variable amounts of amorphous extracellular chromatin within the caseous areas (Fig. 3F).
The larger accumulations were histone 3 negative, but smaller granular aggregates were histone 3 positive (Fig. 3G). Neutrophil leukocytes were present in the ET lumen and ET mucosa at its junction with the NP and in the ET lumen where it opens into the ME (Fig. 3H

Population dynamics of NTHi infection
When two NTHi strains able to be monitored independently, 162lux and 375, were coinoculated into 8-10 week old Jbo/+ mice, 83% of ME gave rise to a mixed culture of both strains at day 1 but 71% of ME gave a monoculture of one or the other, but not both, strains at day 7. In this period ME infection rates remained high, ME titers of each strain were comparable in co-infections, but NTHi 162lux monocultures predominated by day 7 (Fig. 5).
Of interest, in at least one mouse the right and left ME were infected by monoculture of the alternative NTHi strains indicating that each middle ear can operate as a separate compartment for growth and selection within the same animal.

Jbo/+ mouse produces an innate immune response in the ME to NTHi infection
The host innate immune response was assessed in the ME of 8-10 week old Jbo/+ mice following NTHi infection using 8-10 week old Jbo/+ GF mice as a sterile inflammatory baseline control. We chose to study expression of TLR response cytokine genes that are

Azithromycin treatment eliminates NTHi ME infection
Antibiotic resistance in the clinical setting is on the increase for commensal pathogens (Benninger, 2008); this has the potential to impact on treatment regimens for disease caused by NTHi, including OM. To make a preliminary assessment of the Jbo/+ mouse model for One day post-inoculation, the ME infection rate (63%) and titers (10 4.2 CFU/l, 10 3.4 -10 4.9 95% CI, n=15) were similar to those seen following NTHi challenge. However on day 2, unlike with NTHi, there were unexpected deaths in 5 of 15 mice with this relatively low dose, and a small volume that would not be expected to infect the lower airway. The experiment was immediately terminated and retro-orbital blood samples from 3 of 10 survivors indicated bacteremia. In the surviving mice the ME infection rate was 50% and titers of 10 4.5 CFU/l (10 2.8 -10 6.2 95% CI, n=10) were attained. Thus, the Jbo/+ infection model can support ME pathogens other than NTHi but for the pneumococcus this needs to be further explored using appropriately attenuated strains with reduced virulence.

DISCUSSION
A recent review of the use of animals in OM research concluded there is a need to improve our current animal models, develop new ones and use a diversity of species to ensure that differences between any one species and humans do not bias our data (Li et al., 2013). In this work we show that Junbo, a mutant mouse line that develops OM spontaneously, has significant potential for ME infection studies as it can be reproducibly infected with a single IN inoculum of NTHi. NTHi rapidly ascends the ET and colonizes the ME. NTHi infection is detected in bulla fluids at 1 hr, titers peaked between day 7-14 and 20% of individual ME were infected up to day 56. Mouse ME bacterial titers (10 4 -10 5 CFU/l) are comparable to those reported in direct inoculation of the chinchilla ME (10 6 -10 7 CFU/ml) albeit that the volume of material recovered from the mouse is substantially less. Infection in the chinchilla was cleared by day 35 using the same NTHi strain set utilised in our study . The strong association between the incidence of ME and NP infection, the high number of NTHi in the ME relative to NP and exudate along the length of the ET suggests the ME can act as a reservoir for NP re-infection. The implication is that the ME compartment is not an evolutionary dead end for NTHi. The ME niche will impose strong re-sampling the same animal, and therefore represents welfare refinement.

Ethics statement
Full details of these studies were reviewed and approved by MRC Harwell ethical review committee. The humane care and use of mice in this study was carried out under the authority of the appropriate UK Home Office Project License.

Mouse strains
The majority of experiments used Junbo mice that were congenic on a C3H/HeH genetic background (Parkinson et al., 2006)

Histology, immunohistochemistry and in situ hybridization
Middle ear histology in GF and SPF mice was assessed in Haematoxylin and Eosin stained sections of 3 m wax sections as previously described (Cheeseman et al., 2011).
To examine the histology of NTHi infected bullae by immunohistochemistry To maximize RNA integrity for in situ work, a band saw was used to isolate the bullae from fixed heads and EDTA decalcification achieved in 48 hours.
For immunohistochemistry 4 m thick wax sections were cut onto electrostatically charged slides and dried overnight at 37 o C before a final drying at 60 o C for 25 minutes.
Sections were de-waxed in xylene, hydrated through ethanol and washed three times in Tris Buffer. Endogenous peroxidase was blocked using Dako REAL peroxidase blocker (S2023) for 10 minutes following antibody incubations.
Antigen retrieval was performed using Dako proteinase K (S3020) for 20 minutes at room temperature.  numbers recovered, the NP was sampled first in one cohort of n=15 mice; the infection rate was found to be not significantly different (67% versus 80% P=0.6816, Fisher exact test n=15 mice per group) and the titers comparable when the NP is sampled either before or after the ME (10 1.1 and 10 1.6 respectively).
NTHi culture of ME bulla fluids and NP wash.
Bulla fluids and NP washes in PBS were mixed by three 10 sec bursts on a vortex mixer then 10-fold dilutions (10 -1 , 10 -2 ) were made in PBS. 50 l of each ME dilution or NP wash was spread on a BHI-Lev agar plate. The detection limit was 10 CFU/l for the primary bulla fluid preparation and 100 CFU/l for the 10 -1 dilution. The primary bulla fluid suspension was centrifuged at 13000xg for 3 minutes and the pellet frozen on dry ice then stored at - In experiments using antibiotic resistant strains, bulla samples were plated on media supplemented with 300 g/ml streptomycin or 30 g/ml kanamycin. The ME commensal bacteria were assessed using a non-selective plate in parallel. Plates were cultured overnight at 37 O C to calculate NTHi titers. Representative NTHi colonies were examined by phase contrast microscopy to confirm its small coccobacillus morphology.
In the co-infection experiment with NTHi 162lux and 375 (5 x 10 5 CFU each strain IN), a monoculture is defined as ≥10 colonies all of the same strain (either 162lux or 375) on the BHI primary culture plate.

Infection rates and bulla fluid titer
To compare ME infection rates we used an index of infected bulla fluids:

Bioluminescent imaging
Mice were inoculated with NTHi 162lux and bioluminescent signals from the head (oral cavity aspect of the palate after dissecting away the mandible) or, for dual infection studies, the proportion of NTHi colonies on the BHI ME culture plates were imaged using an IVIS Lumina II system (Perkin Elmer).

Statistical analysis
Log10 normalized NTHi titers were analyzed using t-tests (mouse strain susceptibility data, immunization data and antimicrobial data) or by one-way ANOVAs and Tukeys multiple  susceptibility to otitis media has been intensively studied in genetically altered mice using the wide range of reagents available for protein and gene expression analyses in this species.
Following intranasal challenge NTHi infection in wild-type mice resolves spontaneously within 7 days.

Results
We