Cellular content plays a crucial role in Non‐typeable Haemophilus influenzae infection of preinflamed Junbo mouse middle ear

Abstract Non‐typeable Haemophilus influenzae (NTHi) is a major pathogen causing acute otitis media (AOM). The relationship between the cellular content of the middle ear fluid (MEF) during AOM and infection of NTHi is poorly understood. Using the Junbo mouse, a characterised NTHi infection model, we analysed the cellular content of MEF and correlated the data with NTHi titres. The MEF of the Junbo mouse was heterogeneous between ears and was graded from 1 to 5; 1 being highly serous/clear and 5 being heavily viscous/opaque. At seven‐day post‐intranasal inoculation, NTHi was not found in grade‐1 or 2 fluids, and the proportion of MEF that supported NTHi increased with the grade. Analyses by flow cytometry indicated that the cellular content was highest in grade‐4 and 5 fluids, with a greater proportion of necrotic cells and a low‐live cell count. NTHi infection of the middle ear increased the cell count and led to infiltration of immune cells and changes in the cytokine and chemokine levels. Following NTHi inoculation, high‐grade infected MEFs had greater neutrophil infiltration whereas monocyte infiltration was significantly higher in serous noninfected low‐grade fluids. These data underline a role for immune cells, specifically monocytes and neutrophils, and cell necrosis in NTHi infection of the Junbo mouse middle ear.


| INTRODUCTION
Otitis media (OM) is characterised by inflammation of the middle ear and is the most common cause of hearing impairment in children (MacArthur, Hausman, Kempton, Lighthall, & Trune, 2012). Based on variables such as the chronicity of inflammation and the state of middle ear (bulla) fluid, OM is classified into different categories (Myburgh, van Zijl, Swanepoel, Hellström, & Laurent, 2016). A common form is acute suppurative OM, typically caused by bacterial infection; the most common associated bacteria are nontypeable Haemophilus influenzae (NTHi), Streptococcus pneumoniae, and Moraxella catarrhalis (Hood et al., 2016). Interplay between various immune components in the middle-ear cavity and the bulla fluid play a significant part in the onset of disease and subsequent eradication of bacteria (Hernandez et al., 2015). The middle ear cavity is lined with a mucosal membrane containing secretory glands that produce major components of both mucoid and serous fluids (Orita et al., 2002).
Serous fluid is considered to aid mucociliary clearance, and mucoid material provides lubrication to the epithelial cell boundary, thus, acting as the first line of innate immune defence. The anatomy of the middle ear also plays a significant part in the initial clearance of infect-OM reoccurs in 20-30% of the paediatric population (Roos, Hakansson, & Holm, 2001;Veenhoven et al., 2003), and the pathophysiology behind this is still not clearly understood. Persistence of bacteria/ virus/bacterial material in the middle ear might initiate the inflammatory response that leads to recurrence of OM, and the cellular content of the bulla fluid is likely a significant factor that influences clearance (Forséni, et al., 1999). The chinchilla model for OM has indicated a significant role for NTHi biofilm in the middle ear in the modulation of the host cellular response (Sato, Liebeler, Quartey, Le, & Giebink, 1999). Experimental induction of OM in rat and mouse by direct inoculation of bacteria into the noninflamed middle ear causes a rapid cellular response typified by a large increase in granulocytes, macrophages, and a mild increase in the lymphocyte populations (Forséni, et al., 1999;Hernandez et al., 2015).
The pathological chronicity of OM increases during the long-lasting or reoccurring infection in the middle ear. The dynamics of the cellular response against the infecting bacteria is still unclear. In the current study, using a Junbo mouse model of OM, we analysed the correlation between the cellular content of the middle-ear fluid against the ability of NTHi to infect the preinflamed middle ear. The Junbo mouse spontaneously develops chronic middle-ear inflammation under specific pathogen-free conditions (Parkinson et al., 2006). Earlier studies have shown that intranasal inoculation of the Junbo mouse with NTHi typically results in 80% of the middle ears being infected (Hood et al., 2016).
Thus, 20% of Junbo middle ears are not subsequently infected by NTHi.
We show that the type of immune cell and its functional state both contribute to the NTHi infection observed in the inflamed middle ear of Junbo mice.

| Variation in the bulla fluid influences NTHi infection
The Junbo mouse spontaneously develops fluid in the middle ear (Parkinson et al., 2006) and a single intranasal inoculation with NTHi results in infection in a majority of middle ears; ears with no bulla fluid do not support NTHi growth (Hood et al., 2016). Around 80% of the middle ears with fluid in the Junbo mouse are NTHi culture positive at 7 days postinoculation with NTHi 162 sr. Even although these mice were genetically similar and were bred under specific pathogen free conditions, 20% of the middle ears were not infected with the NTHi.
To investigate this variability, we sought to understand the correlation between bulla fluid and NTHi infection load. An initial general observation was the variation in the quality of the bulla fluid, from being serous to highly mucoid and pus like; similar observations are noted in human patients (Carrie, Hutton, Birchall, Green, & Pearson, 1992). On this basis, the Junbo bulla fluid was graded from 1 to 5, wherein 1 being the most serous/clear and 5 being the most viscous/opaque. Middle-ear fluids, which were deemed intermediate between two grades, or contained mixed aliquots of two grades of fluid, were described as intermediate grades 2/3, 3/4, and 4/5. Across 153 middle ears that contained fluid and were sampled, nearly 5% of the bulla fluids were of serous type grade 2 or 2/3, 41% were more dense/slightly viscous grade 3 or 3/4, 51% were pus filled/viscous grade 4 or 4/5, and 3% were very viscous/opaque grade 5 (Figure 1a). Grade 1 highly serous/clear fluids were rarely observed, thus, not included in current analyses.
When overlaying this middle-ear fluid characterisation with the NTHi infection data, 79% of all middle ears were infected with NTHi, and 21% of middle-ear fluids were not infected. A clear correlation between NTHi infection and viscosity of the fluid was observed ( Figure 1a). The Junbo mouse middle ears with grade 2 and 2/3 serous fluid had no NTHi infection (0%) present at seven-day postinoculation, whereas 46% of grade 3, 79% of grade 3/4, 100% of grade 4, and 92% of grade 4/5 fluids were infected. All middle ears with highly viscous grade-5 fluids cultured NTHi at seven-day postinoculation; grade 5 fluid was generally not found in noninoculated mice. The NTHi titres also correlated with the grade of middle-ear fluid ( Figure 1b); the more serous fluids had low-bacterial titres and as the viscosity increased so did the average NTHi titre. Junbo mouse middle ears with the most serous grade-1 fluid were found on rare occasions, and no NTHi were present at seven-day postinoculation. Thus, a direct correlation has been shown between the nature of the bulla fluid and the NTHi bacterial load in the middle ear of the Junbo mouse.
Having established the direct relationship between the viscosity/ opacity of the Junbo bulla fluids and the NTHi bacterial load in the middle ear, next, we examined the total cell count in the bulla fluids as this would seem a logical factor to contribute to the different fluid types and associated survival of NTHi. Significant difference in total cell count was observed in the different grades of fluid ( Figure 1c). The high-viscosity grade 4 and 4/5 fluids had a total cell count of 9.12 × 10 5 and 1.27 × 10 6 cells/μl respectively in noninoculated mice, which was significantly higher when compared with the serous grade 2, 2/3, and 3 fluids, with a total cell count of 1 × 10 5 , 3.14 × 10 5 , and 4.01 × 10 5 cells/μl, respectively. The difference in cell number for different grades of fluid was confirmed by giemsa staining (Figure 1d). To ascertain whether the presence of the NTHi alters the cell number, the cell count was performed on the bulla fluids obtained seven-day postinoculation with NTHi. The total cell count in the serous grade 2/3 fluid (negative for NTHi culture) post-NTHi inoculation was not significantly changed whereas the total cell count significantly increased in grade-4 fluid (positive for NTHi culture), attaining an average cell count of 1.89 × 10 6 cells/μl, consistent with an infiltration of cells in the presence of NTHi.

| NTHi infects Junbo mouse bulla fluids with necrotic cell population
It would appear somewhat counter-intuitive that despite the higher cell count in high-viscosity fluids, NTHi is able to exist effectively in these middle ears; we, thus, investigated the viability state of these cell popu-  A slight reduction in macrophage number was observed in NTHi+ve compared with NTHi−ve ear fluids, but this was not statistically significant.

| NTHi infection in preinflamed middle ear causes changes in cytokine and chemokine levels
The Junbo mouse, a chronic OM model, has a highly inflamed and hypoxic environment in the middle ear (Cheeseman et al., 2011). After intranasal inoculation of the Junbo mouse by NTHi, the cytokines present in the middle-ear fluid of NTHi+ve and NTHi−ve ear fluids were profiled at day 7 of postinoculation ( Figure 5). The levels of granulocyte colonystimulating factor (GCSF), granulocyte-macrophage colony-stimulating factor, and interleukins-2,3,4,5,6,9,10,12, and 13-did not significantly vary following NTHi infection in the Junbo mouse middle-ear fluid ( Figure 5a). Major Th cell-dependent cytokines, IL-17 and INF-gamma (Lucey, Clerici, & Shearer, 1996) (Forséni, et al., 1999;Sato et al., 1999). The Junbo mouse has a perturbed immune response due to the Evi1 mutation that affects NF-kB regulation (Li Konduru et al., 2017), which disturbs middle-ear homeostasis causing a highly inflamed middle-ear environment and OM. The efficient establishment of NTHi infection in this highly inflamed environment with high immune-cell content is somewhat counter-intuitive but mimics the condition found in patients suffering from long-term and recurrent AOM (Sharma & Pichichero, 2013). In  NTHi is reported to be internalised and induce apoptosis of type II alveolar epithelial cells (Goyal, Singh, Ray, Srinivasan, & Chakraborti, 2015). The observations in the current study are consistent with the same ability of NTHi to induce apoptosis being relevant in the middle ear of the Junbo mouse. In the normal inflammatory response, the clearance of these apoptotic and necrotic cells is essential to avoid the per- Numerous studies have focused on the changes in the cytokine and chemokine levels in the middle ear with respect to the onset and progression of OM (Juhn et al., 2008;Sato et al., 1999), and these signalling molecules influence the migration and functioning of the immune cells.
Using the Junbo mouse, we have for the first time analysed the changes in the major cytokines and chemokines in response to the presence of NTHi in an already highly inflamed middle ear. The analysis showed that VEGF-A was twofold elevated in middle-ear fluids infected with NTHi compared with those with no NTHi infection. VEGF-A is the major signalling molecule involved in vasodilation; it promotes cellular migration and is induced under hypoxic condition (Cheeseman et al., 2011). The hypoxic environment in the Junbo mouse middle ear plays a critical role in the chronicity of OM in this mouse (Cheeseman et al., 2011), and our current observation indicates its significance in NTHi infection. Another major molecule showing significant elevation was pro-inflammatory cytokineTNF-α; this plays a major role in stimulating neutrophil degranulation, respiratory burst, and release of platelet activating factor (PAF) (Ogura, Furukawa, Tada, Ikeda, & Yamashita, 2008). PAF induces middle-ear inflammation (Ogura et al., 2008) and is found in higher levels in pus-filled human middle-ear fluids (Tachibana et al., 1996). TNF-α was also significantly elevated in the chinchilla middle ear after pneumococcus inoculation but at a later time point (72 hrs) following infection (Sato et al., 1999). The Th1 dependent INF-γ and Th17 dependent IL- Chemokines like SCF and TPO were elevated in NTHi+ve middle-ear fluids; these not only induce inflammation but also regulate stem cell function. SCF plays a major role in mast cell degranulation (Lukacs et al., 1996) and the differentiation of CD34 + cells into mast cells in the upper respiratory tract (Da Silva & Frossard, 2005). Similarly, TPO plays a role in haematopoietic stem cell differentiation (Lupia, Goffi, Bosco, & Montrucchio, 2012a). Thus, these two chemokines play a significant role in the increase of inflamed state seen in NTHi-infected Junbo mouse middle ears. Another chemokine, CCL3, which inhibits the proliferation of immature progenitor cells but activates the proliferation of mature progenitor cells (Baba & Mukaida, 2014), was not potentiated in the NTHi infected ear fluids. The significance of CCL3 in protecting against NTHi-induced OM in mice is already known (Deniffel et al., 2017), and our data on CCL3 levels in both middle-ear fluids and serum suggests its role in inhibiting the infection of NTHi. Another important chemokine, CCL2, was elevated in the NTHi-infected mice.
CCL2 is known to contribute to inner ear inflammation stimulated by NTHi in the mouse (Woo, Pan, Oh, Lim, & Moon, 2010) and is an important mediator in the recruitment of monocytes; this consequently affects the cellular response (Qian et al., 2011;Raghu et al., 2017). Thus, in the Junbo mouse middle ear, CCL2 is also playing an important part in mediating the cellular response against NTHi.
The current study indicates that the cellular content is a significant factor in the variability observed in the middle-ear fluid in the Junbo mouse, and this affects the ability of NTHi to infect; the reason(s) underlying the variability in the ear fluids remain unclear. In humans also, the chronicity of OM is often associated with the differences observed in the ear fluids (Carrie et al., 1992). The Junbo mice used in our study were genetically identical and raised under SPF conditions, suggesting that inherent and environmental factors might not be significantly contributing to this variability. Examination for other bacteria in the non-NTHi inoculated Junbo mouse middle-ear fluid using different culture media and conditions showed low or no bacterial load. Inoculation of NTHi increased the load of other culturable bacteria in the middle-ear fluid but was not significantly variable across different grades of fluid and did not correlate with the infection of NTHi (data not shown). This suggests that NTHi was the major contributor to the cellular variation that we observed, but an influence of nonculturable bacteria cannot be ruled out; high-throughput microbiome analysis is currently being undertaken. One other noteworthy observation in our study was that even in the same mouse, dif-

| Junbo mouse infection with NTHi
The inoculation of the Junbo mice was carried out as described previously (Hood et al., 2016). Briefly, Junbo mice were inoculated intranasally under gas anaesthesia with 5 μl per nares of NTHi 162 sr cell suspension at a concentration of 1 × 10 8 cfu/ml in PBS-2% gelatine.

| Histology and immunohistochemistry
Middle-ear histology sections and Haematoxylin and Eosin staining of Junbo mice heads were carried out as previously described (Cheeseman et al., 2011). To examine the neutrophil and NTHi content by immunohistochemistry, rabbit antimyeloperoxidase (anti-MPO, Abcam™) and anti NTHi-162 antibodies were used. Anti-NTHi 162 Sr sera was raised in the rabbit (Covalab); this and control nonimmune (prebleed) serum were purified further by affinity chromatography using the Protein G Sepharose®, Fast Flow beads (Sigma) as per the manufacturer's instructions. The purified antibody was assessed by western blot and immunostaining. Unstained histology slides with 4-μm thick samples were dewaxed using xylene. Endogenous peroxidase was blocked using peroxidase blocker (Dako REAL S2023) for 10 min following antigen retrieval by two intervals of heating in a microwave for 7 min in distilled water. Secondary antibody binding was carried out using the rabbit VERTEX ABC kit. Visualisation was achieved using Dako (K3468) liquid DAB+ substrate chromogen system. Counterstaining was carried out using Harris Haematoxylin and mounting in Clearview mounting (Thermo Fisher Scientific). The slides were visualised using a Hamatasu™ NDP Nanozoomer. The wells were washed three times with PBS and primary antibodies against respective chemokines obtained from the R&D systems™ (CCL2-AF479-NA, CCL3-MAB450, CXCL2-AF452-NA, CXCR2-MAB2164) were added and incubated for 2 hr at room temperature with shaking. The wells were washed three times with PBS and respective HRP conjugated secondary antibodies were added. After incubation for 1 hr at room temperature with shaking, the wells were washed three times with PBS, and 100 μl of TMB substrate (Thermo Scientific™) solution was added. The plate was incubated for 15 min at room temperature, and the reaction was stopped by adding 2 N H 2 SO 4 . The colour change was measured at 450 nm on an Epoch Biotek™ plate reader and fold difference with respect to either NTHi−ve MEF or noninoculated Junbo mouse serum was calculated.