The epithelium of the middle ear is mostly a single layer of cubical squamous cells. Some patches of the middle ear epithelium, and especially close to the Eustachian tube, gradually change in a pseudostratified columnar epithelium similar to the mucociliary epithelium of respiratory epithelia [2]. This epithelium is constituted of basal cells, goblet cells producing mucins, and other cells that can be ciliated or not. On top of these patches, mucus is present and protects these regions of the epithelium from the infection. The ciliated cells ensure the movement of the mucus in direction to the Eustachian tube orifice where it is evacuated from the middle ear to the oral cavity.

Mucus glands, a normal feature of the Eustachian tube, can also be present in the middle ear of patients with OM. They constitute invaginations of the epithelium in the lamina propria, regions very rich in goblet cells that are able to produce large quantities of mucins. In healthy subjects that did not have prior disease of the middle ear, very few mucus glands are usually observed, whereas in subjects that had a history of OM events, these glands are more numerous [3]. This suggests that in the middle ear, mucus glands probably appear after several episodes of OM and then remain even after the disease is resolved. In patients with chronic suppurative otitis media (CSOM), the density of mucus glands is very high [4] and appears as a sequelae of CSOM [5]. Studies of the structure of mucus glands in the middle ear showed that the epithelium first invaginates at the location of high-density goblet cells and then different ramifications develop to lead to different structure types and sizes. Nevertheless, it was noticed that mucus glands of the middle ear can degenerate and lose their ability to produce mucins [3] likely because when OM resolves, the ear contains less factors sustaining inflammation and mucin production.

As explained before, the healthy middle ear epithelium contains few goblet cells that are concentrated in some patches of mucociliary epithelium mainly close to the Eustachian tube. But in the case of OM, the simple layer epithelium remodels into a pseudostratified epithelium. Several studies have demonstrated by histology techniques (Hematoxilyn and eosin staining on cuts of paraffin-embedded tissues) that the middle ear epithelium exhibits more secretory cells as well as ciliated cells in numerous parts of the middle ear epithelium [6–8]. Secretory cells are positive to periodic acid Schiff (PAS) staining detecting the presence of glycoconjugates, which are mainly in mucin proteins. Smirnova et al. [9] demonstrated that these mucins are secreted in the MEEs as they were PAS positive using a slot blot.

OM is characterized by the presence of fluid in the middle ear cavity, called effusions that can be serous or mucous. Serous effusions do not contain mucins and are not viscous. On the contrary, mucous effusions are highly viscous and contain a high content of mucins [10]. An in vitro test of transportability of a bead under magnetic attraction also indicated that mucous effusions are less transportable than serous ones, suggesting that mucous effusions are difficult to clear in the middle ear [11]. Serous effusions contain proteins similar to the blood, so it is suggested that serous fluids are the result of a passive transudate of blood components in the middle ear due to a negative pressure in the middle ear likely because of the Eustachian blockade during inflammatory OM [12, 13]. Even if there are some conflicting findings from different research groups, serous effusions are believed to show better outcomes of the disease, whereas mucoid effusions are suggested to predict chronic otitis media with effusions (COME) [11, 14]. The study of Matkovic et al. [15] also contributed to this hypothesis as among 108 effusions collected, only 6 % were mucoid for patients having OM diagnosed for less than 3 months, whereas 95 % were mucoid for patients having the disease for more than 3 months. The large differences in medical outcomes and effusion and middle ear mucosa (MEM) characteristics prove that various cellular and molecular pathways act in the evolution of the disease. The production of large amounts of mucins necessitates the differentiation of goblet cells in the epithelium and the development of mucin glands. Indeed, the reabsorption of water in serous fluids and the concentration of their proteins are believed to participate to turning serous effusions into mucous ones [16]. Contrary to serous effusions, mucous effusion production necessitates the active process of producing mucins (exudates). But as a large number of proteins from the blood are also present in the mucous effusions (as albumin the predominant one), a passive diffusion of proteins and liquid is also probably implicated in the accumulation of mucous effusions.

Ion transport and water channels are also believed to play an important role in bringing water in the middle ear cavity and participate to serous and mucous effusion production. The healthy middle ear has to be kept without fluid contrary to the inner ear for a good transmission of the sound vibrations. Herman et al. [17] suggested the importance of water channels and ion transports. Experiments conducted in Mongolian gerbil’s middle ear cells showed that the absorption of fluid in the middle ear was dependent on an osmotic gradient created by sodium and potassium adenylpyrophosphatase (ATPase)-dependent channels. The impairment of this ion flux has been shown in the lungs of rabbits in response to hydrogen peroxide that is produced during oxidative processes as well as hypoxia, a condition likely to appear in OM [18]. The aquaporins (AQP)1, 4, and 5, channels regulating the water homeostasis in cells, were detected in the Eustachian tube and MEM of rats as well as the epithelial sodium channels (ENaCs) [19]. In experimental OM in rats induced by Eustachian tube obstruction, ENaC and AQP were deregulated from 1 to 8 weeks after Eustachian tube obstruction, suggesting their implication in the water imbalance leading to fluid presence in the middle ear [20].

Effusions are composed of mucins but contain other proteins (antibacterial proteins, cytokines, etc.), lipids, deoxyribonucleic acid (DNA), and bacterial components [10, 14, 21], some of these substances being remains of dead bacteria and epithelial cells. Mucins , the major macromolecular component of epithelial mucus, are very high-molecular-weight proteins constituted of a backbone where numerous sugar side chains are added as a posttranslational modification (glycosylation with glycotransferases enzymes). These glycoconjugates are linked to the mucins in the Golgi and are then stored in secretory granules, waiting to have a signal to merge with the membrane and be released in the extracellular compartment [22]. Mucins are widely studied as their regulation is often a key determinant of diseases as cancer, lung diseases, and gastrointestinal diseases.

Mucins are classified by their protein backbone that is encoded by different mucin genes called MUC. MUC transcripts are big (until 15 kilo bases), so is the protein backbone, accounting for 15–50 % of mucin mass, and can contain 400 to more than 11,000 amino acids [22].The major posttranslational modification of mucins is O-glycosylations, consisting in O-glycans attached to tandem repeats rich in serine and threonine all along the backbone. N-glycosylation is also observed but in a lesser extent. More than 20 human MUC genes have been identified—about the same number in mice. Considering the size of their gene transcript and protein backbone, but also their many glycoconjugates, the analysis of the MUC proteins is difficult. In the respiratory tract, 12 MUC genes have been identified and less in the ear, probably because mucins are more studied in the airways. Studies of mucins in the MEM and MEEs seem to show that MUC5B is the predominant mucin in the ear of patients having OM, whereas healthy subjects show very low levels of mucin [6–8, 23, 24]. But other mucins have been detected, apparently in lower amounts, which are MUC5AC, MUC2, and MUC4 [7, 13, 24, 25]. MUC5B has been detected by transcript analysis and protein assay: Preciado et al. [23] detected MUC5B protein by mass spectrometry in MEEs from COME patients and Lin et al. [7] by immunohistochemistry on the mucosa of patients with mucoid OM, whereas non-inflamed mucosa did not react with either of the antibodies anti-MUC5B and anti-MUC4. It has been noticed that the Eustachian tube of mucoid OM patients had MUC5B, MUC4, and also MUC5AC and MUC1 glycoproteins [8]. From the same study, electron microscopy of secretions from COME patients showed the presence of chain-like polymeric mucin. Some studies have detected the presence of messenger ribonucleic acid (mRNA) transcripts of MUC1, MUC2, MUC3, MUC4, MUC5AC, MUC5B, MUC7, MUC8, MUC9, MUC11, MUC13, MUC15, MUC16, MUC18, MUC19, and MUC20 [24–26], but gene expression does not always reflect the protein production and secretion. Indeed, Thornton et al. [27] showed that mucin gene expression in the airways was not always correlated to the presence of the protein. Mucins can be secreted or attached to the cell membrane. Among the mucins detected in the middle ear, we can notice that MUC5B and MUC5AC are secreted mucins, whereas MUC1 and MUC4 and are membrane-tethered mucins [22].

The overproduction of mucins leads to mucoid effusions that are hard to clear by the middle ear. Efforts have been made to try to prevent mucin overproduction, but a recent article pointed to the necessary presence of MUC5B in the innate immune response of airways and the ear. A study directed by Dr. Christopher Evans showed that the knockout of Muc5b in mice had a fast and dramatic effect on the mortality and morbidity due to infection of the airways leading to systemic infection [28]. Histology of the lungs showed an overproduction of Muc5ac probably to compensate the lack of Muc5b, but failed to protect the airways from infection. The ears were also infected by different bacteria and contained liquid as well as signs of inflammation. Thus, Muc5b glycoprotein is needed in the airways and the ear to protect mice against bacterial invasion and shows its central role in the innate immunity.

MEEs contain other molecules that participate to the defense against pathogens. Antibacterial proteins efficiently kill bacteria and are very important in the innate immunity mechanisms. Defensins are broad-spectrum antimicrobial peptides, small (30–45 amino acids), rich in cationic amino acids, and stabilized by disulfide bounds that protect them from proteases [29]. Defensins have antimicrobial properties towards bacteria and viruses, are able to inhibit some bacteria toxins [30], and have pro-inflammatory activities stimulating cytokine and chemokine production [31]. Surprisingly, defensins have not been studied in patient samples, but in vivo and in vitro studies of experimental OM showed their induction in response to bacterial infection of the middle ear [32, 33]. Human β-defensin 2 (HBD2) was studied in vitro in order to determine the molecular pathways implicated in its induction. In human middle ear epithelial cells (HMEEsCs), HBD2 is under the control of the pro-inflammatory cytokine interleukin (IL)1-β that activates Raf-MEK1/2 (mitogen-activated protein kinase kinase), the mitogen-activated protein kinase (MAPK) pathway [34]. Non-typable Haemophilus influenzae (NTHi) is also able to induce the expression of HBD2 first activating the toll-like receptor (TLR) 2 and then inducing protein 38 (p38) MAPK pathway [32]. HBD1 and HBD2 have also shown their ability to reduce Streptococcus pneumonia (SP), Haemophilus influenzae (Hi), and Moraxella catarrhalis (MC) growth in a liquid broth assay [35]. In Chinchilla, the orthologue of human β-defensin 3, chinchilla β-defensin 1, CBD1, had potent antimicrobial activity against SP, Hi, and MC [36]. Furthermore, chinchillas pretreated with recombinant CBD1 resulted in lower colonization of NTHi in the nasapharynx [37]. But bacteria are able to resist to defensins when they are growing in biofilms: Jones et al. [38] demonstrated that HBD3 binds to extracellular DNA constituting the matrix of NTHi biofilms, leading to the sequestration of HBD3 and thus diminishing the biological activity of an important defense of innate immunity.

Other antibacterial molecules are also part of the innate immune defense. Among them, the lysozyme is a cathelicidin (a cationic peptide) that has various effects, primarily damaging the membrane of bacteria. Lysozyme is present in MEEs of pediatric patients having OM, especially in mucous effusions compared to serous ones [12, 13, 39]. Giebink et al. [40] showed that the concentrations of lysozyme are more important in the MEEs of patients with COME positive for bacteria culture and suggested that this antibacterial agent was not only produced by polymorphonuclear leukocytes but also by the middle ear epithelium that accounted for 50–80 % of the lysozyme in the middle ear. Experimental OM in animals also demonstrated higher lysozyme detection in the middle ear: in response to MC in the Guiney pig [41] and in response to SP in chinchilla, this study also showed that more lysozyme were observed even when heat-killed bacteria were injected in the middle ear, suggesting that the production of lysozyme might be activated in response to membrane components of bacteria. Furthermore, mouse depleted of lysozyme also showed a higher susceptibility to OM development after SP infection [42], underlining the importance of lysozyme in the innate immune defense against bacteria.

Finally, some other antibacterial molecules poorly studied seem to play a role in the middle ear defense to pathogens: surfactant proteins as short palate, lung, and nasal epithelium clone (SPLUNC)-1, small cationic peptides, halocidin, and xylitol [29, 43].

The complement system is a biochemical cascade composed of several peptides normally present as inactive forms. This system can be activated by different sequential cascades of enzymatic reactions in which proteins are sequentially cleaved and activated. The resulting effector molecules are C3a and C5a, also called anaphylatoxins. They are the most potent activation products of the complement that are able to induce a large diversity of effects as bacterial cytotoxicity, induction of pro-inflammatory cytokines production, and inflammatory cell activation [44]. The activation of the complement system depends on three pathways. The classical pathway consists in the recognition of immunoglobulin IgG and IgM complexes formed around pathogens that activate the C1 complex, activating C4 molecules to induce the activation of C3 and C5. The alternative pathway is triggered by carbohydrates, lipids, and proteins found on pathogens: C3 mediates the activation of the cascade. And the lectin pathway recognizes sugars at the microbial surface and leads to the activation of C4, C3, and finally C5.

Mediators of the complement activation have been found in MEEs and the MEM. Recently, He et al. [45] analyzed molecules of the component in effusions of children with recurrent OM by the enzyme-linked immunosorbent assay (ELISA). High amounts of C3a, C5a, and sC5-b9 were detected in the MEEs of patients having OM for more than 6 weeks. The concentration of C5a was also strongly correlated to the concentration of IL-6 and IL-8 pro-inflammatory cytokines, suggesting a link between complement activation and the inflammatory effect they induce. Complement transcript induction was also observed in HMEEsCs in vitro in response to SP and influenza A virus (IAV). Another study was conducted on effusions of patients with COME to assay the complement activation (C3a and C3 cleavage fragments) by ELISA and western blot analysis [46]. High concentrations of complement molecules were found and C3 activation was evaluated at 40 % of the total amount of C3 protein. They also noticed that C3a concentration was higher when effusions stayed longer in the ear and when children had multiple tube insertions, pointing C3a levels as a marker of the chronicity of OM. The complement activation leads to lysis of pathogens; this has been verified by Niarko-Markela and Meri [47] with erythrocytes of Guiney pigs exposed to MEEs from patients with otitis media with effusion (OME). Thirteen of the 38 MEEs tested had direct endogenous hemolytic activity, and 27 enhanced serum-initiated lysis. They also detected high levels of terminal complement complexes demonstrating the strong activation of the complement.

Multiple cell types and especially epithelial cells that are in contact with the external environment express innate immune receptors as Toll Like Receptors (TLRs). In the mucosal environment, mast cells and dendritic cells also express TLRs. The TLRs are pattern recognition receptors that recognize pathogen-associated molecular patterns (PAMPs). The activation of TLRs leads to the production of molecules also implicated in the innate immune response, as chemokines, cytokines, interferons (IFNs), and antimicrobial molecules described before. TLRs are type I transmembrane receptors with an extracellular N-terminal region with leucin-rich repeats and an intracellular toll-IL-1 receptor (TIR) domain. TLRs can form homodimers or heterodimers. The homodimers of TLR4 , TLR5, TLR11, and the heterodimers of TLR2-TLR1 or TLR2-TLR6 bind to their respective ligands at the cell surface, whereas TLR3, TLR7-TLR8, TLR9, and TLR13 localize to the endosomes, where they sense microbial and host-derived nucleic acids. TLR4 localizes at both the plasma membrane and the endosomes. They each recognize specific types of PAMPs, for example, TLR1-TLR2 and TLR1-TLR6 recognize acylated peptides, TLR4 recognizes lipopolysaccharide (LPS), TLR3 targets double-stranded ribonucleic acid (RNA), and TLR9 recognizes bacterial DNA. TLR signaling is induced by their dimerization, dependent on ligand binding. All TLRs except TLR3 have a signaling pathway dependent on myeloid differentiation factor 88 (MyD88), activating the transcription factor NF-ĸB and then the expression of pro-inflammatory cytokines [48].

Several TLRs have been identified in the MEM and MEEs. TLR2, TLR4, TLR5, and TLR9 were found at the level of RNA and proteins in the MEM of both OM and non-OM patients [49]. For TLR2, TLR4, and TLR5, no difference of expression was found between non-OM and OM MEM, but their concentration was lower in the mucosa of patients with CSOM. In consequence, it was suggested that the clinical recovery of OM depends on TLR expression in the middle ear. There are conflicting evidences considering TLRs in MEEs and the correlation with the presence of bacteria. Lee et al. [50] observed effusions of patients with OME having lower TLR2, TLR6, and TLR9 mRNA when they are prone to have persistent OM and the level of TLRs is higher in culture-positive MEEs. Another study demonstrated the inverse for TLR9: less TLR9 mRNA is detected in culture-positive MEEs [51], whereas Lee et al. [52] failed to see any difference in TLR2, TLR4, TLR5, and TLR9 mRNA. Studying TLRs in MEEs might be accurate for assaying their presence in immune cells but does not take into account the epithelial cells playing an important role in the immune defense through TLRs. This might explain the differences described before. Nevertheless, animal studies showed that defects in TLR2 and TLR4 lead to the persistence of inflammation and mucosal metaplasia during OM [53].

Inflammation is a central innate immune response activated by pro-inflammatory mediators (chemokines, cytokines) in order to attract and activate immune cells, stimulate the various innate immune defenses, and initiate the adaptive immune response to pathogens. This is a very efficient process involving different mediators and cells, but also deleterious if it does not resolve when the pathogens are no longer present. Inflammation is suspected to participate to the absence of resolution of OM especially in the case of COME and CSOM.

Animals are useful models to investigate the cellular and molecular mechanisms implicated in OM. They permit to control the type of infection, the different stages of a disease, as well as the genetic background of the biological material. Comparing to in vitro studies, in vivo models allow taking into account the entire immune system and the interaction between different cell types being important in the resolution of infections. But we have to keep in mind that animal models have their limitations as their responses to pathogens might not be the same as the human ones, the differences being dependent on the species chosen. For the study of OM, rodents are widely used: chinchilla, mouse, rat, Guiney pig, and gerbil. According to the literature, mice and chinchillas are the main animals used in laboratories to study OM. Mouse is the first animal models used now as they are small, with a very controllable genetic background, easy to use in laboratories because a high diversity of reagents are compatible with this species. Nevertheless, mice have a very small middle ear, which is less convenient to induce OM by surgery as well as collecting MEEs. Chinchilla offers the possibility to have a bigger middle ear: the review of Ryan et al. [68] compared the middle ear volume observed in different studies. The average middle ear volume of the chinchilla is about 1.5 ml³, whereas the one of a mouse is about 0.05 ml³, so the middle ear volume of the chinchilla is 30 times bigger than the one of the mouse. It is consequently easier to manipulate the middle ear and recover MEEs that are sufficient in quantity to do several biological assays. The anatomy of the chinchilla ear has also been shown to be very close to the human one [69]; they do not often develop spontaneous OM [70], and they show similar responses to virus and bacteria in the course of OM compared to humans even if the pathogens colonizing humans are not usually the same as those of chinchillas (see [71]). Rats are also used in several studies and have the advantage of having a bigger middle ear and more availability of reagents than chinchillas. Several interesting studies used rats to do a time course analysis of OM development.

OM is often induced by experimental obstruction of the Eustachian tube, leading to a negative pressure in the middle ear [72, 73]. Infection by bacteria can be coupled to this procedure to mimic better human OM. Bacterial injection can be made through the tympanic membrane but damaging this membrane lets other contaminants the possibility to enter in the middle ear. Injection via the ventral bulla is preferred as it does not damage the tympanic membrane and avoids contaminations. But it necessitates skills in microsurgery to avoid damaging the vessels and airways around the bulla. Infections post surgery can occur and may modify the immune response in the middle ear. Considering these limitations, Stol et al. [74] developed a noninvasive murine model adapted from a previous rat model. They used a pressure cabin at 40 kPa which induced pneumococci translocation from the nasopharyngeal cavity to the middle ear; the maximum bacteria load appearing 96 h post infection with the bacteria. Inflammation was confirmed with the secretion of IL-1β and TNF-α in the middle ear. This model has the advantage to avoid the limitations due to the surgery but probably does not permit to have homogenous OM development between animals. Another disadvantage should be considered: other parts of the body might be affected by the difference of pressure, especially for medium- or long-term experiments. Finally, pressure cabins might not be easy to use, expensive to buy, and might not permit to expose enough animals at the same time.

Human pathogens are studied in animal models as relevant clinical strains. But they are evolutionary adapted to humans and usually not animals, which can bring bias in these experimental studies. Nevertheless, the effects observed in experimental OM induced in animals and especially mice are very close to the observations made in humans: OM induced in mice having different genetic backgrounds with different strains of SP, Hi, and MC have shown similar inflammatory and mucosal effects even if the duration of the disease, the intensity of the responses, and the ability of resolution of OM where variable [68].

OM was also evaluated in mutant mouse strains; a strategy often used to investigate the implication of a specific gene in the apparition or the course of a disease. Mutations are natural or induced in laboratories. For the study of OM, we have to be careful choosing the type of mutations. Mutations in genes acting in the development of the middle ear might create some morphologic defects, influencing the responses of the middle ear. Mutating central genes in the immunity might also compromise the response to pathogens. And the deletion of a gene sharing similar functions with other genes sometimes leads to compensation mechanisms that may compensate the loss of functionality. Otherwise, this type of biological material offers great possibilities in studying spontaneous OM or pathogen-induced OM.

Transformed middle ear epithelial cells are now widely used to investigate the mechanisms implicated in bacteria effects, especially focusing on inflammatory and mucoid effects. This type of biological material permits to assay the effect of live bacteria, bacteria lysates, purified bacteria proteins, inflammatory mediators, etc., in a homogenous cell type which is useful but lacks the interaction with other cell types, especially the immune cells that produce mediators regulating epithelial cells. Knowing the limitations of cells in vitro, it is a very useful tool that gives us opportunities we cannot have with animals: the analysis of mechanisms implicated in a biological effect is more easy.

The human middle ear epithelial cell line HMEEC-1 was created by Dr. David Lim in 2002 using a retrovirus containing E6/E7 genes of human papillomavirus type 16 to transform primary middle ear epithelial cells from adults [75]. This type of transformation is known to regulate the cell cycle acting on the retinoblastoma (RB) tumor suppressor gene limiting the repression of the cell cycle and mediating the degradation of p53 protein also implicated in cell cycle repression [76, 77].

The mouse middle ear epithelial cell (mMEEsC) line was made by Dr. Jizhen Lin laboratory in 2005 [78]. Middle ear epithelial cells were isolated from mice and transformed by the large T-antigen of the simian virus 40 (SV40) A-gene. These cells have the property to be temperature sensitive: At 33 °C, the SV40 antigen is active and stimulates the cell cycle. But at 37/39 °C, SV40 is inactivated and cells differentiate, expressing markers of epithelial cells such as keratins and collagens. In our laboratory, we have noticed that these cells can be cultured several weeks at air liquid interface and form a single layer epithelium.

Other cell types were used: the middle ear cell line from chinchillas immortalized by SV40 [79] and the primary chinchilla middle ear epithelial cells (CMEEsCs) [80] or primary middle ear epithelial cells from adults successfully differentiated at air liquid interface in a ciliary and secretory epithelium [81]. Our laboratory tried to culture middle ear epithelial cells from children middle ear but because of the low amount of cells available during these procedures, we were unable to successfully grow them.

MEEs from patients with OM contain high concentrations of a panel of pro-inflammatory cytokines (see Sect. 3.3.1). In vivo studies were conducted to try to replicate the conditions of infection occurring in OM in order to analyze the expression and secretion of pro-inflammatory cytokines in the MEM and in MEEs mainly in mice, chinchillas, and rats (Table 7.1 and 7.2).