Mutation in Phex Gene Predisposes BALB/c-PhexHyp-Duk/Y Mice to Otitis Media

Genetic susceptibility underlying otitis media (OM) remains to be understood. We show in this study that mutation in Phex gene predisposes the BALB/c-PhexHyp-Duk/Y (abbreviated Hyp-Duk/Y) mice to OM, which occurs at post-natal day 21 (P21) with an average penetrance of 73%. The OM was identified by effusion in the middle ear cavity and/or thickening of middle ear mucosae, and was characterised by increase in goblet cells, deformity of epithelial cilia and higher expression of proliferating cell nuclear antigen (PCNA) in cells of the middle ear mucosae. Moreover, the transcription levels of Tlr2, Tlr4, Nfkb1, Ccl4, Il1b and Tnfα in the ears of the Hyp-Duk/Y mice at P35 were significantly upregulated, compared to those of the controls. Higher expression levels of TLR2, TLR4, NF-κB and TNF-α in the middle ears were demonstrated by immunohistochemistry (IHC). However, the OM in the mice was not prevented by azithromycin administration from gestational day 18 to P35. Further study showed that, in contrast to the low mRNA levels of Phex gene in the ears of the Hyp-Duk/Y mice, the mRNA level of Fgf23 was significantly elevated at P9, P14, P21 and P35. Meanwhile, mRNA levels of EP2 (PGE2 receptor), which expressed in the middle ear epithelia as demonstrated by IHC, were already increased at P14 even before the occurrence of OM, indicating that PGE2, an inflammatory mediator, is involved in the OM development in the mutants. Taking together, Phex mutation primarily up-regulates gene expression levels in FGF23 mediated pathways in the middle ears, resulting in cell proliferation and defence impairment at the mucosae and subsequently bacterial infection. The Hyp-Duk/Y mouse is a new genetic mouse model of OM.


Introduction
Otitis media (OM) continues to be one of the most common childhood infections [1]. Human predisposition to OM is affected by multiple factors, including Eustachian tube (ET) structure and function, immune status, innate mucosal defence, genetic susceptibility and pathogen exposure [1]. Increasing evidence supports an important role for heredity in susceptibility to OM [2]. Animal models with OM linked to a defined genetic lesion are important to reveal pathogenesis and underlying genetic pathways linked to OM [3].
X-linked hypophosphatemic rickets (XLH) is a dominant disorder characterised by hypophosphatemia resulting from impaired renal tubular reabsorption of inorganic phosphate [4]. In humans, XLH is caused by mutation in the Phex gene which codes for a 749-amino acid protein. Five mutations in the mouse Phex gene have been reported: Hyp, Hyp-2J, Hyp-Duk, Gy, and Ska1. The Hyp-Duk mutation is a spontaneous intragenic deletion involves at least 30 kb containing Phex exons13 and 14 as indentified by Southern blot analysis [5]. Because of the common phenotypes (including small body size, slightly abnormal skull shape and short tail) among Phex Hyp-Duk /Y (Hyp-Duk/Y) mutants, it can be inferred that the mutation results in loss of PHEX protein function in the ears. Actually, auditory brainstem response (ABR) threshold analysis showed that Hyp-Duk/Y mice have substantial hearing impairment compared to normal controls and to other Phex mutants [5].
In this investigation, we further characterise Hyp-Duk/Y hemizygotes, which present a high incidence of chronic OM. We present evidence of OM in the mutants and a possible mechanism for OM development. Our data imply that Phex gene mutation predisposes Hyp-Duk/Y mice to OM primarily by FGF23 mediated pathways. OM in the Hyp-Duk/Y mice may ultimately cause conductive, and possibly sensory, hearing loss.

Mice and animal care
The BALB/cAnBomUrd-Phex Hyp-Duk mice were originally housed in The Jackson Laboratory (Bar Harbor, ME, USA) research facilities and were relocated to Case Western Reserve University and maintained through sibling inbreeding. Mainly male mice, mutants (Hyp-Duk/Y, n = 122) and controls (+/Y, n = 109) ranging in age from post-natal day 9 (P9) to 20 weeks, were included in this study. All mice were genotyped by PCR using conditions and primers described previously [6]. Care and use of animals were approved by the Institutional Animal Care and Use Committee of The Jackson Laboratory (TJL, Bar Harbor, ME) and of Case Western Reserve University.

Haematoxylin/Eosin staining
Fifty-five mutant and 43 control mice from 2 to 20 weeks of age were anaesthetised and perfused through the heart left ventricle with PBS followed by Bouin's (haematoxylin/eosin or H&E staining) or by 4% paraformaldehyde (for all others) fixative. Bullae were dissected, perfused with fixative, immersed in same for 48 h, decalcified with Cal-EX solution for 6 h (or 10% EDTA, 4-7 days), embedded in paraffin, serially sectioned at 7 mm thickness, and stained with H&E for light microscopy (Leica DM4500 B, Leica Microsystems, Wetzlar, Germany) at 5 to 636 final magnification.

Scanning electronic microscopy (SEM)
Bullae were isolated from skulls of 4 control (+/Y) and 4 mutant (Hyp-Duk/Y) mice at P21 or P150. Samples were placed in 2.5% glutaraldehyde in cacodylic acid in 0.1 M phosphate buffer (pH = 7.2) for 36 hours. After decalcification in 10% EDTA for 4-7 days, middle ear cavities were dissected from bullae and dehydrated in serial solutions of ethanol (60%, 70%, 80%, 95%, 100%) for 15 minutes each, with a second soak in 100% ethanol. Each middle ear cavity was subjected to CO 2 critical point drying, sputter-coated with 60/40 gold-palladium, and viewed under high resolution scanning electron microscopy (Hitachi S-4500, Japan).  Mayer's mucicarmine method Goblet cells in the epithelium of the middle ear mucosae were identified by Mayer's mucicarmine staining, following the protocol of Electron Microscopy Sciences (Hatfield, PA). Briefly, sections from both ears of mice at age 12 weeks (5 mutants and 4 controls, prepared as in the previous paragraph) were deparaffinised in xylene and hydrated with distilled water. Sections were first stained in Weigert's haematoxylin, approximately 7 min and washed in running water for 10 min. Second, sections were stained in mucicarmine for 60 min at room temperature and rinsed quickly in distilled water. Thirdly, sections were counterstained in metanil yellow, rinsed in distilled water for 2-4 sec and dehydrated in 95% and 100% ethanol, sequentially. After clearing in xylene, sections were mounted with coverslips using permount for microscopic observation.

Proliferating cell nuclear antigen (PCNA) immunostaining
Specimens were prepared as described above (Mayer's Mucicarmine Method). Immunohistochemistry was performed according to standard protocols. From 4 mutant and 4 control mice at

Bacterial identification and antibiotic prophylaxis
Four 5-week-old mice from each genotype (Hyp-Duk/Y and +/Y) were randomly chosen and anaesthetised. Mice were euthanised by CO 2 asphyxiation. Under sterile conditions, bullae were removed; middle ears were isolated and washed with sterile PBS. One hundred ml of each PBS lavage were inoculated onto individual BBLTM TrypticaseTM Soy Agar plates with 5% sheep blood (TSA II, Fisher, Pittsburgh, PA) and plates were incubated at 37uC, 5% CO 2 for 18 h. Colonies were sorted, counted, and further identified. Staphylococci were identified by standard microbiologic features, with coagulase-negative Staphylococci (CNS) differentiated from S. aureus by positive tube-coagulase test.
Azithromycin (50 mg/kg/d) was therefore added to drinking water (final concentration of 0.25 g/l) [8] for Phex -Hyp-Duk /+ mice (n = 6, mated to +/Y mice) at gestational day 18 (G18) and drug treatment was maintained through lactation, after which the drug was given to the litters in drinking water. At P35, antibiotic-treated control (+/Y, n = 4) and mutant (Hyp-Duk/Y, n = 5) litters were sacrificed and the bullae of ears were isolated for H&E staining and microscopy, as described in histological preparation.

Semi-quantitative RT-PCR for evaluation of gene transcription levels in the ears
Four to five Hyp-Duk/Y mutant and 3-4+/Y wild-type male littermates ranging in age from P9 to P35 were used for the experiments. Mouse treatment and total RNA preparation from the middle and inner ears were performed following the methods described previously and the mRNA levels of specific genes were evaluated by semi-quantitative RT-PCR [7]. Additional RT-PCR primers were listed in Table 1.

Fluorescence immunohistochemistry (IHC) assays in middle ears
For detection of PHEX, TLR2, TLR4, NF-kB and TNF-a protein expression in the middle ears of Hyp-Duk/Y mice, cryosections were used and the detail methods were described in the supplementary section. To localize EP2 expression in the middle ear, paraffin sections were used following the procedure described [7]. Goat anti-EP2 polyclonal antibody (sc-22195, 200 mg/ml, Aanta Cruz Biotechnology, INC.) was used at 1:200 dilution. Donkey anti-goat secondary antibody conjugated to Alexa Fluor-568 was used at 1:400. Statistical method ANOVA test was used for statistical analysis. P,0.05 was considered to be significant.

Exon deletion yields functionally null PHEX protein in Hyp-Duk/Y mouse ears
The RT-PCR products from 2 samples in each group were randomly chosen for agarose gel electrophoresis (Figure 1 A), in which one single band from each sample was amplified. Sizes of Phex-specific RT-PCR products from Hyp-Duk/Y samples were smaller (70 bp and 137 bp) than products from control mice (252 bp and 319 bp) for both primer pairs. Sequencing of RT-PCR products confirmed that exons 13 and 14 were deleted from the Phex cDNA, leading to an open reading frame shift after aa 468 and introducing a stop codon at aa 520 (Figure 1 B). The deletion in the Phex gene does not alter splicing of mRNA from the mutant gene. This is the first time the Hyp-Duk/Y mutation has been sequenced at cDNA level to determine the specific nature of the mutation in the ears.
PHEX protein expression in the middle ears of Hyp-Duk/Y and control mice (P35) was determined by immunostaining with two antibodies (PHEX-H-176 and PHEX-C-13) that recognise peptides flanking the altered sequence ( aa 468 to 520) in the PHEX protein from aa positions 151 to 326 (H-176) and 530 to 560 (C-13), respectively. The antibody H-176 gave strong signals, mainly in middle ear mucosal cells, in both mutant and control mice. However, C-13 only gave signals in middle ears (mainly epithelial cells in mucosae and osteocytes in the bony wall) of control mice, and failed to stain any proteins in corresponding structures in mutant mice, indicating absence of the C-terminal portion of the PHEX protein, resulting from truncation in the mutant (Figure S1).
Histopathologic evidence of inflammation in the middle ears of Hyp-Duk/Y mice Ear sections from 55 Hyp-Duk/Y mice and 43 age-matched littermate control (+/Y) mice from 2 to 20 weeks of age were examined for histopathology. Overall, OM with effusion in the middle ear space and/ or thickened mucosae occurred in at least one ear of Hyp-Duk/Y mice at (or after) 3 weeks of age with an average incidence rate of 73% (35/48, Table 2). However, only two of 43 control mice at ages 3 weeks and 20 weeks, respectively, presented with OM by histological examination. In OM mice, mucosae of the middle ears and Eustachian tubes showed general thickening with proliferation of epithelial cells and lamina propria connective tissue ( Figure 2). The effusion was serous and contained variable numbers of inflammatory cells, mainly consisting of polymorphonuclear cells.

Increase in goblet cells and mucosal cell proliferation in the middle ears of Hyp-Duk/Y mice
Sections of each ear from mutant (Hyp-Duk/Y) mice and control (+/Y) mice at age 12 weeks were stained by the Mayer's mucicarmine method to visualise goblet cells (Figure 3 A, 3B), which have been demonstrated at high density in secretory OM [9]. In control mice, goblet cells were rare, but in Hyp-Duk/Y mutants, goblet cells were found scattered among other cells in the middle ear cavity epithelium, especially in cuboidal ciliated epithelium. PCNA is a reliable marker of cells undergoing active proliferation [10]. Strong PCNA staining was detected in the middle ear mucosae of Hyp-Duk/Y mutant mice, shown in Figure 3D, while epithelial cells of the middle ears in control mice were virtually PCNA negative ( Upregulation of pro-inflammatory cytokine expression in the middle ears of Hyp-Duk/Y mice Inflammation caused by bacterial infection is mediated by TLR signaling pathways and is characterised by high level production of pro-inflammatory cytokines which contribute to otitis media pathogenesis [11]. Therefore, transcription levels of Tlr2, Tlr4 and genes (Ccl4, Ccl2, Il1b and Tnfa) encoding pro-inflammatory cytokines were examined by semi-quantitative RT-PCR in mutant (Hyp-Duk/Y) and control (+/Y) mice. At P35, mRNA levels of the above genes were significantly upregulated (P,0.05 or 0.01) in the ears of mutant mice compared to those of controls ( Figure 5). Expression of TLR2, TLR4, NF-kB and TNF-a was also examined by in situ fluorescent IHC staining of sections from middle ears of +/Y and Hyp-Duk/Y mice at age P35 ( Figure S2). The four proteins were mainly expressed in the middle ear mucosal layers of both mutant and control mice, whereas the staining intensity of antibodies for each of the four proteins was much stronger in middle ear mucosae of Hyp-Duk/Y mice than that in the middle ears of +/Y mice. The results indicate that TLRmediated signaling pathways were activated in the middle ear immune response in OM.

Pathogenic bacteria isolation from middle ears of the Hyp-Duk/Y mice
To investigate the causative agent of OM infection in Hyp-Duk/ Y mice, we cultured middle ear lavage samples from 5-week-old mutant and control mice. Under microaerophilic conditions, the only type of bacteria isolated from middle ears of mutant (Hyp-Duk/Y ) mice were coagulase-negative Staphylococci (CNS), a Gram positive bacterium which has been frequently isolated from middle ear effusions in human otitis media patients [12,13]. Though CNS were also isolated from wild-type (+/Y) littermates, the mean CFUs (colony-forming units) of CNS in middle ears of mutant mice (55629.76) were significantly higher than in controls (15615.75) (P,0.01). However, OM in these mice could not be prevented by azithromycin administered from G18 to P35, which was demonstrated by histological observation (data not shown).

Upregulation of PHEX mediated gene expression in the ear of Hyp-Duk/Y mice
Phex gene mutation in Hyp-Duk/Y mice was previously correlated with elevated levels of fibroblast growth factor 23 (FGF23), which subsequently increases mouse prostaglandin E2 (PGE2) production in kidney [14]. Therefore, a time course determination of the transcription levels of Phex, Fgf23 and Ep2 (PGE2 receptor gene) were carried out in the mouse ears. The mRNA levels of Phex were significantly downregulated at P9, P14, P21 and P35 in the mutant mice, whereas those of the Fgf23 gene were upregulated, compared with the levels of the control mice. Moreover, mRNA levels for Ep2 were not increased in Hyp-Duk/Y mice at P9 and became significantly higher than controls at P14, P21 and P35 ( Figure 6A-D). Further fluorescent IHC assays of EP2 showed that EP2 expression was localized in the membrane and cytoplasm of the middle ear epithelial cells of both +/Y and Hyp-Duk/Y mice, but with strong staining intensity in the mutants at both P14 and P35 ( Figure 6E-H). The results indicate that PGE2 levels are increased in the ears of Phex mutants even at P14.

Discussion
In this study, we found that Phex mutation predisposes the Hyp-Duk/Y mice to OM. The single-gene mouse model of OM is similar to Jeff (Jf) mice, which present with fluid and pus in the middle ear cavity [3]. Onset of OM in Hyp-Duk/Y mice occurs after 3 weeks of age in about 73% of Hyp-Duk/Y mutants (from 3 weeks to 20 weeks of age). The OM in Hyp-Duk/Y mice was characterised by effusion and/or mucosal cell proliferation indicated by increased goblet cell density and by PCNA staining, a marker of active cell proliferation that is associated with otitis media [15,16]. Effusion in the middle ears attributes to local infection, activation of NF-kB and upregulated levels of cytokines [17][18][19]. Effusion can also be exacerbated by epithelial cilia deformity or by Eustachian tube constriction or obstruction [8].
Our further study implicates PHEX-FGF23 pathways in OM pathology of Hyp-Duk/Y mice. PHEX has been identified as causative agent of XLH and autosomal dominant hypophosphatemic rickets/osteomalacia (ADHR) [5,20]. The native PHEX protein was reported to inactivate a phosphaturic factor, which is considered to be fibroblast growth factor 23 (FGF23) [21]. Circulatory FGF23 is elevated in most patients with XLH [22], and high level of intact FGF23 despite hypophosphatemia is a clinical indicator in XLH diagnosis [23]. While PHEX is expressed primarily in cells of bone lineage, the main protein effects on renal phosphate wasting and impaired vitamin D metabolism occur in the kidney. FGF23 regulates urinary phosphate excretion to maintain systemic phosphate homeostasis. The exact mode of action of the phosphaturic effects of FGF23 is not fully understood and is an intense area of research [24]. PHEX and the sibling protein DMP1(dentin matrix protein 1) regulate growth factor FGF23 expression in osteocytes by controlling the pathway involving FGF receptor (FGFR) signaling [25]. In cultured kidney cells, FGF23 inhibits phosphate transport [26], probably via an FGF receptor 3c which is linked to the MAPK pathway [27]. FGF23 and alpha-klotho, a transmembrane protein, together stimulated osteoblast proliferation [28]; they also increased phosphorylation of ERK1/2, P38, JNK, AKT, IkB and GSK-3b and BrdU incorporation in renal and intestinal epithelial cell lines from human [29]. Activation of FGF receptor and Wnt/b-catenin is a newly identified molecular pathway stimulated by inactivation of PHEX in osteoblasts from Hyp mice [30]. The above studies suggest that FGF23, by interacting with FGF receptors, can initiate downstream signaling events by multiple pathways.
FGF23 increases urinary and renal production of PGE2, which inhibits proximal tubule phosphate transport in Hyp mice [14]. PGE2 in different tissues may have different functions. We therefore tested whether increased FGF23 can induce production of PGE2 in the ears of Hyp-Duk/Y mice. Direct determination of PGE2 expression in mouse ears is not feasible; thus, transcription of PGE2 receptor (EP2) was measured. mRNA levels of Ep2 were significantly upregulated in the ears of Hyp-Duk/Y mice not only at P21 and P35, but also at P14 (before the OM occurs), indicating that PGE2 would likely show a correspondent increase in the time course with OM development. As upregulation of Fgf23 mRNA levels in the ears of Hyp-Duk/Y mice was correlated with the levels of EP2 from P14 to P35, we postulate that OM in Hyp-Duk/Y mice is primarily mediated by FGF23/PGE2 pathways.
PGE2 can be released from various cell types (such as macrophages and endothelial cells) and mediates inflammation through its potent vascular permeabilization activity [31]. High concentration of PGE2 in middle ear effusions has been documented and correlated with the degree of inflammation during OM in humans [32]. PGE2 increases transepithelial ion transport rate by increasing intracellular cAMP content in a middle ear cell line (MESV); such a mechanism could diminish the periciliary sol layer to impair mucociliary clearance in OM [33]. In addition, PGE2 can cause production of TNF and interleukin IL-1b, which direct neutrophil trafficking into the middle ear cavity [34]. These cytokines can exacerbate OM in Hyp-Duk/Y mice, as in humans, by inflammation and tissue destruction [35].
Bacterial infection is a secondary factor which augments middle ear pathology. Coagulase-negative Staphylococci (CNS) is currently a major cause of nosocomial and health-care related infections [36]. This bacterium was isolated from the middle ears of an OM mouse model of Down syndrome in our previous study [37]. The higher CNS density in the middle ears of Hyp-Duk/Y mice may trigger, at least partially, elevation of mRNA levels of pro-inflammatory cytokines in TLR2-mediated pathway. The increased TLR4 levels in the ears of Hyp-Duk/Y mice were probably caused by infection of Gram-negative bacteria that were unable to grow at the culture conditions. As application of azithromycin from G18 to P35 failed to prevent OM in Hyp-Duk/Y mice, we concluded that bacterial infection was secondary to the primary pathological alteration mediated by FGF23/PGE2 in middle ears.
In summary, Phex gene mutation induced primary structural and functional middle ear abnormalities and, with secondary bacterial infection, resulted in OM in the Hyp-Duk/Y mice (Figure 7). Chronic inflammation could damage the fragile structures of the middle ear thus causing conductive and even sensory hearing loss. The Hyp-Duk/Y mouse is therefore a promising model for studying the molecular mechanism of OM and for testing and targeting drugs to prevent OM.

Materials and Methods S1
(DOC)