MyD88-Dependent Immune Response Contributes to Hearing Loss in Experimental Pneumococcal Meningitis

Hearing loss is the most common long-term complication in pneumococcal meningitis, affecting up to 26% of survivors [1]. This makes bacterial meningitis one of the leading causes of acquired hearing impairment in children and young adults. The site of the lesion is the cochlea, where a severe suppurative labyrinthitis develops [2, 3]. Although little is known about the induction of the cochlear immune response during pneumococcal meningitis, recent studies have uncovered crucial cerebral mechanisms of immune activation in pneumococcal meningitis. Among them, ligation of bacterial components on Tolllike receptors (TLRs) seems to be important [4]. Nearly all TLRs use a common intracellular adapter protein, MyD88. On its association with a TLR, MyD88 signals for activation of NF-κB and subsequent up-regulation of proinflammatory mediators. The role for MyD88 in pneumococcal meningitis was recently demonstrated by the intracisternal infection of MyD88^−/− mice infected with Streptococcus pneumoniae: a reduction of the cerebral immune response was associated with higher bacterial titers [5]. Here, we established the first mouse model of meningitis-associated labyrinthitis and studied the impact of MyD88 deficiency on the development of suppurative labyrinthitis and sensorineural hearing loss by use of knockout mice.

Materials and methods. The animal model used has been described elsewhere [5]. In brief, meningitis was induced by transcutaneous intracisternal injection of 1.5 × 10⁵ cfu of S. pneumoniae type 3 (provided by B. Grabein, Max von Pettenkofer Institut, Munich, Germany) into the cisterna magna under anesthesia with halothane. Then, mice were allowed to wake up. All mice that were studied >24 h received intraperitoneal therapy with ceftriaxone (100 mg/kg daily) for a total of 4 days. At the end of each experiment, mice were anaesthetized with 100 mg/kg ketamine and 5 mg/kg xylazine, and cerebrospinal fluid (CSF) samples were obtained for CSF white blood cell (WBC) count. After deep anesthesia was induced with ketamine and xylazine, mice were perfused transcardially with 15 mL of PBS containing 10 U/mL heparin. Temporal bones were dissected, fixed in 4% formalin, decalcified in PBS containing 10% EDTA, and embedded in paraffin. All animal experiments were approved by the government of Upper Bavaria, Germany.

For measurement of auditory brain-stem responses (ABRs), mice were anaesthetized with ketamine and xylazine. Needle electrodes were placed over both mastoids (negative), the vertex (positive), and the neck (ground). Impedances were controlled <5 kΩ. Square-wave click impulses (duration, 100 μs; frequency, 20 Hz) and tone bursts of 1 and 10 kHz (duration, 4 ms; frequency, 23.4 Hz) were delivered by use of earphones (E-A-RTONE3A; Auditory Systems). ABRs were amplified (× 250,000), band-pass filtered (150–10,000 Hz), and averaged (n = 1000) by use of Neuroscreen Plus (Jaeger-Toennes). To determine the hearing threshold, we started with wave impulses of 105 dB and reduced the intensity in 5-dB steps. The lowest intensity that elicited ABRs was considered to be the hearing threshold. If ABRs could not be elicited at 105 dB, stimulusintensities of up to 130 dB were tested. In wild-type and MyD88^−/− mice, hearing thresholds were determined 1 day before infection and before the end of the experiment. Accordingly, hearing loss was calculated as ΔdB.

For histological analysis, 7-μm midmodiolar sections of mouse temporal bones were deparaffinized, rehydrated, and stained with hematoxylin-eosin. The area of each section of the spiral ganglion in the basal cochlear turn was measured, and intact-appearing spiral ganglion neurons (containing a nucleus) were counted within [3]. Spiral ganglion neuronal density was expressed as neurons/0.01 mm². Severity of acute suppurative labyrinthitis was compared between MyD88^−/− mice and wildtype control mice by use of a cochlear inflammation score that has been described elsewhere [3]. Three sections were investigated for each cochlea, and the mean was calculated for each readout parameter.

Experimental groups were as follows. First, meningitis-associated hearing loss and the corresponding cochlear histopathologic alterations were characterized during acute pneumococcal meningitis (6 h after infection, n = 12 ears of 6 mice; 24 h, n = 10 ears of 5 mice; 30 h, n = 12 ears of 6 mice; 48 h, n = 6 ears of 3 mice; 72 h, n = 6 ears of 3 mice) and after recovery (5 days, n = 8 ears of 4 mice; 2 weeks, n = 8 ears of 4 mice). Second, MyD88^−/− mice and wild-type control mice of the same age and weight were infected simultaneously as described above and then investigated 24 h after infection. The degree of cochlear inflammation was compared between infected MyD88^−/− mice (n = 16 ears of 9 mice) and infected wild-type mice (n = 27 ears of 15 mice). Two (MyD88^−/−) and3 (infected wild-type control) ears broke because of the difficult preparation and could not be evaluated histologically. Hearing impairment was assessed on the basis of ABRs (for MyD88^−/− mice, n = 10 ears of 5mice; for infected wild-type controlmice, n = 12 ears of 6 mice). Infected MyD88^−/− mice were not studied for 124 h, because of their high mortality. This high mortality was found to be caused by an aggravation of systemic complications [5].

To determine cerebellar bacterial titers, cerebelli were dissected and homogenized in 1 mL of sterile PBS. Cerebellar homogenates were diluted serially, plated on blood agar, and cultured for 24 h at 37°C in 5% CO₂. The detection limit was 1 × 10³ cfu/mL of homogenate.

Data were analyzed by use of SYSTAT software (version 9; SPSS). Hearing thresholds between experimental groups were compared by 2-sample t test. A paired t test was used when the hearing-threshold development over time was analyzed. α-adjustment was used for multiple comparisons. P < .05 was considered to be significant. Data are displayed as mean ± SD values.

Results. All infected mice developed clinical signs of infection, such as piloerection, motor deficits, impaired vigilance, and weight loss as early as 6 h after infection. The maximal clinical picture was seen at 30 h after infection. Bacterial cultures from cerebellar homogenates were positive from 6 h until 72 h after infection, and CSF pleocytosis was observed (maximum at 24 h after infection, 14,509 ± 9902 cells/μL). Ceftriaxone therapy (initiated 24 h after infection) resulted in an almost complete remission of the clinical picture and a reduction in CSF WBC count to normal (2 weeks after infection, 113 ± 126 cells/μL).

A substantial elevation of hearing thresholds was detected as early as 24 h after infection for click stimuli, 1-kHz stimuli, and 10-kHz stimuli (figure 1A). The strongest hearing impairment was observed for high frequencies (10 kHz) (figure 1B). Throughout the course of the experiments, no further changes in the hearing capacity of the mice could be seen (figure 1A).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1.

In infected mice, hearing thresholds (A) were significantly elevated starting 24 h after infection and remained high until 2 weeks after infection (click stimulus). At any time point after infection, higher frequency hearing (10-kHz stimulus) was more strongly impaired than lower frequency hearing (1-kHz stimulus) (B). Hearing loss in panel B is shown as ΔdB at 24 h after infection. Spiral ganglion neuronal density decreased over time (C), indicating neuronal injury and death. In panels B and C, asterisks indicate P < .05. In comparison to uninfected control mice (D–F), mice with acute meningitis-induced labyrinthitis (G–I) showed dense infiltration of the scala tympani by granulocytes (asterisk, H). Hemorrhage in the spiral ganglion was a consistent finding (I). Two weeks after infection (K–M), damage had occurred to hair cells (arrow, L) and destruction of the lateral cochlear wall could be observed, as indicated by loss of cellular density in the spiral ligament (pound sign, L). Most striking, heavy loss of intact neurons was evident in the spiral ganglion (M). Staining was with hematoxylin-eosin. The displayed pictures show a selection of representative stainings.

During the acute stage, in all cochleae of infected mice, dense granulocytic inflammation of the perilymphatic spaces (scala tympani was affected more than scala vestibuli), but not of the scala media, was observed (figure 1G and 1H). The most severe inflammation was seen in the basal turn of the cochlea (figure 1G and 1H). After successful antibiotic therapy (2 weeks after infection), suppurative labyrinthitis was replaced by fibrocytic occlusion of the perilymphatic spaces, which was accompanied by the formation of blood vessels (figure 1K and 1L). Cellular density in the spiral ligament had declined, and the organs of Corti in basal sections of the cochlea were severely damaged or even nonexistent in 5 of 8 ears (figure 1L). Furthermore, the neuronal density in the spiral ganglion had declined significantly (figure 1C and 1M).

Clinical score and bacterial titers from cerebellar homogenates were significantly higher in MyD88^−/− mice than in infected wild-type control mice (cerebellar bacterial titers, 6.96 ± 0.53 vs. 6.08 ± 0.53 cfu/mL of homogenate; P< .05), whereas the immune response was diminished (CSF WBC count, 1775 ± 698 vs. 7687 ± 5888 cells/μL; P< .05). Even though MyD88^−/− mice also developed hearing loss until 24 h after infection, this hearing impairment was significantly smaller than that in wild-type control mice (figure 2). This was the case for square-wave impulses, 1-kHz tone stimuli, and 10-kHz tone stimuli. Furthermore, at 24 h after infection, granulocytic infiltration of the labyrinth was significantly lower in MyD88^−/− mice than in wild-type control mice, whereas dense bacterial growth could be observed in gram-stained sections (figure 2).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 2.

MyD88-deficient mice developed significantly less hearing impairment until 24 h after infection as wild-type control mice (A). This was accompanied by a reduced inflammatory response (B) and increased bacterial density in the cochlea (C, infected wild-type mice; D, infected MyD88^−/− mice; staining was with Gram stain; asterisks indicate P < .05). The displayed pictures show a selection of representative stainings.

Discussion. The current pathophysiologic concept of meningitis-associated hearing loss has been derived from animal models, including models using rabbits, guinea pigs, gerbils, and rats [3, 6–9]. These models mimic the electrophysiologic and histopathologic characteristics of meningitis-associated hearing loss quite well and have been used for the investigation of inhibitors/scavengers (uric acid and Mn[III]tetrakis[4—benzoic acid]porphyrin [MnTBAP]) [10], metabolic factors (superoxide dismutase and neuronal derived growth factor) [6], antibiotic therapy [9], and adjunctive treatment options (dexamethasone and N-acetyl-L-cystein) [7, 10]. Creation of the present mouse model of meningitis-associated labyrinthitis establishes a new tool to study meningitis-associated hearing loss via the use of knockout mice.

All infected mice in this study developed hearing loss until 24 h after infection, indicating robustness of the model and high reproducibility. Closely resembling the symptoms from humans with pneumococcal meningitis and from other animal models of pneumococcal meningitis—associated hearing loss [2, 3], the acute cochlear alterations in infected mice was dominated by severe granulocytic inflammation of the perilymphatic spaces, whereas the endolymph-filled scala media remained unaffected. Again, the observed aggravation of the inflammation in the basal portion of the scala tympani is consistent with the observation that hearing was significantly more affected in the high-frequency range, which is located in the basal part of the cochlea [10]. After recovering from meningitis (2 weeks after infection), neuronal loss in the spiral ganglion, fibrocytic occlusion of the perilymphatic space, damage to the spiral ligament, and loss of hair cells were the predominant findings, as has been described previously [2, 3].

Once in the perilymphatic spaces of the cochlea, the bacteria can multiply easily because the inner ear is an immunoprivileged site (similar to the brain). One common feature during uncontrolled bacterial proliferation is bacterial autolysis. This leads to the liberation of bacterial cell-wall components and endotoxins, and an immune response is initiated. In this study, we show for the first time that MyD88-dependent pathways play a substantial role in the activation of the immune response during pneumococcal suppurative labyrinthitis. As mentioned above, MyD88 is an intracellular adapter molecule for TLRs. In vitro experiments have shown that peptidoglycan and lipoteichoic acid of pneumococci are mainly recognized by TLR2 [11]. Recently, we demonstrated the significance of TLR2 in murine pneumococcal meningitis by showing that TLR2^−/− mice intracisternally infected with S. pneumoniae have aggravated intracranial complications, more-severe clinical symptoms, and higher bacterial titers, compared with wild-type mice [4]. TLR4, which can bind pneumolysin, might also play an important role in the initiation of the innate immune response during S. pneumoniae infection [12]. Furthermore, MyD88 is involved in the signaling of interleukin (IL)—1β, which is known to play a potent role in immune activation during pneumococcal meningitis [13]. IL-1β expression is dependent on the activation and translocation of NF-κB. Thus, this might provide a positive feedback loop during the induction of the inflammation. Indeed, IL-1β was recently reported in the cochleae of gerbils [14] and mice (S.K. and M.K., unpublished data) with pneumococcal meningitis. Thus, the striking effect of MyD88 deficiency on cochlear inflammation suggests that TLRs and/or IL-1β might play an important role in immune activation during pneumococcal meningitis—associated labyrinthitis.

The reduced inflammatory reaction in the cochlea of MyD88-deficient mice was associated with a diminished development of sensorineural hearing loss, compared with that in wild-type controlmice. Previous studies have already pointed toward the inflammatory response as a significant contributor to the development of cochlear damage and hearing loss in meningitis-associated labyrinthitis [15]. Inflammatory cells develop large amounts of reactive nitrogen and oxygen species during pneumococcal meningitis. In a recent study, we could show that the administration of peroxynitrite scavengers and antioxidative drugs (MnTBAP and N-acetyl-L-cystein) reduces meningitis-associated hearing loss and its histopathologic correlates [10]. Similar results were obtained by the administration of superoxide dismutase [6]. Our observation that the attenuated inflammatory response in the inner ear of MyD88-deficient mice was associated with less hearing impairment further strengthens the concept of a role for the immune response in the development of meningitis-associated hearing loss.

In conclusion, we established the first mouse model of meningitis-associated hearing loss. Using MyD88-deficient mice, we demonstrated a significant role for MyD88 in the development of suppurative inflammation in meningitis-associated labyrinthitis. This study provides convincing evidence that the host immune response significantly contributes to the development of meningitis-associated hearing loss.

• Received August 2, 2006.
• Accepted October 24, 2006.

• © 2007 by the Infectious Diseases Society of America