Bacterial Cytolysin Perturbs Round Window Membrane Permeability Barrier In Vivo: Possible Cause of Sensorineural Hearing Loss in Acute Otitis Media

Abstract

The passage of radioiodinated streptolysin-O (SLO) and albumin through the round window membrane (RWM) was studied in vivo. When applied to the middle ear, SLO became quantitatively entrapped in this compartment and no passage to the cochlea occurred. However, flux of radioiodinated albumin through the toxin-damaged RWM was observed. We propose that the passage of noxious macromolecules, such as proteases, from a purulent middle-ear effusion may be facilitated by pore-forming toxins, resulting in cochlear damage and sensorineural hearing loss.

A serious sequela of otitis media (OM) is damage to the inner ear, leading to sensorineural hearing loss (19, 22, 23), which is limited to the cochlea basal turn (21). The etiology of sensorineural hearing loss has remained elusive, but previous investigations indicate that permeability of the three-layered round window membrane (RWM) is of central importance (12, 25). Some studies have addressed the effects of various bacterial products on the permeability of the RWM (11–16), but bacterial cytolysins have largely been absent from this list. To first provide information on the possible effects of these toxins, we previously analyzed the effects of streptolysin-O (SLO), the major cytotoxin elaborated by Streptococcus pyogenes A, (1, 5, 6) on the permeability of resected RWMs in an experimental system originally developed by Lundman et al. (17). SLO was employed because this toxin is related molecularly to pneumolysin, the major cytotoxin of Streptococcus pneumoniae (1, 3, 5, 6, 8, 9, 24, 26) and is available in highly purified form in our laboratory (4, 7, 27). When the RWM surrounded by an intact bony frame was excised and embedded in a Plexiglas sheet separating two buffer chambers, low concentrations of SLO provoked rapid breakdown of the RWM permeability barrier. Fluxes of Na⁺, [¹⁴C]mannitol, serum proteins, and radioiodinated SLO were demonstrable (10).

It now became important to determine whether a flux of macromolecules across toxin-damaged RWMs might indeed occur in vivo. To address this question, we employed audioradiography and measurements of radioactivity to detect the passage of iodinated SLO and albumin, respectively, from the middle to the inner ear.

Recombinant SLO was purified from Escherichia coli (27). Solutions containing 20 μg of native SLO per ml contained no detectable lipopolysaccharide as determined by a quantitative Limulus assay (Kabi Vitrum Diagnostica, Stockholm, Sweden). The proteins were stored lyophilized and reconstituted with buffer plus 1% bovine serum albumin and 2 mM dithiothreitol. The toxin was radioiodinated to a specific activity of 10 mCi/mg under retention of functional activity (20).

Thirty-five healthy guinea pigs (body weight, 250 to 400 g) were used. Ear infection was excluded by the examination of the external auditory canal and the tympanic membrane. Operations were undertaken under general anesthesia with 80 mg of ketamine (50%) and 4 mg of thiazine (5%) per kg of body weight and 0.1 ml of fentanyl. The bulla was prepared and carefully opened by a standard retroauricular approach (2). The tympanic membrane and auditory ossicles, except the stapes, were removed, and the round window niche was exposed.

A total of 0.1 μCi of radioiodinated SLO plus 0.08 and 0.4 μg of unlabeled SLO were applied to the round window niche in a volume of 0.1 ml of Hanks’ balanced salt solution (HBSS) containing 1% bovine serum albumin and 2 mM dithiothreitol. After 2 h, the round window niche was washed with HBSS and the animal was sacrificed. The cochlea was resected and fixed with an irrigation of 2% glutaraldehyde from the round to the oval window (18). Then the cochlea was fixed again, decalcified, embedded in paraffin, and cut into 10-μm sections. The sections were dipped in a Kodak NTD 3 photoemulsion, incubated for 14 days, and stained with methylene blue. In positive controls, radioiodinated SLO was directly applied into the cochlea and so had immediate access to the cochlear epithelia and the organ of Corti. Experiments with unlabelled SLO without tracer served as negative controls.

In analogous experiments, 2,000,000 cpm of radioiodinated albumin plus 0.08, 0.4, or 2 μg of unlabelled SLO were applied in a volume of 0.1 ml to the round window niche. After 2 h, the round window niche was washed with HBSS and the animal was sacrificed. The cochlear endo- and perilymph were collected with a Hamilton syringe, and radioactivity was determined in a gamma counter (Packard Cobra II). Negative controls included animals that received radioiodinated albumin into the round window niche without the use of SLO.

In the first experiments, we sampled the endo- and perilymph of the cochlea following application of radioiodinated SLO to the RWM in the presence of 0.08 to 2 μg of unlabelled toxin and albumin. However, radioactivity was never recovered, indicating that the toxin could not pass the membrane.

Autoradiography was then used to locate the tracer. As shown in control autoradiographs, where radioiodinated SLO was directly applied into the cochlea, the tracer produced a homogeneous distribution of grains along all cochlear epithelia and the organ of Corti (Fig. 1). This demonstrated that SLO could bind to target cell substrates in the inner ear. When the SLO tracer was applied with 0.08 (n = 3) or 0.4 μg (n = 11) of unlabelled SLO into the round window niche, however, no radioactivity could be detected in the cochlea, and all of the SLO tracer was detected in the middle-ear compartment (Fig. 2). This finding was consistent with the lack of detectable flux of the radiotracer to the inner ear and showed that SLO quantitatively binds to target cells and becomes entrapped in the middle-ear mucosa.

FIG. 1.

Control autoradiography after direct application of radioiodinated SLO into the cochlea. A homogeneous distribution of the tracer (grains are indicated by an X) along all cochlear epithelia and the organ of Corti (C) is revealed. E, endolymphatic space; P, perilymphatic space (magnification, ×200). 

FIG. 2.

SLO applied to the RWM does not enter the inner ear. After the application of radioiodinated SLO plus 0.4 μg of unlabelled SLO into the round window niche, no flow of the tracer can be demonstrated. Only natural background activity is shown in the inner ear (IE). The stria vascularis (SV) is dark due to its content of melanocytes. Outside the cochlea, grains (indicated by an X) are revealed, the middle-ear mucosa has entrapped the tracer. ME, middle ear. Magnification, ×200. 

Albumin contrasts with SLO in displaying no affinity to cellular structures. To assess flux of this molecule, direct measurement of radioactivity in the peri- and endolymph was undertaken. No flux of albumin occurred when the tracer was applied with 0.08 μg of SLO to the middle ear. However, flux of the tracer was observed when 0.4 μg of SLO was applied, and flux was massive when 2 μg of SLO was employed (Fig. 3). As mentioned above, no measurable flux of radiolabelled SLO was observed under the same experimental conditions.

FIG. 3.

Measurement of flux of radioiodinated albumin across the RWMs. Radiolabelled albumin (2,000,000 cpm) was applied in a volume of 0.1 ml together with 0.08, 0.4, or 2 μg of SLO to the round window niche. After 2 h, radioactivity in the endolymph was determined. Values are means ± standard deviations (error bars) (n = 4 in each group). The passage of albumin was observed when 0.4 and 2.0 μg of SLO were applied with albumin.

These findings are the first to establish that a bacterial cytolysin can damage the RWM, causing leakage of macromolecules to the perilymph in situ. A priori, the concentration of SLO required to cause such massive breakdown of the RWM permeability barrier may appear high (4 to 20 μg/ml). However, a concentration of 20 μg/ml equates with only about 100 molecules of toxin per μm³, the approximate cytoplasmic volume of a gram-positive bacterium. In the case of pneumolysin, which is retained in the cytoplasm, far-higher concentrations must be present within the bacteria. It is thus of special interest first that pneumolysin is structurally and functionally similar to SLO and second that pneumolysin is produced by all pneumococcal strains, which are the most common cause of bacterial OM. In actual infections, it is therefore readily conceivable that autolysing bacteria located in immediate contact with the RWM will release sufficient quantities of their toxin to cause the gross permeability defects observed in this study. Ionic disequilibrium caused by ion fluxes could lead to severe disturbances of inner-ear functions, and the passage of macromolecules, such as bacterial proteases, may cause toxic damage to the organ of Corti. A simple explanation for the clinical syndrome of sensorineural hearing loss occurring during pneumococcal OM thus emerges. The present results indicate that the passage of cytolysins such as SLO or pneumolysin themselves would not be expected to contribute to disturbances of inner-ear function, because the toxins probably will become quantitatively entrapped in the middle ear due to their binding to cells lining this cavity. In the in vitro system used previously (10), the entire epithelial lining of the middle ear was absent, and the RWM alone may have been unable to trap all of the tracer, leading to the deviant finding compared to that of the present study. Thus, SLO could become a tool to transiently permeabilize the RWM, making possible the introduction of various agents into the inner ear.