Lipopolysaccharide decreases cochlear blood flow dose dependently in a guinea pig animal model via TNF signaling

To evaluate the effect of Lipopolysaccharide (LPS), a bacterial endotoxin on cochlear microcirculation, and its mode of action.

Additionally, people who suffer chronic hearing loss from otitis media are more likely to suffer from depression or anxiety-related issues. 4 Otitis media is commonly caused by a variety of bacteria, among the most common are the gram-positive bacteria Streptococcus pyogenes or pneumoniae, Staphylococcus spp., and the gramnegative bacteria Klebsiella spp., Pseudomonas spp., Proteus spp., Haemophilus influenzae, Moraxella catarrhalis, and Escherichia coli. 5,6 Several of these express lipopolysaccharide (LPS), an endotoxin which is situated in the outer membrane of gram-negative bacteria, and which is a considerable inflammation mediator. It has been established that injection of lipopolysaccharide into the middle ear is a viable animal model for acuta bacterial otitis media. 7,8 When it comes to permanent sensorineural hearing loss, it is commonly believed that this is caused by transition of toxins though the round 9 and oval 10,11 window membrane, causing inflammation of the cochlea (suppurative labyrinthitis). 12 These findings had been reproduced in animal models of otitis media with suppurative labyrinthitis. 13 Cochlear microcirculation is an important functional parameter of the cochlea-and impairment of cochlear microcirculation coincides with decreases in partial oxygen pressure 14 and increases in hearing thresholds. 15,16 There has been evidence that permanent decreases in cochlear microcirculation form the common final pathophysiological pathway of numerous inner ear pathologies, including sudden sensorineural hearing loss, 16 endolymphatic hydrops 17,18 , and noise trauma. 15 Until recently, the pathologies of the inner ear were strictly divided into those that were considered to be of a microcirculatory origin and those that were of an inflammatory origin. However, it has been observed that inflammatory cytokines like tumor necrosis factor (TNF) may impair cochlear blood flow, rendering this division of entities to be inaccurate. 19 Hence, we decided to investigate the effect of LPS on cochlear microcirculation and its mode of action.

| Animals
Animals were healthy female Dunkin-Hartley guinea pigs purchased from Envigo Laboratories, Venray, The Netherlands, aged 8-12 weeks, with a bodyweight of 250-400 g. After an acclimation period of 1 week, animals were included in the experiment.

| Anesthesia and surgical preparation
The animal model used has been described 20 and used 16,21-23 extensively. Anesthesia was induced by intramuscular injection of 85 mg/ kg bodyweight (b.w.) of ketamine and 8.5 mg/kg b.w. of xylazine.
Injections were repeated every 30 min with half of the dosage to keep up anesthesia. Once depth of anesthesia was sufficiently deep, as was controlled by corneal and toe-pinching reflexes, the hair above the ear and the neck was mechanically and chemically removed. After this, about 0.5 ml of lidocaine with epinephrine was injected into the tissue and an i.v. catheter was inserted into the jugular vein. Then, the ear and the muscle tissue covering the temporal bone were removed. The lateral temporal bone was then removed as well as the ossicles and thus the cochlea exposed.
Vessels covering the cochlea were removed by simply wiping them off with a microsponge, and a rectangular window of approximately 400 × 400 µm was then carefully carved into the cochlea, at the height of the second turn. Then, fluoresceinisothiocyanatedextrane (FITC, Molecular weight of 500.000, purchased from Sigma-Aldrich, Deisenhofen, Germany) was injected intravenously as a contrast material. Thus, the blood flow of the stria vascularis could be visualized.

| Microscopy and calculation of cochlear blood flow
Cochlear microcirculation was then recorded using a Leica M205FA binocular microscope with a Leica EL 6000 light source and the proprietary Leica Application Suite (all Leica, Wetzlar, Germany) software. (Video S1 Figure 1A,B) The obtained videos were then stored digitally for later off-line quantification of capillary diameter using CapImage (Dr. Zeintl Engineering, Heidelberg, Germany)-a software that has been designed specifically for this purpose. ( Figure 1C) Breathing excursions are corrected for manually.
Initially, three representative vessels were chosen at random. In these vessels, at each point in time, capillary diameter and intravascular blood velocity were quantified by measuring each value thrice and then taking the average of the three measurements. Cochlear blood flow was then calculated by the formula that was specifically proposed by Wayland 24 for this purpose: where v is the velocity of the blood flow (in µm/s), d is the diameter of the representative capillary (in µm), and q is the blood flow (in pl/s). To correct for interindividual differences, the changes for each animal are reported as relative changes in blood flow, given as arbitrary units (AU).

| Experimental protocol
Overall, 25 guinea pigs were included in this study. Each one was randomly assigned into one of five groups of five each. After the surgical preparation had been finished, basal values were recorded for about 3 min before treatment started. Afterward, the first treatment was then applied topically for 20 min. To achieve this, the bulla was filled with the dissolved LPS (or sterile saline 0.9% as carrier, respectively) until the window carved into the cochlea was covered with a layer of fluid. First treatment consisted either of Placebo (sterile saline 0.9%) or LPS (1 mg, 10 µg, or 100 ng per ml dissolved in sterile saline 0.9%) in various concentrations for 20 min. Then, the bulla was rinsed with saline for approximately 10 min and cochlear microcirculation was then quantified.
Following this, the same concentration of LPS was then again applied for another 20 min. Finally, the bulla was rinsed again for 10 min and microcirculation was quantified again. Following this, the animals were euthanized.
In a second part of the experiment, the lowest concentration that yielded a significant decrease in cochlear microcirculation after 30 min of exposure was chosen. Subsequently, the stria vascularis was topically pretreated with etanercept, a TNF-antagonist, as has been reported previously, 25 for 20 min in the same manner as LPS was applied and cochlear microcirculation was quantified. Following this, the lowest concentration of LPS that yielded significant results after 20 min exposure (10 µg/ml) was applied for 20 min and cochlear microcirculation was quantified again.

| Statistics
To detect significant differences, we fitted linear mixed models that included a random effect for the animal and were estimated using a restricted maximum likelihood approach. A p value <0.05 was considered to be significant. The software used for this was Project R (Build 3.2.5 for Windows, The R Project for Statistical Computing, http://www.r-proje ct.org/).

| Effect of Placebo on cochlear microcirculation
Topical application of sterile saline leads to no significant changes in cochlear microcirculation, which remained at .998 arbitrary units (AU) ± .060 AU. Subsequent application of placebo again showed no change in cochlear microcirculation, which was at .996 ± .074 AU.

| Effect of 100 ng/ml LPS on cochlear microcirculation
Application of 100 ng/ml LPS caused cochlear microcirculation to drop to .968 ± .148 AU and subsequently to .893 ± .163 AU after

| Effect of Etanercept / LPS on cochlear microcirculation
Application of etanercept (5 ng/ml) caused cochlear microcirculation to drop .963 ± .155 AU and subsequently after application of 10 µg LPS to .931 ± .147 AU. These changes were not significantly different compared to the group that had been treated with placebo.
The original dataset including the measures of capillary diameter and intravascular blood velocity is available as online supplemental material to this manuscript. In addition to this, we were able to show that the effect of LPS on cochlear microcirculation shows dose-dependent effect.

Independent of how high the concentration of LPS was applied,
we never observed a decrease greater than 20% compared to basal values. Yet this effect was seen faster in those guinea pigs treated with higher concentrations of LPS. This suggests that the effect of LPS on microcirculation is mediated by a cellular receptor that is activated in a stronger fashion when high concentrations of LPS are present. Fittingly, we have also been able to prove that the effects of LPS on cochlear microcirculation can be abrogated by previous application of etanercept. Hence, it seems highly probable that the effects of LPS are-at least partially-mediated by tumor necrosis factor. TNF is regularly released as a response to stress by fibrocytes of the spiral ligament, 32 as has also been described for fibrocytes in other tissues. 33 TNF has been shown to cause cochlear pericytes to contract and thus reduce cochlear blood flow. 34 Additionally, LPS may directly affect cochlear pericytes, as has been observed in different tissues 35 as well as in cochlear pericytes after TNF exposure in vitro. 36 It has also been shown that in the central nervous system, pericytes are among the first cells to sustain damage in persistent hypoxia 37,38 and-if hypoxia persists long enough-may enter a rigor mortis like state, thus persistently impairing capillary blood flow. 39 Considering the similar physiological properties, a similar mechanism seems well probable. In return, a persistent decrease in cochlear While this is certainly not the only way otitis media may cause cochlear damage in suppurative labyrinthitis, it seems probable that persisting impairment of cochlear microcirculation plays a major role. This view is further supported by the fact that impairment of cochlear microcirculation has also been suggested to be causative of sudden sensorineural hearing loss 16 cycle may also prove useful in these cochlear pathologies.

ACK N OWLED G EM ENTS
Open Access funding enabled and organized by ProjektDEAL.

CO N FLI C T O F I NTE R E S T
The authors Ihler, Freytag, Kloos, Spiegel, Haubner, Canis, Weiss, and Bertlich declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Original data are available as online supplementary material.