The sensory epithelium of the cochlea does not contain microvessels 
, yet the density of GJ plaques in cochlear supporting cells is exceptionally high 
. The intercellular coupling provided by these GJs is suspected to play nutritive roles by facilitating local glucose and other metabolite transfers 
. We, therefore, concentrated our effort on comparing differences in GJ-mediated ionic and metabolic coupling in the cochlear supporting cells between the WT and Cx30 null mice. We first examined whether the Cx30 null mutation affects the formation of GJs in this region of the cochlea by immunolabeling for Cx26 and Cx30 and examining the results using the flattened cochlear preparation. Consistent with the genetic deletion of the Gjb6
(Cx30) gene, no Cx30 immunoreactivities were detected (Fig. S1B
). Previous studies using Western blotting also confirmed that these mice do not express the Cx30 protein 
. Extensive Cx26 immunoreactivities clustered apparently as GJ plaques, however, were found in the membrane of cochlear supporting cells of Cx30 null mice ( & Fig. S1A
). Comparing the size of the evident GJ plaques between WT () and Cx30 null () mice, the density of Cx26 immunoreactivities in the Cx30 null mice was apparently decreased. Nonetheless, these presumably homomeric Cx26 GJ plaques were observed in the membrane of all cochlear supporting cells that normally co-express Cx26 and Cx30 (). These results were consistent with previously published data showing that the cellular expression pattern of Cx26 is unaltered in the cochlear sections of Cx30 null mice 
. Next, we quantified the effect of the Cx30 null mutation on intercellular ionic coupling by measuring GJ conductance between adjacent cochlear supporting cells using two-electrode patch clamp recordings conducted with the flattened cochlear preparation (). The intercellular currents recorded in WT mice () were eliminated by GJ blockers (2 mM of octanol () or 200 µM flufenamic acid (FFA, )), indicating that GJs mediated the electrical coupling. A significant amount of GJ-mediated current remained in the Cx30 null mice (). Average GJ conductance (Gj
) derived from the I–V relation was 9.1±2.2 nS (the total number of cells (n) recorded was 10) and 8.3±1.3 nS (n
8) in WT and Cx30 null mice, respectively (). The small reduction of GJ conductance in the Cx30 null mice was statistically not significant (p>0.05). These whole cell recording data indicate that the absence of Cx30 did not appreciably reduce ionic coupling among cochlear supporting cells.
Assessment of ionic coupling in the supporting cells of WT and Cx30 null mice.
We next investigated whether the absence of Cx30 affected GJ-mediated metabolic coupling among cochlear supporting cells by using dye diffusion assays performed with the flattened cochlear preparation. This approach was first validated by performing dye injections with propidium iodide (PI) into single cells at various locations in the cochlea of WT mice. PI is a well-characterized fluorescent dye known to pass through cochlear GJs 
. The dye was readily transferred to multiple cells when single cell injections were made into fibrocytes located in the lateral wall (, n
10), Claudius cells (, n
18) and outer sulcus cells (, n
26). As a negative control, we injected PI into the marginal cells since these cells are not coupled by GJs 
. For all the marginal cells we tested (n
8), the dye stayed in the injected cell for more than 30 minutes (). Two other fluorescent dyes (e.g., Lucifer yellow (LY), 2-NBDG) used in this study showed similar results when injected into marginal cells. The failure of intercellular transfer of the membrane impermeable dye among marginal cells supports the hypothesis that GJs mediate the intercellular diffusion of these fluorescent dyes.
Validation of the dye diffusion assay performed with the flattened cochlear preparation.
We extended the in situ
dye injection assay by injecting fluorescent dyes possessing different charges and molecular weights. In Claudius and outer sulcus cells of WT mice, both negatively (LY, ) and positively charged (PI, ) dyes diffused to many cells in a few minutes. Alexa 568 (MW
−1), in contrast, showed minimal intercellular diffusion among the WT cochlear supporting cells (, n
10). This dye apparently represented an upper limit for a negative-charged dye to be transferred across cochlear GJs. Alternatively, a fluorescent dye commonly used to evaluate intercellular glucose transfer, 2-NBDG 
, showed remarkably fast diffusion among cochlear supporting cells. Both the rate and extent of diffusion of 2-NBDG were dramatically greater than those demonstrated by LY and PI (, ).
Time-lapse recordings of intercellular dye transfer in WT and Cx30 null mice after fluorescent dye was injected into a single outer sulcus cell.
After validating the in situ dye injection assay, we compared the efficiency of intercellular biochemical coupling among cochlear supporting cells of WT and Cx30 null mice. To quantify differences in metabolic coupling among these cells, we performed time-lapsed recordings to assess the time course of PI () and 2-NBDG () diffusion. Data obtained from WT and Cx30 null mice at various time points after the start of dye injections showed that the extent of the fluorescent dye diffusion gradually increased with time (). Quantifying these data by plotting the number of dye recipient cells as a function of time after the injection indicates that for both the positive (PI, ) and negative (LY, ) dyes, the number of cells receiving dye transfer in the sensory epithelium of Cx30 null mice was consistently less at all time points. 2-NBDG showed an exceptional ability to pass through cochlear GJs among the supporting cells. Injections of 2-NBDG into a single Claudius cell quickly diffused to nearly 40 cells within 1 minute (). On average, about 80 cells received the D-glucose analogue through the GJ-mediated diffusion in about 20 minutes in the WT mice. In contrast, less than 30 cells received 2-NBDG in the cochlea of Cx30 null mice during the same time period (). Similar differences were observed between the WT and Cx30 null mice when 2-NBDG was injected into a single outer sulcus cell (). These data demonstrated that Cx30 null mutation caused a deficit in GJ-mediated metabolite transfer among cells in the sensory epithelium of the cochlea.
Comparison of the number of cells (y-axis) received fluorescent dye through GJ-mediated diffusion in WT and Cx30 null mice.
To directly test whether GJs facilitated glucose uptake from the blood circulation into the cochlear supporting cells, we perfused the 2-NBDG through the cardiovascular system. Uptake of 2-NBDG after systemic application was examined by measuring the intensity of 2-NBDG fluorescence in the supporting cells. In the WT mice, 2-NBDG fluorescence was clearly observed in the cochlear pillar cells (arrow in ), Claudius cells (region labeled 3 in ), outer sulcus cells (region labeled 4) and in the region of the stria vascularis (labeled 6 in the ). NBDG fluorescence was relatively weak in the inner (region labeled 1 in ) and outer (region labeled 2 in ) hair cell regions. The specificity of 2-NBDG fluorescent signal was confirmed by using excess non-fluorescent glucose to compete with the 2-NBDG. A large molar ratio of glucose to 2-NBDG efficiently diminished the 2-NBDG fluorescence (). The loading of 2-NBDG was dramatically reduced in the cochlea of Cx30 null mice () or when the GJs were blocked by octanol (2 mM, ), supporting the hypothesis that the 2-NBDG uptake was dependent on the normal function of GJs. Quantification of the results showed that the 2-NBDG loading was reduced to 41.5±3.8% and 54.3±3.8% (n
5) in the Claudius and outer sulcus cells, respectively, of WT control mice after GJs were blocked by 2 mM octanol (). In the cochlea of Cx30 null mice where homomeric Cx26 GJs were still present (), averaged 2-NBDG loading (n
5) was 65.1±8.3% and 37.3±9.4% of the WT control values in the Claudius cells and outer sulcus cells, respectively. These values were further reduced to 6.3±4.9% and 3.6±5.2% (n
5) when excess glucose was used to compete with the 2-NBDG in the systemic applications. In summary, our results indicate that GJs are required for efficient transfer of glucose from the blood circulation into the cochlear supporting cells, and the Cx30 null mutation significantly decreases this important GJ-mediated metabolic coupling in sensory epithelium of the cochlea.
Use of 2-NBDG as a fluorescent tracer of glucose to study the dependence of glucose uptake by cochlear supporting cells on the GJs.
Intracellular glucose deficiency is known to affect mitochondria function and increase the generation of reactive oxygen species (ROS) 
. Therefore, we compared the level of ROS in the cochlear supporting cells of WT and Cx30 null mice (). After ROS generation related intracellular fluorescent intensity was stabilized, we compared the H2
DCFDA fluorescence from randomly-selected Claudius and outer sulcus cells (n
278, from 6 mice). Comparing to WT mice, average intensity of CM-H2
DCFDA fluorescence in these cells was 50.3% higher in the Cx30 null mice (), which is statistically highly significant (p<0.01).
Comparison of intracellular ROS levels in the cochlear supporting cells of WT and Cx30 null mice.