Modulating the water channel AQP4 alters miRNA expression, astrocyte connectivity and water diffusion in the rodent brain

Abstract

Aquaporins (AQPs) facilitate water diffusion through the plasma membrane. Brain aquaporin-4 (AQP4) is present in astrocytes and has critical roles in normal and disease physiology. We previously showed that a 24.9% decrease in AQP4 expression after in vivo silencing resulted in a 45.8% decrease in tissue water mobility as interpreted from magnetic resonance imaging apparent diffusion coefficients (ADC). Similar to previous in vitro studies we show decreased expression of the gap junction protein connexin 43 (Cx43) in vivo after intracortical injection of siAQP4 in the rat. Moreover, siAQP4 induced a loss of dye-coupling between astrocytes in vitro, further demonstrating its effect on gap junctions. In contrast, silencing of Cx43 did not alter the level of AQP4 or water mobility (ADC) in the brain. We hypothesized that siAQP4 has off-target effects on Cx43 expression via modification of miRNA expression. The decreased expression of Cx43 in siAQP4-treated animals was associated with up-regulation of miR224, which is known to target AQP4 and Cx43 expression. This could be one potential molecular mechanism responsible for the effect of siAQP4 on Cx43 expression, and the resultant decrease in astrocyte connectivity and dramatic effects on ADC values and water mobility.

Figure 1

siAQP4 treatment induces a decrease in ADC values explained by decreased levels of AQP4 and Cx43 expression. (A) We propose that decrease of AQP4 after siAQP4 injection affects the gap junction protein Cx43 and astrocyte connectivity. Our hypothesis was that these off-target effects are mediated by miRNAs targeting both AQP4 and Cx43. (B) ADC values were reduced by 45.8% in the cortex 3 days after siAQP4 injection even though AQP4 levels were only decreased by 24.9%.

Figure 2

siAQP4 treatment induces decreased levels of AQP4 and Cx43 expression in primary astrocyte cultures. (A) siAQP4 induced a reduction of AQP4 at 6 and 24 h in primary astrocyte cultures measured by Western blot. The red band at 30 kDa corresponds to AQP4 protein (left panel) and the red band at 43 kDa is the Cx43 protein (right panel). The level of expression of both proteins is normalized to actin (in green at 45 kDa). (B) AQP4 shows a progressive decrease up to 24 h. Cx43 levels were decreased at 24 h, suggesting an indirect effect of siAQP4 on Cx43 expression. (*p < 0.05 compare to CTL for AQP4; **p < 0.05 compare to CTL for Cx43; values are represented as mean ± SEM).

Figure 3

In vivo treatment of siAQP4 induces decreased expression of Cx43 in the cortex. Expression of Cx43 was decreased in the cortex 3 days after siAQP4 treatment as shown by Western blot assay with actin (45 kDa) used as a loading control (A), and by immunohistochemistry (B). In siGLO control animals, Cx43 protein was colocalized with GFAP in the astrocytic endfeet but was almost absent from the cortex of the siAQP4-treated rats (B). AQP4 and Cx43 also colocalize around blood vessels in the cortex of siGLO animals (C) but were both decreased after siAQP4 treatment (D), as shown by immunoreactivity quantification (A.U. Arbitrary Unit; E–F). Scale bars: B: 20 µm, D: 50 µm. *p < 0.05; values are represented as mean ± SEM.

Figure 4

In vivo treatment of siAQP4 does not affect the gap junction protein Cx30. There was no difference in Cx30 staining between siGLO and siAQP4-treated animals as shown by immunohistochemistry in the cortex 3 days after injection (A,B). Quantification of Cx30 immunoreactivity in the cortex confirmed the absence of effect of siAQP4 on Cx30 expression (A.U. Arbitrary Unit). Scale bars: A, B: 50 µm.

Figure 5

Effects of siAQP4 on astrocyte dye-coupling. (A) Scrape loading experiments (white dotted line) on astrocyte cultures revealed the effect of siAQP4 on astrocyte connectivity, evidenced by a decreased spread of Lucifer yellow (green) in the siAQP4-transfected astrocytes. Rhodamine B-Dextran loading (red) was similar and did not diffuse at distance from the site of scrape loading in both conditions. (B) Quantification of the fluorescence of Lucifer yellow confirmed the functional consequence of siAQP4-induced decrease of Cx43 (A.U. Arbitrary Unit). Scale bars: A, B: 50  µm; *p < 0.05; values are represented as mean ± SEM.

Figure 6

Effects of siCx43 on AQP4 and Cx43 expression. Intracortical injection of siCx43 induced a decreased expression of Cx43 and no change in AQP4 expression, as shown by Western blot with tubulin (55 kDa) as a loading control (A.U. Arbitrary Unit; A), and by immunohistochemistry in the cortex (B,C for Cx43, D,E for AQP4). Scale bars: B, C: 50 µm; D, E: 100 µm; *p < 0.05; values are represented as mean ± SEM.

Figure 7

Effects of siCx43 treatment on T2 and ADC values. Representative T2WI and DWI pictures from siGLO and siCx43 treated mice (A). siCx43 treatment did not change the ADC values (B) or the T2 values (C) in contra or ipsilateral cortex when compared to siGLO.

Figure 8

Modulation of miR19a, miR23a, miR130a, miR224, miR381 and miR384-5p expression after siAQP4 and siCx43 treatment. Relative expression of miRNA in the cortex of siAQP4- (A) and siCx43- (B) treated rat pups. RNU6-2 was used as a normalization control. Schematic representation of our hypothesis (C) and results (D) which suggest a siAQP4-induced up-regulation of miR224 which mediates the down-regulation of his target Cx43. Interestingly, siCx43 treatment induced a down-regulation of miR19a and miR224, which in turn, were not able to decrease AQP4 expression; *p < 0.05; values are represented as mean ± SEM.