Aqp9 Gene Deletion Enhances Retinal Ganglion Cell (RGC) Death and Dysfunction Induced by Optic Nerve Crush: Evidence that Aquaporin 9 Acts as an Astrocyte-to-Neuron Lactate Shuttle in Concert with Monocarboxylate Transporters To Support RGC Function and Survival

Abstract

Aquaporin 9 (AQP9) is an aquaglyceroporin that can transport lactate. Accumulating evidence suggests that astrocyte-to-neuron lactate shuttle (ANLS) plays a critical role in energy metabolism in neurons, including retinal ganglion cells (RGCs). To test the hypothesis that AQP9, in concert with monocarboxylate transporters (MCTs), participates in ANLS to maintain function and survival of RGCs, Aqp9-null mice and wild-type (WT) littermates were subjected to optic nerve crush (ONC) with or without intravitreal injection of an MCT2 inhibitor. RGC density was similar between the Aqp9-null mice and WT mice without ONC, while ONC resulted in significantly more RGC density reduction in the Aqp9-null mice than in the WT mice at day 7. Positive scotopic threshold response (pSTR) amplitude values were similar between the two groups without ONC, but were significantly more reduced in the Aqp9-null mice than in the WT mice 7days after ONC. MCT2 inhibitor injection accelerated RGC death and pSTR amplitude reduction only in the WT mice with ONC. Immunolabeling revealed that both RGCs and astrocytes expressed AQP9, that ONC predominantly reduced astrocytic AQP9 expression, and that MCTs 1, 2, and 4 were co-localized with AQP9 at the ganglion cell layer. These retinal MCTs were also co-immunoprecipitated with AQP9 in the WT mice. ONC decreased the co-immunoprecipitation of MCTs 1 and 4, but did not impact co-immunoprecipitation of MCT2. Retinal glucose transporter 1 expression was increased in Aqp9-null mice. Aqp9 gene deletion reduced and increased the intraretinal L-lactate and D-glucose concentrations, respectively. Results suggest that AQP9 acts as the ANLS to maintain function and survival of RGCs.

Keywords: Aquaporin 9; Astrocyte-to-neuron lactate shuttle; Monocarboxylate transporter; Optic nerve crush; Retinal ganglion cells; Scotopic threshold response.

Fig. 1

Physiological condition of Aqp9 KO mice. a Immunostaining of AQP9 in liver. WT mice show intense AQP9 immunoreactivity (IR), whereas in Aqp9 KO mice, IR is almost completely lost. Scale bar indicates 50 μm. b–d Time course of changes of parameters in WT and Aqp9 KO mice. b Body weight. c Blood glucose concentrations. d. Intraocular pressures. Blood glucose levels and intraocular pressures were measured in awake mice between 9 am and 11 am. Error bars indicate standard error of the mean (n = 10 each). N.S., not significant

Fig. 2

Effect of Aqp9 gene deletion and ONC on AQP9 expression in retina. a RT-PCR and 2% agarose gel electrophoresis to confirm Aqp9 gene deletion in Aqp9 KO mice (KO). Gapdh was used as a loading control. b Western blot analysis. The upper panel shows constitutive and decreased expression of AQP9 protein in retinal homogenates of WT and Aqp9 KO mice, respectively. The middle panel compares retinal AQP9 expression between WT mice with sham operation (WT sham) and those that underwent ONC (WT ONC) 1 week prior. The bottom panel depicts β-actin protein expression, which was used as an internal control. c Quantitative comparison of an AQP9/β-actin ratio in the western blotting data between WT (gray solid bar) and Aqp9 KO (hatched bar) mouse retinas. Error bars designate standard error of the mean (SEM). n = 3 each, **p < 0.0001, unpaired t test. d Quantitative comparison of an AQP9/β-actin ratio in the western blotting data between WT mice with sham operation (sham: gray bar) and those that underwent ONC (ONC: blue bar) 1 week prior. Error bars designate SEM. n = 3 each, *p < 0.01, unpaired t test. e AQP9 and RBPMS immunoreactivity (IR) in inner retinas. f AQP9 and GFAP IR in the inner retinas. Scale bars indicate 25 μm

Fig. 3

Expression of AQP9 and MCT1, 2, and 4 and effect of Aqp9 gene deletion and ONC on their expression. a Immunoreactivity (IR) of AQP9 and MCTs in WT and Aqp9 KO mice 7 days after sham operation (WT sham and KO sham, respectively) and ONC (WT ONC and KO ONC, respectively). A scale bar indicates 50 μm. b–d Co-localization of MCT1, MCT2, and MCT4, respectively, with AQP9 in GCL. Scale bars indicate 25 μm. Note that all isoforms of MCTs are expressed in GCL and at least partly co-labeled with AQP9 in WT mice with the sham operation. e Representative western blots for AQP9 and MCTs1, 2, and 4. WT sham, WT mice with a sham operation; KO sham, Aqp9 KO mice with a sham operation; WT ONC, WT mice 7 days after ONC; KO ONC, Aqp9 KO mice 7 days after ONC. β-Actin was used as an internal control. f–h Quantitative analyses for MCT1, MCT2, and MCT4 expression levels, respectively, relative to β-actin levels. Error bars represent standard error of the mean (SEM). n = 5; **p < 0.001, *p < 0.01, ANOVA with the Bonferroni post hoc test. i Immunoprecipitation (IP) of retinal homogenates with an anti-AQP9 antibody. The upper panel shows auto-probing with the anti-AQP9 antibody for the IPs, while the lower panel depicts western blotting for the IPs with MCT1, MCT2, and MCT4 antibodies. Note that all of the MCTs (1, 2, and 4) are detectable in the IPs for AQP9 in WT sham. j–m Quantitative analyses for AQP9, MCT1, MCT2, and MCT4 levels, respectively, in the IPs of AQP9 relative to the respective levels in WT mice with sham operation. Error bars indicate SEM. n = 3; p < 0.001, unpaired t test. n.s., not significant

Fig. 4

Effects of Aqp9 gene deletion, ONC, and intravitreal injection of MCT2 inhibitor on RGC density in WT and KO mice. a Photomicrographs of TUBB3-immunolabeled cells in whole-mounted retinas of WT and Aqp9 KO (KO) mice 7 days after a sham operation (sham) or ONC. The scale bar indicates 25 μm. b Quantitative comparisons of the density of TUBB3-immunolabeled cells between WT and Aqp9 KO mice with or without ONC. Error bars indicate standard error of the mean (SEM). n = 8 each. **p < 0.001 and *p < 0.01, ANOVA with post hoc Bonferroni test. n.s., not significant. c Photomicrographs of retrograde-labeled cells in whole-mounted retinas of WT and Aqp9 KO mice 7 days after ONC or sham operation with FG injection into the superior colliculus. The scale bar indicates 100 μm. d Quantitative comparisons of the density of FG-labeled cells. Error bars designate SEM. The density significantly declined in the WT retina after ONC (n = 8, **p < 0.0001). The reduction was more evident in the Aqp9 KO mouse retinas (n = 8, *p < 0.001). n.s., not significant. e Photomicrographs of TUBB3-immunolabeled cells on flat-mounted retinas. WT or Aqp9 KO mice underwent sham operation with a 1-μL intravitreal injection of 10 mM 4-CIN (4-CIN), ONC with a vehicle injection (ONC + Veh), or ONC with the 4-CIN injection (ONC + 4-CIN), and retinas were dissected 7 days later. The scale bar indicates 25 μm. f Quantitative comparisons of TUBB3-immunolabeled cell densities among different mouse conditions. Error bars indicate SEM. n = 6; **p < 0.0001, *p < 0.05, ANOVA with post hoc Bonferroni test. Veh, vehicle; n.s., not significant. Note that the 4-CIN injection alone did not impact the TUBB3-immunoreactive cell density either in WT or Aqp9 KO mice and that an additive effect of 4-CIN in the reduced TUBB3-immunoreactive cell density after ONC was observed in WT mice, but not in Aqp9 KO mice

Fig. 5

Dark-adapted ERG responses elicited at a light intensity of − 0.15 log sc td s a Representative superimposed ERG responses that were recorded from a WT mouse with sham operation (a), a WT mouse with ONC 7 days prior (b), an Aqp9 KO mouse with sham operation (c), and an Aqp9 KO mouse after ONC (d). Note similar a- and b-waves of ERGs among the four groups of mice. b Representative ERG responses superimposed that were recorded from a WT mouse and an Aqp9 KO mouse with four different treatments, i.e., sham operation (a), ONC (b), sham operation and intravitreal 4-CIN injection (c), and ONC and 4-CIN injection (d). Note again that there are no apparent differences in a- and b-wave latencies and amplitudes among the treatments in both WT and Aqp9 KO mice. c Quantitative comparisons of latencies of a- and b-waves among groups. d Quantitative comparisons of amplitudes of a- and b-waves among groups. Error bars indicate standard error of the mean. n.s., not significant (n = 6, ANOVA)

Fig. 6

Scotopic threshold responses (STRs) elicited by a series of incremental dim light stimuli. a Representative series of superimposed STRs that were recorded from a WT mouse with sham operation (black waves), a WT mouse with ONC (blue waves), an Aqp9 KO (KO) mouse with sham operation (gold waves), and an Aqp9 KO mouse after ONC (purple waves). Note that amplitudes of positive STRs (black arrow) are unequivocally reduced in both WT and Aqp9 KO mice with ONC compared to those with sham operation at light intensities of − 5.1, − 4.6, and − 4.1 log sc td s. b Representative series of superimposed STRs that were recorded from a WT mouse and an Aqp9 KO mouse with four different treatments: sham operation (black waves for the WT mouse and gold waves for the Aqp9 KO mouse), ONC (blue waves for the WT mouse and purple waves for the Aqp9 KO mouse), intravitreal 4-CIN injection (green waves for both WT and Aqp9 KO mice), and ONC and 4-CIN injection (red waves for both mice). The positive STRs are more reduced for the WT mouse with ONC and 4-CIN injection compared with the WT mouse with ONC alone, while the negative STRs (red arrow) are similar among the WT mice with the four different treatments. In contrast, the reduction of the positive STRs in the Aqp9 KO mice with ONC and 4-CIN injection is similar to that in the Aqp9 KO mice with ONC alone, while the negative STRs are reduced in the Aqp9 KO mice with ONC alone and those with ONC and 4-CIN injection at the light intensities of − 5.1, − 4.6, and − 4.1 log sc td s. c Quantitative comparisons of amplitudes of pSTR and nSTR among groups. Error bars indicate standard error of the mean. n = 6; **p < 0.01, *p < 0.05, ANOVA with post hoc Bonferroni test. n.s., not significant

Fig. 7

Upregulation of GLUT 1 and GLUT3 expression by Aqp9 gene deletion and ONC. a Immunofluorescence for GLUT1 and GLUT3 in wild-type and Aqp9 KO mice 7 days after sham operation (WT sham and KO sham, respectively) and ONC (WT ONC and KO ONC, respectively). The scale bar indicates 50 μm. b Western blots for GLUT1 and GLUT3 protein expression levels in WT and Aqp9 KO mice with and without ONC. β-Actin was used as a control. c, d Quantitative analyses for GLUT1 and GLUT3, respectively, relative to β-actin intensity. Error bars indicate standard error of the mean. n = 3; *p < 0.01, ANOVA with post hoc Bonferroni test. n.s., not significant

Fig. 8

Relative ratios of l-lactate and d-glucose concentrations in mouse retinas measured by colorimetric assays. a Retinal l-lactate concentrations in mice with different conditions relative to those in WT mice that underwent sham operation with an intravitreal injection of vehicle (control). b Retinal d-glucose concentrations in mice with different conditions relative to d-glucose concentrations in controls. WT, wild-type mice; KO, Aqp9 knockout mice; ONC, optic nerve crush performed 7 days prior; 4-CIN, α-cyano-4-hydroxycinnamate. Error bars indicate standard error of the mean. n = 8; **p < 0.01, *p < 0.05, ANOVA with post hoc Bonferroni test. n.s., not significant

Fig. 9

The energy transportation in normal (a) and optic nerve crush (ONC) (b) between astrocytes and retinal ganglion cells (RGCs). Aquaporin 9 (AQP9) and monocarboxylate transporters (MCTs) along with lactate transport from astrocyte to RGCs. ONC lowers the expression of AQP9 and all isoforms of MCT and disrupts interactions of AQP9 with MCTs1 and 4, which lead to the lower concentration of lactate in the retinal tissue. To compensate for reduced energy supply via lactate, glucose transporter (GLUT) expression is upregulated to increase the direct uptake of glucose into RGCs

Fig. 10

Expression of Brn3a/GFAP and monocarboxylate transporters (MCT)1, 2, and 4 in ganglion cell layer (GCL). a MCTs and Brn3a immunoreactivity (IR) in inner retinas. b MCTs and GFAP immunoreactivity (IR) in inner retinas. Scale bars indicate 50 μm