Figure 1.
Generation of Cx57-Cre mice and exclusive Cre expression in horizontal cells.
A, Targeting scheme. The coding region of the wild-type Cx57 gene is located in exon 2 and 3. After electroporating the targeting vector (pKW-DTR-fret-Cre), a large part of exon 2 was replaced by a DTR-eGFP construct, flanked by frt sites, and the coding sequence for the NLS-Cre recombinase followed by a neomycin resistance cassette (neoR). After Flp- mediated recombination, the coding sequence for the NLS-Cre recombinase gets under the control of the Cx57 promoter. DTR-eGFP: fusion protein of the diphtheria toxin receptor (DTR) and eGFP; NLS: nuclear localization signal. B, Southern blot analysis of ScaI-digested genomic DNA to test for correct recombination. A signal of 7.2 kb confirms the correct targeting of the Cx57-DTRfrtCre construct to the Cx57 locus. The functioning of the frt sites and successful Flp-mediated excision of DTR were verified by a signal of 8.7 kb. C-E, Projections of confocal stacks of vertical retina sections from Cx57Cre/Cre mice labeled with antibodies against calbindin (magenta; C, E), a marker for mouse horizontal cells, and Cre recombinase (green; D, E). Only calbindin-positive horizontal cells show Cre recombinase immunoreactivity in their nuclei (arrows) in the distal INL; no other retinal cell type is Cre-positive. F, The Nomarski image shows the normal retinal layering. ONL/INL, outer/inner nuclear layer; OPL/IPL, outer/inner plexiform layer; GCL, ganglion cell layer. Scale bar: C-F, 20 µm.
Figure 2.
Cre expression in GluA4fl/fl:Cx57+/Cre mice leads to reduced GluA4 immunoreactivity in the outer retina.
A, Generation of the conditional GluA4-deficient mice by crossing Cx57+/Cre mice with GluA4fl/fl mice [22]. Expression of Cre recombinase leads to Cre-mediated ablation of the GluA4 gene only in Cx57-positive cells. B-D, Confocal projections of the OPL and distal INL from a GluA4fl/fl:Cx57+/Cre mouse labeled for calbindin (magenta; B, D) and Cre recombinase (green; C, D). Each calbindin-positive horizontal cell is also positive for Cre recombinase. Arrowheads point to blood vessels labeled by unspecific binding of the secondary anti-mouse antibody. E-J, Double labeling for calbindin (magenta) and GluA4 (green) in GluA4fl/fl (E-G) and GluA4fl/fl:Cx57+/Cre mice (H-J) confirmed the selective deletion of GluA4 in the OPL (arrows) but not the IPL of GluA4fl/fl:Cx57+/Cre mice. OPL/IPL, outer/inner plexiform layer; INL, inner nuclear layer; GCL, ganglion cell layer. Scale bars: B-D, 40 µm; E-J: 20 µm.
Figure 3.
GluA4 and Cx57 expression are reduced in GluA4fl/fl:Cx57+/Cre mice.
A-L, Expression of GluA4 (A-D), Cx57 (E-H), and GluA2/3 (I-L) were quantified in mice in GluA4-expressing (A, B, E, F, I, J) and GluA4-deficient horizontal cells (C, D, G, H, K, L) labeled with anti-calbindin antibodies (B, D, F, H, J, L). Maximum projections of two confocal images are shown. M, The number of immunoreactive puncta per 500 µm2 was determined in both genotypes and normalized to the counts in GluA4fl/fl control mice. Cx57 and GluA4 immunoreactivity were reduced in GluA4-deficient horizontal cells (GluA4fl/fl:Cx57+/Cre) while GluA2/3 immunoreactivity was not significantly different between genotypes. Similar results were obtained when the area fraction was determined that was occupied by immunoreactive puncta (N). Values in M and N are given as mean ± SD. **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. n = 6 to 10. Scale bars: A-L: 20 µm.
Figure 4.
GluA4 protein levels are reduced in retina homogenates from GluA4fl/fl:Cx57+/Cre mice.
A, Western blot analysis of GluA4 expression in retina homogenates from GluA4fl/fl and GluA4fl/fl:Cx57+/Cre mice. GluA4-specific antibodies detected a single protein band migrating at ∼100 kDa (arrowhead). Expression levels are lower in GluA4-deficient mice than in controls. B, To control for total protein amount, the blot was stripped and incubated with antibodies against the housekeeping protein α-tubulin. A single immunoreactive band of ∼50 kDa was detected (arrowhead) revealing that protein amounts were comparable in homogenates from both genotypes. Molecular mass marker proteins are indicated on the right (kDa).
Figure 5.
Glutamate- and AMPA-induced currents are reduced in horizontal cells lacking the GluA4 subunit.
A, B, Representative current traces recorded from GluA4-expressing (A) and GluA4-deficient horizontal cell somata (B) at a holding potential of -70 mV. No significant inward current was elicited by Ringer’s solution. However, inward currents were induced by glutamate (Glu, 1 mM) or AMPA application (100 µM). Drug-induced currents were still present but strongly reduced in GluA4-deficient horizontal cells (B). C, Quantification of the glutamate- and AMPA-induced currents in both genotypes. In GluA4-deficient horizontal cells, current amplitudes were significantly reduced by about 75% compared to control cells. Please note that current amplitudes are normalized to 10 pF capacitance of cell membrane to account for differences in cell size. Small circles represent values for individual cells whereas large, black-filled circles represent the means; error bars show SD. n specifies the number of measured cells. D, Dose-response curve for glutamate, measured for both genotypes (GluA4fl/fl: n = 11; GluA4fl/fl:Cx57+/Cre: n = 7). Data were normalized to Imax and fitted to the Hill equation to obtain EC50 values (GluA4fl/fl: 128 ±38 µM; GluA4fl/fl:Cx57+/Cre: 99 ±32 µM) and Hill coefficients (GluA4fl/fl: 1.71 ±0.15; GluA4fl/fl:Cx57+/Cre: 1.81 ±0.57). Differences were not significant (EC50: p = 0.12; Hill: p = 0.56). Values are given as mean ± SD. ****, p < 0.0001.
Figure 6.
The residual glutamate-induced current in GluA4-deficient horizontal cells is mainly mediated by AMPA receptors.
A-D, Representative current traces recorded from GluA4-expressing (A, C) and GluA4-deficient horizontal cell somata (B, D) at a holding potential of –70 mV. When the AMPA receptor antagonist GYKI52466 (100 µM) was present, glutamate (Glu, 1 mM; A, B, E) only induced a detectable current in GluA4-expressing cells (A,E) but not in GluA4-deficient cells (B, E). AMPA-induced responses were completely abolished by GYKI52466 in both genotypes (C-E). E, Quantification of glutamate- and AMPA-induced currents upon GYKI52466 application. GYKI52466 reduced the amplitudes of glutamate- and AMPA-induced currents by 98.9% in controls and 99.3% in GluA4-deficient mice, indicating that most of the glutamate-induced current is mediated by AMPA receptors. Current amplitudes in E are normalized to 10 pF capacitance of cell membrane to account for differences in cell size. Small circles represent values for individual cells; large, black filled circles represent the means; error bars show SD. n specifies the number of measured cells. *, p < 0.05.
Figure 7.
Concanavalin A makes a small contribution of kainate receptors to glutamate responses visible.
A, B, Representative current traces recorded from GluA4-expressing (A) and GluA4-deficient horizontal cell somata (B) upon application of the glutamate and GYKI52466 while concanavalin A (ConA, 1 mg/ml), known to potentiate kainate receptor-mediated currents, was continuously present in Ringer’s solution. In both genotypes, glutamate induced a discernible current when AMPA receptors were blocked, suggesting a small contribution of kainate receptors to the overall glutamate response. C, Quantification of glutamate-induced currents in the presence of ConA. In both genotypes, responses to glutamate in the presence of GYKI52466 and ConA were significantly higher than to Ringer’s application. Current amplitudes in C are normalized to 10 pF capacitance of cell membrane to account for differences in cell size. Small circles represent values for individual cells; large, black filled circles represent the means; error bars show SD. n specifies the number of measured cells. *, p < 0.05; **, p < 0.01.
Table 1.
Primers used for mouse genotyping.