Deficient glucose and glutamine metabolism in Aralar/AGC1/Slc25a12 knockout mice contributes to altered visual function

Animals

All animal experiments were conducted in accordance with procedures approved in the Directive 86/609/EEC of the European Union, the Ethics Committee of the Universidad Autónoma de Madrid, and the Universidad de Alcalá and in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Aralar/AGC1^+/− mice [31] were crossed to produce Aralar/AGC1^−/− and Aralar/AGC1^+/+ control littermates in Dr. Jorgina Satrústegui’s laboratory (Madrid, Spain) and bred with free access to water and a standard diet under a 12 h:12 h light-dark cycle (maximum 350 lux). Genotype was determined with PCR using genomic DNA from mice tails with the following primers: sense primer-a (mAra 3′-LTR-F3: 5′-GTT CTC TAG AAA CTG CTG AGG-3′) for mutated alleles, sense primer-b (mAra int-13F1: 5′-GAT GTG AGA ACT CAC CAG TGT-3′) for wild-type alleles, and antisense primer-c (mAra int-13B: 5′-ACC ACC ACC AGC GTG TCA GC-3′) for the mutant and wild-type alleles. PCR conditions were 35 cycles (94 ºC, 1 min; 58 ºC, 1 min; and 72 ºC, 1 min) for DNA amplication in a 25 μl mixture containing 200 ng DNA, 0.5 units Go Taq ® G2 Flexi DNA polymerase (Promega Biotech Ibérica, Alcobendas, Spain); 0.4 microMolar primers, 0.8 mM dideoxynucleotide Mix (Niborlab, Sevilla, Spain) and 2 mM MgCl₂ (Promega Biotech Ibérica, Alcobendas, Spain). Mutant (406 bp) and wild-type (271 bp) fragments were separated by electrophoresis on a 2% agarose gel [31].

Immunohistochemistry

Light-adapted animals were euthanized with cervical dislocation. The eyes were marked for orientation, fixed for 2 h in 4% paraformaldehyde (PFA), and embedded in optimum cutting temperature (Tissue-Tek®-OCT™; Sakura Tinetek Europe, Zoeterwoude, Netherlands; frozen sections) or fixed overnight in formalin and included in paraffin before slicing in the horizontal plane (nasal-temporal orientation). Paraffin-embedded sections (5 μm) were stained with hematoxylin and eosin for morphometric analysis. Briefly, retinal thickness was measured as the outer segment (OS), outer nuclear layer (ONL), inner nuclear layer (INL), and inner plexiform layer (IPL) length along the horizontal plane at approximate 450 μm intervals from the optic nerve head (ONH) toward the periphery with ImageJ. For inmunohistochemistry, retinal frozen sections (15 μm) were blocked (0.2% Triton X-100, 2% serum in PBS: 1X: 136 mM NaCl, 8 mM Na₂HPO₄, 2.68 mM KCl, 1.96 mM KH₂PO₄, pH 7.4).) for 1 h before overnight incubation with the indicated antibodies (Appendix 1). Antibody was washed, and the retinal sections were incubated with the appropriate secondary antibodies (Appendix 1) and 4',6-diamidino-2-phenylindole (DAPI) as the nuclear marker. Autofluorescence was quenched with Sudan Black treatment. Images were obtained at 40X magnification in a Leica TCS-SP5 microscope. The number of cells was estimated as nuclear counts (DAPI) in rows in the ONL and the INL and along the ganglion cell layer. The perimeter was measured along the outer plexiform layer (OPL) with ImageJ. Reagents were obtained from Sigma-Aldrich (Madrid, Spain) unless otherwise specified.

ERG recordings

Dark-adapted (>12 h) animals were anaesthetized with an intraperitoneal injection of saline solution (NaCl 0.9%), containing ketamine (70 mg/kg; Ketalar, Parke-Davis, Wellington, New Zealand) and xylazine (7 mg/kg; Rompun, Bayer, Leverkusen, Germany), and before recording, the pupils were dilated with one to two drops of 1% tropicamide (Alcon Cusí S.A., El Masnou, Barcelona, Spain). To preserve the corneal surface from desiccation, a drop of 2% methyl-cellulose was applied (Methocel, Ciba Vision, Hetlingen, Switzerland). Three recording electrodes (ground, reference, and corneal) were used (Burian-Allen, Hansen Ophthalmic Development Lab, Coralville, IA). The corneal electrode (contact lens type) was placed in the visual axis 5 mm from the cornea. In all experiments, animal handling was performed under indirect dim red light (>620 nm) followed by 5 min in complete darkness before the recordings. Mice were kept at 37 °C on a heating pad (Hot-Cold, Pelton Shepherd Industries, Stockton, CA) during the entire procedure. Full-field ERG was the technique of choice. For low-intensity (<−2 log Cds/m²) stimuli, a Ganzfeld dome, which ensures homogeneous illumination of at least 120° in the central retina, was used, whereas for higher-intensity stimuli (>−2 log Cds/m²), a single light-emitting diode was placed close to the eye. The recorded electrophysiological response was amplified and filtered (CP511 AC amplifier; Grass Instruments, Quincy, MA), and digitalized (ADInstruments Ltd, Oxfordshire, UK). The whole process was controlled with Scope version 3.8.1 software (Power Lab, ADInstruments Ltd) [41,42]. The stimulation protocols were designed according to the International Society for Clinical Electrophysiology of Vision [43]. Six types of standard ERG responses were recorded with the protocols described in Appendix 2. Dim scotopic response (DSR), rod (b-scot), mixed (a-wave and b-wave), and oscillatory potential (OP) responses were recorded sequentially under dark background conditions, and cones (b-phot) and flicker responses were recorded following 5 min light-adaptation with background white light (50 Cd/m²). To test the effect of reducing metabolic stress by illumination, animals were light-adapted for 5 min (50 Cd/m²), and then the scotopic mixed response was recorded at different times in scotopic conditions. To test the effect of aralar absence on inner retinal cells, two identical saturating light flashes were applied with inter-stimulus of 0.3, 0.5, 1, and 2 s (double flash), with the protocol described for the mixed response. In those experiments, the mixed response in the overnight dark-adapted conditions was first acquired for each animal and used for normalization.

The DSR and rod response amplitudes were measured from baseline to the peak from the scotopic recordings; the a-wave amplitude was measured from baseline to the first trough from the mixed response or to the response amplitudes at 3 and 8 msec (in Appendix 3), and the b-wave and cone response amplitudes from the first trough to the peak from the mixed (b-wave) and photopic (b-phot) responses. OP and flicker were determined as the maximum amplitude between the trough and the peak of the waves.

Statistical analysis

Data are presented as mean ± standard error of the mean (SEM) of both eyes averaged per animal (n). The significance of the differences between genotypes was determined with unpaired two-tailed Student t tests or ANOVA followed by Bonferroni’s test. A p value of less than 0.05 was considered statistically significant.

Aralar-deficient retinas have normal morphology and architecture

We first performed histological analysis of the control and aralar-deficient retinas (Figure 2, Figure 3) to evaluate the possibility of degeneration or under-development in aralar-deficient mice. Figure 2 shows the characteristic hematoxylin and eosin staining of the typical control (Figure 2A-D) and aralar-deficient retinas (Figure 2E-H) at 17 days old. The distribution and length of the layers were not altered by the absence of aralar (compare the images in the top and bottom rows in Figure 2 and in the corresponding histograms, Figure 2I-L).

The photoreceptors are the retinal cells that first respond to light. Therefore, the number of photoreceptors will have an impact on retinal activity. We found no differences in the photoreceptor outer segment length (Figure 2I) or in the photoreceptor distribution (Figure 3A-C,G-I) or numbers (Figure 3M). Similarly, the distribution of the inner retinal cells (Figure 3D-F,J-L) was unaltered as was the number of bipolar cells (Figure 3M) and ganglion cells (datanot shown). Furthermore, the number of pyknotic nuclei (less than 3–5%) was similar in both genotypes (data not shown). Mass spectrometry analysis of metabolites showed similar levels of taurine and cGMP (Figure 3N), two metabolites present mostly in photoreceptors. Eye size estimated from the retinal perimeter along the outer plexiform layer (Figure 3O,P) also was unchanged.

Dark-adapted ERG responses are diminished in photoreceptors from aralar-deficient mice

Next, we used electroretinography to analyze retinal cell function in aralar-deficient mice and in control siblings after overnight dark-adaptation (Figure 4). ERG waveforms obtained in scotopic (dark) and photopic (with background light) conditions in response to different light intensities can be used to study the behavior of the different retinal cells. Therefore, we recorded the electrical response to light after overnight dark-adaptation to increasing light flashes first in scotopic conditions and then in photopic conditions. Figure 4A shows the responses to dim (DSR), low (b-scot), and high-intensity (mixed) light flashes acquired in darkness. The responses mainly correspond to the activity of ganglion and amacrine cells, rod bipolar cells, and the whole retina [40,42]. The mixed response can be separated in the a-wave (initial downward response) that corresponds to the activity of the rods and cones followed by the upward response of the inner retina cells (b-wave). The b-wave can be filtered to obtain the OP that reflects the activity of ganglion cells and amacrine cells [40,42]. Aralar-deficient retinas show light-evoked ERG activity, but the ERG wave amplitude is reduced by half (non-significant decrease by 42±10.23% in DSR; and significant decreases by 57±6.5%, b-scot; 42±5.43%, a-wave; 45±5.35%, b-wave; 44±7.17%, OP) at the different light intensities (Figure 4B) when compared to that of the control mice. The amplitude of the a-wave of the aralar-deficient retinas was decreased at the trough (Figure 4C) but also at 3 and 8 msec after the end of the flash pulse (Appendix 3), when there was still no bipolar cell activation, excluding an undue contribution of the altered b-wave initiation. The implicit times (time to reach half-maximal response of the b-wave) and the initial photoreceptor response (the a-wave initial slope) also were altered (Table 1). Furthermore, the flash intensity to elicit a 10% maximal response was similar in the aralar-deficient mice when compared with the control mice (Table 1, Figure 4B). We also calculated the b- to a-wave ratios and the DSR to b-wave ratios and found no differences between genotypes (Table 1), which points to photoreceptors as the primary affected cells.

Next, we tested the response to light in photopic conditions, when the rod response is saturated, to isolate the cone response. We found a decrease in the response amplitudes in the aralar-deficient retinas for single (b-phot, 42±4.2%) and high-frequency (flicker, 26.4±7.16%, statistically insignificant) stimulation (Figure 4A,C). In summary, our results show that aralar-deficient mice, although not blind [44], have a significant visual deficiency that appears to originate in photoreceptors.

Light-adaptation counteracts the impairment of responses in aralar-deficient retinas

Based on the critical role played by aralar in glucose oxidative metabolism [25,27], we hypothesized that the diminished photoreceptor responses in aralar-deficient retinas reflect the impairment of the energy production needed to maintain the dark current and glutamate vesicle recycling in darkness. Because light reduces the ATP demand of retinas [5], we reasoned that illumination would decrease metabolic stress and improve performance during subsequent ERG recordings in scotopic conditions.

To test this, we dark-adapted mice overnight and then illuminated them for 5 min (background white light, 50 Cd/m²). This exposure to illumination decreased the ERG amplitude (Figure 5A), due to rhodopsin bleaching and desensitization. The ERG response then recovers over the next hour (Figure 5C,D). In the aralar-deficient retinas (Figure 5B), there is almost no effect on amplitude just after the 5 min illumination (Figure 5C,D), and the following recovery is more efficient than in the control retinas (compare % recovery at each time point between genotypes in Figure 5E,F). This is consistent with an improved energetic state of the aralar-deficient retinas caused by the decreased metabolic workload during the illumination period.

Paired-flash test in aralar-deficient retinas unveils a reduction of bipolar cell response

Aralar is also expressed by neurons of the inner retina that may be similarly affected by a lack of aralar when facing an increase in energy demand. Furthermore, the synaptic transmission at the ribbon synapse in aralar-deficient retinas may be affected by the reported reduction in glutamine levels reflecting the impaired de novo synthesis [33] that adds to the diminished protection of glutamate from oxidation in the absence of aralar [32,35]. These metabolic alterations would affect the b-wave independently of the a-wave alteration.

To investigate this possibility, we measured the influence of aralar on the ERG inner retina component using a paired-flash stimulation protocol. This protocol has been used previously to study the recovery of a-wave or photoreceptor activity [45,46]. We extended the application of the technique to study the recovery of postsynaptic activity (b-wave). We reasoned that if glutamate recycling was impaired or the repolarization of bipolar cells delayed, the b-wave measured in a second flash should show a further decrease in the aralar-deficient mice readily observable as a decrease in the b- to a-wave ratios.

In the control retinas (Figure 6A), the first flash abolishes the photoreceptor response (a-wave) for the second flash at the shortest interval times, but the second flash gradually recovers to 70% of the initial value at the longer interval time employed (Figure 6C,D). The decrease in the photoreceptor response is mirrored by a decrease to about 40% of the inner retina response (b-wave), which recovers to 90% of the initial value (Figure 6E,F). In the aralar-deficient retinas (Figure 6B), the a-wave, although reduced in amplitude, recovers with a similar time course as the control retinas (Figure 6C,D). This is consistent with the reported normal levels of phototransduction proteins in aralar-deficient mice [35]. However, the b-wave decrease is bigger than in the controls at shorter time intervals (25% and 26% in the aralar-deficient retinas versus 35% and 38% in the control retinas at 0.3 and 0.5 s, respectively, Figure 6F). To account for the contribution of the first flash recovery to the second one, especially at the shortest inter-stimulus time, we subtracted the single flash waveform from that of the double flash (Appendix 4) and found a similar reduction in the b-wave amplitude in aralar-deficient retinas. We also found a slight increase in the a-wave amplitude at short times (0.3–1 s, Figure 6C) in the aralar-deficient retinas that is likely due to the delayed b-mixed wave initiation and would not explain the strikingly different b- to a-wave ratio behavior between genotypes (Figure 6G).

Aralar-deficient retinas are morphologically normal but have reduced dark-adapted ERG response

Photoreceptor degeneration is a leading cause of blindness [47]. The decrease in ERG amplitude we observed also occurs in retinal degeneration models where the retina cell numbers are decreased [41,42,48,49]. However, this is not always the case as in other animal models such as monocarboxylate transporter 3 knockout [50], carbonic anhydrase knockout [51] and in hypoglycemia [20] or in early stages of diabetes [52], a decrease in ERG amplitudes was observed before cell loss.

The results shown in Figure 2 and Figure 3 are consistent with previous findings of similar levels of rhodopsin, recoverin, and COX1, major markers of photoreceptor development, in aralar-deficient retinas [35]. The results are also in accordance with the absence of neurodegeneration in the central nervous system, as observed with cresyl violet staining [31] and NeuN immunolabeling [53] in the brain of aralar-deficient mice. Thus, a decrease in photoreceptor numbers as the explanation for the reduced amplitude of ERG waves is not at the basis of the reported phenotype, at least at the age under examination that is near the survival limit for these knockout animals.

Photoreceptors have high levels of aralar [34] and other MAS components [54], contain about 55–65% of retinal mitochondria [8], and are among the most metabolically demanding types of cells, energetically and from the point of view of glutamate homeostasis. Given the role of aralar in glucose oxidative metabolism [25,27], glutamine de novo synthesis [33], and glutamate homeostasis [35], photoreceptors would be the type of cell that is most likely to be affected by aralar genetic ablation.

Signal transmission in the retina is vertical from first-order neurons (photoreceptors) to second- and third-order neurons (bipolar and ganglion cells, respectively), so any defect in photoreceptor activity also will affect responses from the following layers of cells, with a certain gain relationship between two adjacent cell layers [40]. We found a decrease in all ERG wave elements (Figure 4) in the aralar-deficient retinas, but the fact that the ratio of waves corresponding to adjacent layers is maintained between genotypes (b- to a-mixed, DSR to b-mixed, Table 1) indicates that photoreceptors are the cells primarily affected by aralar deficiency.

Illumination partly rescues the effect

Aralar deficiency in retinas increases the lactate to pyruvate ratio by twofold [35]. The deficiency also diminishes incorporation of ¹³C from ¹³C-glucose into tricarboxylic acid cycle intermediates in isolated retinas [36]. These findings are consistent with an inefficient supply of pyruvate to mitochondria, as also occurs in cultured neurons [27]. Although retinas are highly glycolytic [4,6], some glucose is oxidized in mitochondria, and its metabolism increases in darkness when the energy demand is higher [4-6,8,17]. The diminished oxidative metabolism of glucose in aralar-deficient retinas may contribute to a state of diminished energy production and slower dark current and vesicle recycling. The deficiency would have its greatest impact after a long period of high metabolic demand, such as overnight dark-adaptation. The consequence would be the lower light-evoked ERG amplitude responses that we observed under those conditions (Figure 4). Illumination decreases the metabolic stress and allows the recovery of aralar-deficient retinas that return more efficiently to preillumination response values than the control counterparts (Figure 5). The light-induced decrease in the ERG response in photopic versus scotopic conditions is similar in both genotypes (reduction by 69±1.33% versus 66±2.7% of b-phot compared with b-mixed wave in control and aralar-deficient retinas, Figure 4) and argues against incomplete light adaptation as an explanation for the improved response at the shortest time point in aralar-deficient retinas. Although the involvement of other factors [55] in the “super-recovery” of aralar-deficient retinas after illumination cannot be excluded, the immediacy of the super-response is consistent with the role of the reversal of metabolic stress in the aralar-deficient retina.

In normal retinas, the recovery of the ERG response to dark-adapted levels is largely dependent on the amount of rhodopsin bleached by the adapting illumination [56,57]. Diabetic rats recover faster from light-adaptation than control animals, possibly because the diabetic animals have reduced levels of rhodopsin [58]. We do not expect this to be the case in aralar-deficient retinas, which have similar levels of rhodopsin and recoverin as the control retinas [35]. Interestingly, decreased levels of aralar [59], glutamine [60], and glutamate [61] synthesis have been reported in diabetic retinas.

Metabolic alteration of inner retina cells decreases b-wave

Another remarkable finding of our study is the substantial decrease in the second response in the paired-flash ERG analyses (Figure 6). This is consistent with a limitation in synaptic transmission that is reflected in the diminished responses of inner retinal cells (b-wave component; Figure 4). This could be due either to altered glutamate cycling, to an energy deficit of the inner retina neurons, or both (Figure 1).

Retinas have developed a system based on continuous glutamate release to the invaginated ribbon-type synapses that greatly improves the signal-to-noise ratio at the expense of the need for a high turnover of glutamate vesicles and ATP-consuming rates [38,62]. Light decreases the release rate, and thus, the glutamate concentration at the synaptic cleft reduces the saturation of the postsynaptic metabotropic receptors and depolarizing ON bipolar cells via a G-protein-mediated transduction pathway [38,63]. In the control retinas, glutamate is protected from oxidation [35] and stored back in vesicles at the photoreceptor synapse for the next release round. However, in aralar-deficient retinas, glutamate is less well protected from oxidative metabolism [35] and may be consumed oxidatively as an alternative fuel to compensate for impaired glucose utilization. Glutamine, a major glutamate precursor, also can be used as an oxidative substrate as shown by the partial recovery of the ERG amplitude in monocarboxylate transporter-inhibited retinas [22]. The impaired pathway for glutamine synthesis in these animals [33] further reduces glutamate availability. The consequence would be less glutamate to refill synaptic vesicles in photoreceptors and to fully saturate the postsynaptic receptors at the ribbon synapse. This is consistent with the reduced b-wave of the second flash in Figure 6. Pharmacological inhibition of glutamate availability with inhibitors of either glutamine synthetase or glutamate uptake [64-66] similarly affects the b-wave.

Although light hyperpolarizes photoreceptors and decreases their energetic burden [5], light has mixed effects on inner retina neurons. When the ON bipolar cells sense a decrease in glutamate, they depolarize and experience increased ATP demand to fuel the ion pumps that repolarize them (Figure 1). In cultured neurons, the aralar-MAS pathway activates mitochondrial metabolism in response to an increased workload [27]. Therefore, ON bipolar cells that are deficient in aralar may fail to increase energy production to match the increased workload caused by light. This would delay repolarization so that the ON bipolar neurons do not respond efficiently to a second flash. The reduced b- to a-wave ratio in aralar-deficient retinas (Figure 6G) is consistent with impairment of glutamate recycling and with an energy deficit within ON bipolar cells.

Conclusions

In this report, we have shown that aralar deficiency alters the response of the retina to light. The reduced light-evoked amplitude of ERGs is not associated with decreased cell numbers but can be traced back to the known effect of aralar on glucose and glutamine and glutamate metabolism. Our findings suggest that there could be a previously missed visual phenotype associated with the aralar-MAS deficiency in humans with global cerebral hypomyelination (OMIM 612949).