Creatine and phosphocreatine are required to maintain ATP needed for normal retinal function and development. The aim of the present study was to determine the distribution of the creatine transporter (CRT) to gain insight to how creatine is transported into the retina. An affinity-purified antibody raised against the CRT was applied to adult vertebrate retinas and to mouse retina during development. Confocal microscopy was used to identify the localization pattern as well as co-localization patterns with a range of retinal neurochemical markers. Strong labeling of the CRT was seen in the photoreceptor inner segments in all species studied and labeling of a variety of inner neuronal cells (amacrine, bipolar, and ganglion cells), the retinal nerve fibers and sites of creatine transport into the retina (retinal pigment epithelium, inner retinal blood vessels, and perivascular astrocytes). The CRT was not expressed in Müller cells of any of the species studied. The lack of labeling of Müller cells suggests that neurons are independent of this glial cell in accumulating creatine. During mouse retinal development, expression of the CRT progressively increased throughout the retina until approximately postnatal day 10, with a subsequent decrease. Comparison of the distribution patterns of the CRT in vascular and avascular vertebrate retinas and studies of the mouse retina during development indicate that creatine and phosphocreatine are important for ATP homeostasis.
- glutamine synthetase
the retina is a highly differentiated tissue that is able to transduce a light signal into electrical energy and encode it for subsequent transmission to other parts of the central nervous system. The retina has high and fluctuating energy demands. Light stimulation of photoreceptor cell outer segment leads to the hydrolysis of cGMP, the closing of cGMP-gated cation channels, and the electrical response (9, 24). The regeneration of cGMP from ATP and GTP during phototransduction (3, 16) has been estimated to exceed the capacity of the glycolytic pathway to produce ATP (22). Another major energy-consuming process is maintenance of the dark current by the Na+-K+-ATPase pump in the photoreceptor inner segments (4, 51). In this case, mitochondria located close to the plasma membrane of inner segments of photoreceptor cells supply ATP from oxidative metabolism for Na+-K+-ATPase activity (2, 4, 52, 53). However, generation of cGMP occurs away from sites of ATP production and the energy transfer requires creatine/phosphocreatine (Cr/PCr).
The retina contains high levels of creatine. In chicken, the total retinal creatine (Cr/PCr) concentration is ∼3 mM and may be 15 mM within photoreceptor cells (47). PCr is also relatively high in the adult mouse (17.6 μmol/g dry wt) and in the dark-adapted (25.19 μmol/g dry wt) or light-adapted (30.48 μmol/g dry wt) frog retina (25, 26). Phosphocreatine is produced by phosphorylation of creatine by creatine kinase (CK) after ATP synthesis in mitochondria. Reversal of this reaction by cytoplasmic creatine kinase regenerates ATP at sites of high energy utilization (54). The localization of creatine kinase isoforms supports the existence of a PCr circuit in the highly polar photoreceptor cells (20). Mitochondrial CK was present in the inner segments of bovine rod and cone cells, whereas the cytoplasmic brain isoform of CK was also located in the rod outer segments. This suggested that the outer segment CK and Cr/PCr play an important role in phototransduction by providing energy for the visual cycle.
Most tissues take up creatine from the blood. Molecular cloning studies identified rabbit muscle and brain cDNAs encoding a high-affinity Na+- and Cl−-dependent creatine transporter (18). The creatine transporter (CRT) exhibited significant homology to the GABA and norepinephrine transporters (17, 34) and other members of the Na+ and Cl−-dependent neurotransmitter (1, 31) or solute-carrier 6 (SLC6) family of transporters (10). The Na+, Cl−-dependent creatine transporter (SLC6A8) has been shown to be responsible for the absorption of creatine by intestinal epithelia (35) and the transport of creatine across the blood-brain barrier (29, 33). Mutations in the CRT gene (SLC6A8) result in the absence of creatine in the brain and a novel form of X-linked mental retardation, characterized by expressive speech-language delay, epilepsy, developmental delay, and autistic behavior (27, 38, 39). Symptoms from CRT deficiency cannot be corrected by creatine supplementation and persist despite the presence of creatine biosynthetic enzymes, l-arginine:glycine amidinotransferase (AGAT) and S-adenosyl-l-methionine:N-guanidinoacetate methyltransferase (GAMT) in the brain (7).
The blood-retinal barrier (BRB) shields the retina from the circulating blood. Nutrients must be obtained from the choroidal circulation in avascular retinas (e.g., rabbit), while vascularized retinas (e.g., human and rat) also receive a blood supply from vessels that enter the retina inner surface. Tight junctions are present between retinal pigment epithelial cells (outer BRB) and retinal capillary endothelial cells (inner BRB) (21). Cr/PCr are lost from tissues due to spontaneous conversion to creatinine (∼1.7%/day), which diffuses out of cells and is excreted by the kidney (60). Creatine required by the retina must be obtained by transport across the BRB. The CRT is present in both luminal and abluminal membranes of rat retinal capillary endothelial cells, where it mediates creatine transport across the inner BRB (29). However, the distribution of the CRT protein within the retina has not been studied. CRT mRNA is present in bovine and mouse retina (23, 40). The inability to detect the CRT in photoreceptor cells by in situ hybridization (23) was a surprising result, given the known importance of the Cr/PCr in these cells (20, 47). To clarify the supply of creatine to the retina, we applied an affinity-purified CRT antibody to avascular and vascular retinas and the developing mouse retina. The transporter was localized to photoreceptors, a variety of inner neuronal cells and to sites for blood-to-retina transport (blood vessels, perivascular astrocytes, and retinal pigmented epithelium), but not in Müller cells. The CRT distribution in adult mouse retina mirrors that at found at very early stages of development. These studies highlight the importance of Cr/PCr for ATP homeostasis in the retina.
MATERIALS AND METHODS
All procedures were performed in accordance with the University of Auckland Animal Ethics Committee and in adherence to the Association for Research in Vision and Ophthalmology statement for the use of animals in research. The eyes were collected immediately after animal death by other experimenters [guinea pig, rabbit, and chicken embryonic mouse eyes (46) or from the local abattoir for cow eyes]. The rat eyes were collected using procedures previously published (8, 30). We examined adult retinas from: n = 6 Sprague-Dawley rats, n = 4 Cavia porcellus guinea pigs, n = 2 New Zealand White rabbits, n = 4 Gallus gallus chickens, n = 4 cows, and n = 4 C57/BL6 mice. The analysis of CRT immunolabeling through the mouse development was conducted on mice derived from a DBA line or C57BL/6 mice at all ages (total of 54 retinas were examined). No differences in CRT labeling were encountered among these mice lines. The posterior eyecups were immersed in 4% (wt/vol) paraformaldehyde/0.01% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, from 30 min to 1 h. After fixation, the eyecups were washed in 0.1 M phosphate buffer and cryoprotected in graded sucrose solutions (10%, 20%, and 30% wt/vol). For all species, central, mid-, and far-peripheral retinas were examined, and with the exception of the chicken retina, the labeling pattern was identical at these different retinal eccentricities.
Antibody against CRT.
A rabbit polyclonal antibody was raised against a purified COOH-terminal fragment of the bovine CRT protein, expressed as a fusion protein with glutathione-S-transferase (GST) (13). Antibodies specific for the COOH-terminal 21 amino acids were affinity purified on a column containing immobilized fusion protein (GST-CC2) and stored in phosphate buffer at −20°C. Antibody specificity for the retina was assessed by preincubating 0.45 μg of the CRT antibody with 25 μg of GST-CC2 fusion protein at 4°C for 2 h. The preincubated antibody was diluted to 1:100 and applied to 12-μm-thick retinal sections of different species and age groups.
Analysis of the CRT in membranes from retina and choriocapillaris by Western blotting.
Retinas and the choriocapillaris (RPE and choroid) were quickly removed from bovine eyes and stored in phosphate-buffered saline during transport back to the laboratory. The tissues were homogenized separately in 5 volumes of ice-cold homogenization buffer [10 mM HEPES-NaOH, pH 7.4, and 0.25 M sucrose containing a cocktail of protease inhibitors (Roche Complete Mini)] with a 15-s burst on a Polytron homogenizer at speed setting 5. The homogenate was centrifuged at 1,000 g for 10 min at 4°C (model SS-34 rotor, Sorvall) to remove cell debris and nuclei. The supernatant was centrifuged at 10,000 g for 10 min and further centrifuged at 100,000 g (Beckman 70Ti rotor) for 1 h at 4°C to produce a crude membrane fraction. The supernatant was decanted, and the crude membranes were resuspended in homogenization buffer and aliquots were stored at −70°C. The protein content was determined using a commercial assay kit (Bio-Rad DC Protein Assay). Samples of retinal and choriocapillaris crude membranes were run on 10% SDS polyacrylamide gels and subjected to Western blot analysis as described previously (13). For comparison, membranes from a clonal human embryonic kidney (HEK)-293 cell line, stably expressing high levels of the CRT [∼3% of the membrane protein (50)] were run as a positive control. The CRT was detected by incubating the membrane with the affinity-purified, COOH-terminally directed antibody (0.25 μg/ml for 2 h), followed by incubation with a 1:3,000 dilution of goat anti-rabbit, horseradish peroxidase-labeled antibody (Bio-Rad) and chemiluminescent reagent (Amersham Pharmacia Biotech, ECL Plus). Control experiments were carried out by preabsorbing the CRT antibody with the GST-CC2 fusion protein. The CRT antibody (2.5 μg) was incubated with the GST-CC2 protein (50 μg) in 1 ml of 20 mM Tris·HCl-0.137 M NaCl, pH 7.6, containing 0.05% Tween (TBST) and 1% nonfat milk powder for 1 h at 4°C. The antibody was diluted 10 times with TBST/1% nonfat milk powder before incubation with the membrane blots.
Immunocytochemical labeling was performed by indirect immunofluorescence against the CRT antibody. The antibodies were diluted in 3% normal goat serum, 1% bovine serum albumin, and 0.5% Triton X-100. A mixture of the CRT antibody and a range of cell markers were applied overnight to the retina. The labeling was detected using secondary antibodies conjugated to Alexa Fluor 488 (green fluorescence) or Alexa Fluor 594 (red fluorescence). A selection of antibodies were employed in the identification of retinal cell types (19); amacrine and ganglion cells were identified with calretinin (BD Transduction, 1:1,000) and parvalbumin (Sigma, 1:1,000). Calbindin (Sigma, 1:1,000) was used as a marker of horizontal cells and amacrine cells, whereas brain nitric oxide synthase (1:1,000, Sigma) and glutamic acid decarboxylase (1:500; GAD65; BD Pharmingen) were used as markers of amacrine cells. Markers for bipolar cells were anti-Go protein, α-subunit (1:150; Chemicon), and α-isoform of PKC (PKC clone MC5, Sigma, 1:100). Müller cells were identified by using glutamine synthetase (BD Transduction Labs, 1:3,000); astrocytes were identified by glial fibrillary acidic protein (GFAP, Chemicon, 1:1,000); recoverin (Chemicon, 1:1,000)-detected photoreceptor cells, and by some bipolar cells. For double immunocytochemistry, recoverin was conjugated to Alexa fluor dye A594 using a protocol adapted from the Zenon rabbit IgG labeling kit (Molecular Probes). All fluorescent specimens were viewed using confocal microscopy. Images correspond to a maximum intensity projection of a 5-μm-thick stack of images unless otherwise specified. The images were optimized by adjusting the brightness and contrast using Adobe PhotoShop software (Mountain View, CA).
Specificity of CRT antibodies by Western blotting of membranes from bovine retina and immunostaining of rat retina.
To identify the CRT, membranes from bovine retina were analyzed by Western blotting (Fig. 1). The affinity-purified CRT antibodies were initially characterized against membranes from a HEK-293 cell line stably expressing high levels of the CRT (14, 50). The major transporter component was a broad ∼70-kDa band. Smaller (∼55-kDa) and larger (∼130-kDa) species, presumed to correspond to nonglycosylated and aggregated CRT (55), respectively, were also detected. Staining of all three components was effectively reduced by preincubation of the antibody with the GST-CC2 fusion protein. An ∼65-kDa band was seen in membranes from bovine retina. Detection of this band was blocked by preadsorption with the GST-CC2 fusion protein, confirming the specificity of the immunostaining. Similar results were also obtained for membranes prepared from choriocapillaris (data not shown). It was necessary to load large amounts of protein onto the gels and use sensitive chemiluminescence detection methods to detect the CRT in retina membranes. This resulted in other components being detected on the Western blots. However, preadsorption of the antibody with the CRT fusion protein did not reduce immunostaining of these components, indicating these components were unrelated to the CRT. Immunostaining of sections of rat retina with the CRT antibodies are shown in (Fig. 1, B–E). The strong staining of the rat retina (Fig. 1B) was effectively blocked by preadsorption of the antibody with the CRT fusion protein (Fig. 1D).
Immunolocalization of the CRT in vascular and avascular retinas.
The identification of cell types expressing the CRT was determined by immunofluorescence microscopy. Analysis of sections from retinas also enabled comparison of the CRT distribution between mammalian and chick retinas and vascular (cow and rat) and avascular retinas (rabbit and guinea pig). In mammalian retinas, strong labeling was found in the photoreceptor inner segment and the outer nuclear layer, with less immunoreactivity associated with amacrine, bipolar, and ganglion cells (Fig. 2). The CRT was present in the outer and inner retina of cow and rat retinas, both of which have a dual blood supply (Figs. 2, A and C). A similar distribution was evident in the avascular retina of the rabbit and guinea pig (Figs. 2, B and D). Labeling of the CRT in the photoreceptor outer segments was weak in cow and rat retinas compared with the strong labeling, particularly in the rat photoreceptor inner segments. Significant labeling of photoreceptor outer segments was seen only in the avascular guinea pig and rabbit retinas (Figs. 2, B and D). The nerve fiber layer was labeled in all species with examples shown in the rabbit (Fig. 2B), rat (Fig. 2C), and guinea pig (Fig. 2D). To determine whether astrocytes of the nerve fiber layer expressed CRT, we labeled the retina with GFAP. In a vertical section of the rat retina, it is difficult to show the symmetrical stellate form of GFAP-positive astrocytes encountered outside the optic nerve area (41, 57). However, GFAP labeling of the short radiating processes of astrocytes (Fig. 2H) clearly exclude expression of CRT in these cells (Fig. 2I). On the other hand, CRT expression on the blood vessels (Fig. 2F) running in the nerve fiber layer is colocalized with both processes and cell body of perivascular astrocytes (Fig. 2G). Immunolabeling for the CRT was also evident in the cells of the cow inner blood vessels (Fig. 2A). In all retinas examined, the CRT was expressed in cells within the sclera, extraocular muscles, choroid, and the retinal pigmented epithelium. Figure 2E shows examples of labeling of these structures from the rat eye.
The avian retina displayed a pattern of labeling similar to that seen in the mammalian retina (Fig. 3). The CRT labeling of the chicken retina was particularly interesting because of the reduced signal in the photoreceptor inner segments with a concurrent increase in labeling within the outer nuclear layer close to the highly vascularized pecten (Fig. 3A). Away from the pecten, photoreceptor inner segments labeled intensely (Fig. 3, B and D) with the retinal pigment epithelium (Fig. 3D) and vasculature (Fig. 3, C and E) also labeling for the CRT. Within the inner retina, CRT immunoreactivity was confined to subpopulations of amacrine cells (Fig. 3, A and B) and multiple strata within the inner plexiform layer, of which the most immunoreactive stratum colocalized with processes from CRT immunoreactive ganglion cells that projected toward the inner plexiform layer (data not shown). Figure 3C shows that CRT labeling in the pecten is localized in all cells and particularly in the endothelial cells of the blood vessels (Fig. 3E). Despite the differences in the labeling of the inner segment/outer nuclear layer in the region adjacent to the pecten, labeling of the inner retina appeared to be uniform at different eccentricities.
Localization of CRT through development of mouse retina.
We investigated whether the pattern of localization of the CRT transporter would be associated with anatomical and functional changes in the developing mouse retina. The CRT was expressed as early as embryonic day 15.5, an age when the retinal layers are not yet distinguished (Fig. 4). Labeling was restricted to those cells that have migrated to the vitreal margin of the developing retina. At birth and soon after birth [postnatal day (P)4; see Fig. 5A], the strongest labeling was evident in the outer and inner retina, with some amacrine cells and ganglion cells strongly immunoreactive. No labeling was observed in the middle of the neuroblastic layer. No differences in the pattern of labeling for the CRT were evident at different eccentricities. At P6, CRT labeling was uniform along the inner plexiform layer and somata adjacent to this synaptic layer were also labeled. At this age, the nerve fiber layer was immunoreactive for the CRT (Fig. 5B). By P8, labeling for the CRT was localized to cells in the outer and inner parts of the inner nuclear layer, probably differentiated bipolar and amacrine cells. The outer nuclear layer displayed intense labeling. The diverse soma size and location of the CRT immunoreactive cells in the ganglion cell layer indicated that ganglion cells and amacrine cells expressed the transporter (Fig. 5C). In addition, three distinct immunoreactive strata were evident in the inner plexiform layer. By P10, CRT labeling on the outer nuclear layer was intense, with labeling of the inner segments (just distal to the asterisk in Fig. 5D) also evident. Most cells in the inner nuclear layer and ganglion cell layer were labeled, as was the inner plexiform layer. The nerve fiber layer was intensely immunoreactive (Fig. 5D). By P14, CRT immunoreactivity was clearly localized to terminals of a sparse population of photoreceptors likely to be cones and to all photoreceptor inner segments (Fig. 5E). Labeling of the inner retina was reduced, as was the number of labeled amacrine and bipolar cells. From P17 to adult, CRT localization appeared in a few amacrine cells, bipolar cells, and ganglion cells, with the strongest immunoreactivity on the photoreceptor inner segments, outer nuclear layer, and inner plexiform layer (Fig. 5, F–H). CRT labeling was restricted to a few of the amacrine and bipolar cells, yet many cells within the ganglion cell layer were labeled. Immunoreactivity for the CRT in the adult mouse retina (Fig. 5H), was similar to the adult rat retina (Fig. 2C), with the outer nuclear layer showing reduced labeling compared with ages P8–P17.
Cellular identity of CRT-positive cells.
We used a range of cellular markers to identify the neurochemical signature of cells that were immunoreactive for the CRT (Fig. 6). Figure 6A shows several rabbit cone photoreceptors labeled for the CRT showing their long inner segment, soma, and axon. Rod photoreceptor cells are also labeled but show less immunoreactivity. Labeling was evident in cells in the amacrine and ganglion cell layer, with intense labeling of the nerve fiber layer; the CRT-immunonegative processes of Müller cells traversing the inner plexiform layer are identified with double arrowheads (Figs. 6, A and B). The absence of CRT on Müller cells is evident in the combined CRT and glutamine synthetase image (Fig. 6C). Glutamine synthetase labeling was observed around amacrine cells (thin arrow) or in processes that run parallel to the CRT labeling in the outer nuclear layer.
Similar to the rabbit retina, a few focal points of CRT and glutamine synthetase colocalization were observed in the mouse retina (Fig. 6, P–V), but predominantly they appeared to be individual labeled areas in the outer nuclear layer, where processes are running in close proximity (Fig. 6, D–F). Serial optical reconstruction of a photoreceptor cell in the outer nuclear layer showed clearly that glutamine synthetase labeling of the membrane of Müller cells (red) did not colocalize with CRT labeling, which was associated with the photoreceptor. An oblique section through the outer nuclear layer labeled with recoverin, and CRT showed that the CRT signal appeared at sites of soma-soma contact (Fig. 6G). Colocalization of CRT with neuronal markers indicated that the immunolabeling of the CRT in the inner retina was not specific for a population of cells, because it was colocalized with a variety of cell markers (19). Parvalbumin was expressed by amacrine and ganglion cells of the rabbit retina (Fig. 6H), although not all the parvalbumin-positive cells were immunoreactive for the CRT. Similarly to the rabbit retina, not all of the calbindin (Figs. 6, I and J) or calretinin (Figs. 6, K and L) amacrine cells expressed the transporter in the mouse retina. Calbindin was employed as a marker for horizontal cells of the mouse retina; however, no colocalization with CRT was observed. The cellular markers also labeled the stratification of amacrine cell dendrites and axons in the inner plexiform layer, where colocalization of the amacrine cell markers and CRT was observed. The colocalization in the inner plexiform layer was also observed for the GAD65 marker of GABAergic amacrine cells (Fig. 6M). In all the mammalian retinas, several types of amacrine cells and ganglion cells were labeled with CRT. Horizontal cells and rod bipolar cells were also CRT positive. In the chicken, on the other hand, CRT labeling of the inner retina identified mostly amacrine cells.
Throughout mouse development, the pattern of Müller cell labeling closely matched the pattern of CRT labeling in the outer nuclear layer, especially at P8, when increased labeling of outer and inner retina was observed (Fig. 6, P–R). However, high-resolution images did not show colocalization of the glutamine synthetase marker with CRT at P8 (Fig. 6O) or at P17 (Fig. 6, T–V). At P8, recoverin labeled the somata of developing photoreceptor cells (Fig. 6N), and CRT was delimiting the recoverin-labeled cells. At P17, double labeling of the outer nuclear layer with recoverin and CRT showed colocalization of both signals in the photoreceptor plasma membrane of the somata and at the inner segments (Fig. 6S).
We have studied the expression of the CRT within the retina with an affinity-purified antibody and confocal immunofluorescence microscopy. Strong CRT immunolabeling was observed for photoreceptors and several other types of neurons in the vertebrate retina, but strikingly, no staining was seen in glial Müller cells. We have noted significant differences in CRT expression in vascular and avascular vertebrate retinas. Studies of the mouse retina during development indicated that Cr and PCr are important for ATP homeostasis.
The identification of the CRT in tissues has been problematic because of the specificity of available antibodies (44). The affinity-purified antibodies used in the present study are directed toward a unique COOH-terminal epitope and have been shown to recognize both transiently expressed (13) and purified recombinant CRT (50). Differences between the 65-kDa CRT detected in retina and choriocapillaris membranes and the 85- and 71-kDa forms seen in rat brain and retina (29, 33) may reflect tissue-specific glycosylation. We have estimated the CRT to represent ∼3% of HEK-293-CRT cell membrane protein (50). Because ∼100 times more retina/choriocapillaris than HEK-293-CRT membrane protein was needed to detect the CRT on Western blots (Fig. 1), we estimate the CRT to represent <0.03% of retina membrane protein. A similar low abundance in membranes from other tissues may account for difficulties in the detection of the CRT (44).
The CRT was expressed in key locations (retinal pigmented epithelium, inner retinal blood vessels, and perivascular astrocytes) for the transport of creatine from the circulation into the retina. In the chicken, strong CRT labeling was found in the pecten, which contains both endothelial cells lining blood vessels and pigmented cells (Fig. 3C). Localization of the CRT to astrocytes of the inner retinal blood vessels of mammalian retina (Fig. 2, A and E) agrees well with the recent study (29) showing CRT expression in rat retinal capillary endothelial cells and a critical role for the CRT in the uptake of creatine across the inner BRB. In addition, we have shown that only perivascular astroglia express the transporter, indicating a specialized function for the different astrocytes within the nerve fiber layer. Perivascular astroglia are intricately related to the development of inner retinal vessels (57) and have recently been linked to regulation of communication within the nerve fiber layer (48). Our identification of the CRT in the retinal pigmented epithelium indicates that creatine can also be obtained from the choroidal circulation.
Localization of the CRT in photoreceptors links predominant creatine uptake to sites of high-energy phosphate production and is consistent with the Cr/PCr shuttle being used to transfer high-energy phosphates from inner to outer segments of photoreceptors. Creatine kinase activity is found in the inner segment of bovine, mouse, and chicken cone and rod photoreceptors, in close association with the mitochondria-rich portion of the inner segment, and in the membranous disk and plasma membrane of the outer segment of the bovine photoreceptors (20, 43, 47, 49). Stable levels of ATP are encountered in the photoreceptor outer segments in the frog, salamander, and mouse even after stimulation of the phototransduction pathway (12, 26, 28, 36).
CRT labeling was not restricted to any particular type of cell. Most types of neurons in the vertebrate retina contained the CRT, suggesting that most cells in the inner retina can take up creatine. The Müller glia cells play an important role in the delivery of nutrients to neurons and supply lactate to photoreceptors (37). Without transport by Müller cells, creatine would have to enter the inner retina through the CRT in astrocytes. The absence of the CRT in Müller cells also suggests that neurons may be more metabolically independent of glia than has been suggested previously and that Müller cells do not depend on Cr/PCr to maintain ATP levels. It remains possible that Müller cells or other cell types not containing the CRT may be able to synthesize creatine. To our knowledge, there have been no studies identifying the creatine biosynthesizing enzymes, AGAT and GAMT, in the retina. In brain, however, GAMT has been shown to be highly expressed in oligodendrocytes and olfactory ensheathing glia, moderately in astrocytes, and at very low levels in most neurons, leading to the suggestion of a novel neuron-glial relationship for brain energy homeostasis (45). Additional studies are required to investigate whether Müller cells can synthesize creatine and release it by a mechanism not involving the CRT.
The metabolic requirement of the mouse retina progressively increases with development, as measured by the increasing levels of oxygen uptake, CO2 production, and lactic acid formation (25, 32). However, overall PCr and ATP levels in the developing mouse retina decrease with age (25). It appears that the CRT expression has a similar tendency: PCr and CRT expression show a small peak at P10 before decreasing gradually with age (25). At P10, rods are still differentiating in the outer part of the outer nuclear layer (56), suggesting increased retinal metabolism before the onset of mature photoreceptor metabolism (26). Similarly to the developmentally related levels of creatine kinase mRNA in the chicken retina (42, 49), the changes in CRT immunolabeling in the mouse retina suggest that the PCr-Cr cycle is in place before photoreceptor function is established. Thus early immunolabeling for the CRT is consistent with a dependence on PCr not only for photoreceptor metabolic demand but also for the early development of the outer nuclear layer. Though CRT immunolabeling is observed as early at embryonic day 15.5, an age at which the photoreceptor development has not started and retinal energy requirements are not yet related to retinal function, labeling is confined to the developing ganglion cells (5, 56). In particular, increase in the immunolabeling of the transporter in the inner retina seems to be associated with the expansion of the inner blood vessels at approximately P4 through P7 and particularly associated with the pattern of selective regression of capillaries in the second and third weeks of development (6, 11, 15). Indeed, we observed that in the adult retina, CRT-positive bipolar cells were always in close proximity to capillaries, indicating that factors present in the blood, even creatine itself, may be regulating CRT expression. However, the stratified staining of the inner plexiform layer in all vertebrate retinas, including the avascular chicken retina, suggests that creatine uptake is not restricted to those cells in contact with or in close proximity to the blood vessels.
In summary, we have identified the creatine uptake sites involving the transfer of creatine from the blood into the retina and also to neural cells within the retina. The failure of creatine supplementation to reduce the progression of eye symptoms in patients with gyrate atrophy of the choroid and retina may be due to progressive damage from the inability to maintain ATP levels from early stages of development. It is also known that oral administration of creatine brings only small increases to creatine levels in the human brain (55), so further studies of factors that regulate creatine uptake may be warranted.
This work was supported by a professorship (to M. Kalloniatis) funded by the Robert G. Leitl estate and also supported in part by grants from the Health Research Council of New Zealand 02/226 (to M. Kalloniatis) and the Auckland Medical Research Foundation (to D. L. Christie).
We thank Joanna Dodd for purifying the CRT antibody used in these studies.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society