1.

Introduction

The retina comprises cells from the central nervous system (CNS) for the transduction of light messages into electrical signals.^{1, 2, 3} The photoreceptors (rods and cones) are the sites of light capture,⁴ while the retinal physiological repertoire is due to downstream diverse inner-retinal neurons.³

The rods, where photoreception in dim light begins,^{3, 5, 6} consist of a rod inner segment (RIS) and a rod outer segment (ROS). The RIS contains the nucleus and other organelles. The ROS is composed of a stack of sealed, flattened membrane vesicles—the disks—that contain the photopigment rhodopsin (Rh) and are surrounded by the plasma membrane. The ROS is constantly renewed;^{7, 8} new disks are formed at the cilium of the outer segment from evaginations of the plasma membrane and are displaced toward the apical tip, where old disks are shed by the rod pigmented epithelium (RPE).^{8, 9} In past years we have investigated adenosine tri-phosphate (ATP) production in the ROS.¹⁰ Recently, we reported a new protocol for the imaging of disks for immunofluorescence confocal laser scanning microscopy (CLSM),¹¹ and the first comprehensive proteomic analysis of purified rod disks, which were obtained by combining the results of 1-D or 2-D gel electrophoresis separation of disk proteins to matrix-assisted laser desorption/ionization–time-of-flight (MALDI-TOF) or liquid chromatography electrospray ionisation tandem mass spectrometry (nLC-ESI-MS/MS) mass spectrometry techniques. Proteins involved in vision as well as in aerobic metabolism were found, among which were the five complexes of oxidative phosphorylation.¹²

Several studies have addressed the immunohystochemical analysis of the retina, mostly by inclusion or on previously fixed tissue.¹³ For example, the retinal histoarchitecture was recently evaluated by light microscopy on sections of the inferior portion of the eye wall, fixed and embedded in epoxy resin.¹⁴ CLSM was also employed to analyze the structure of isolated mammalian retinas by immunocytochemical analyses,¹⁵ but this technique was applied only to fixed retinas with the exclusion of vital dyes.¹⁶ A technique for corneal confocal microscopy imaging (Retina Tomograph II Rostock Cornea Module) was developed.¹⁷ Retinal in vivo studies were conducted mainly for noninvasive diagnostic scope by scanning-laser ophthalmoscopy (SLO),^{18, 19} optical coherence tomography (OCT),^{20, 21} and optical Doppler tomography (ODT).²²

Permeant cationic lipophylic dyes have been employed for the investigation of mitochondrial membrane potential and intracellular distribution, by both regular microscopy and CLSM.²³ In fact, these probes, which are essentially carbocyanine and rhodamines, show a potential-dependent partitioning across highly polarized mitochondrial membranes. These techniques allow the imaging of mitochondria in living cells to obtain useful information on both mitochondrial morphology and cellular functionality.

In this paper we describe a new technique aimed at imaging the mitochondria of a living retina. We developed a new ex vivo technique that allows the morphologic and functional characteristics of the posterior eyecup to remain virtually unchanged. A CLSM imaging study of whole bovine retinas was conducted that employed two lipophylic fluorescent vital mitochondrial dyes, namely MitoTracker Deep Red 633 (MT Deep Red) and JC-1. Treatments were performed by incubating probes in the living eye-semicup retinas. Then the imaging was performed by mounting either the eye semicups or retinas onto a coverslip chamber. By using this procedure, we could image not only the mitochondria of the various retinal cell types, but also the ROS. Such unexpected staining was quite surprising in that it seemed specific, being inhibited by inhibitors of mitochondrial respiration or protonophores.

2.

Materials and Methods

3.

Results and Discussion

Many lipophylic fluorescent probes interact with actively respiring membranes and show changes in their fluorescent characteristics as they redistribute between compartments. We employed MT Deep Red and JC-1 to image the mitochondria of a whole retina by a new ex vivo approach. The novelty of the procedure is in the treatment of the samples, which maintains the physiological and morphological features of the eye, and in the possibility of imaging a living retina. Eye semicups from freshly detached bovine eyes, with the retinas still attached to the RPE, were filled with MR containing $1\mathrm{mM}$ glucose, $50\mu \mathrm{g}∕\mathrm{ml}$ ampicillin, and a protease inhibitor cocktail, then incubated for $10\min$ . Then probes were added to the solution and it was incubated for $15\min$ more. Finally, the retinas, which by this time had spontaneously detached from the RPE, were not fixed or washed and were mounted on the coverslips.

Figure 1 shows a CLSM color-coded 3-D projection of MT Deep Red fluorescence of a portion of a whole bovine living retina, treated as described above. MT Deep Red is a photostable permeant cationic fluorescent mitochondrial probe,²⁶ that is known to be taken up by functional mitochondria, due to their high membrane potential, without fluorescence quenching. The figure shows that, quite unexpectedly, fluorescence appears to be distributed on the ROS. The morphology of the ROS is clear enough to allow unambiguous identification of its structures.^{27, 28} Moreover, considering the relative $z$ depth of the picture as translated by the color table, the ROS structures appear fluorescent for about $30\mu \mathrm{m}$ , i.e., a half of their length, which is $60\mu \mathrm{m}$ (with $1.2\text{-}\mu \mathrm{m}$ mean diameter in the cattle^{27, 28}). Interestingly, the ROS distal tract is known to be active in phototransduction, thanks to the modified composition of disk lipids.^{29, 30}

Fig. 1

Color-coded 3-D projection of MT Deep Red fluorescence in a living bovine retina. The color table translates the colors into the relative $z$ depth of the image. The scale bar is $30\mu \mathrm{m}$ .

Figure 2 contains CLSM images showing the fluorescence of MT Deep Red in representative portions of a whole bovine living retina. In Fig. 2a, both the mitochondria belonging to other retinal cell types and some detached ROS structures are visible. The different mitochondrial fluorescence intensities may represent variable planes of focus. In fact, mitochondria undulate in and out of the plane of observation. Variations in the mitochondrial accumulation of cationic probes were reported to be potential-dependent.²³ Figure 2c shows a portion of a whole retina with fluorescent rods still attached that seem to be imaged in perspective from their apical zone, as judged by the absence of mitochondria on the plane nearer to the objective. No other neurons of the retina were visible in the many pictures acquired from the imaged whole retinas. MT Deep Red ROS fluorescence appears to be more intense in the apical tract of the ROS, which is functional in phototransduction.^{29, 30} The objects shown in Fig. 2b, are the 3-D reconstructions of a part of the fluorescent tips in Fig. 2c, as determined by an Image J software plug-in.

Fig. 2

CLSM images of MT Deep Red fluorescence in a living bovine retina. All the scale bars are $30\mu \mathrm{m}$ . (A) Mitochondria of the various cell types of the retina and some detached ROS are clearly visible. (B) 3-D reconstruction made on a portion of the image stack visualized in panel C. (C) Maximum projection $(z=84\mu \mathrm{m})$ of serial confocal sections on an undulate part of the retina. (D) Maximum projection $(z=52\mu \mathrm{m})$ of serial confocal sections of an undulate part of the retina after preincubation in the presence of rotenone and antimycin-A.

Impairment of mitochondrial function with inhibitors of electron transport in the respiratory chain is known to cause a decrease of MT Deep Red fluorescence in mitochondria. To verify the specificity of the ROS staining, in a parallel set of experiments we exposed the living retinas to both rotenone, a specific inhibitor of NADH dehydrogenase (complex I of the respiratory chain) and antimycin A, an inhibitor of complex III of the respiratory chain, before adding fluorescent probes. Inhibitors ( $10\text{-}\mu \mathrm{M}$ final concentration, each) were added to the MR in the eye semicup $5\min$ after the start of incubation. MT Deep Red was added $5\min$ later, and after 15 more min, the retinas were imaged. Figure 2d shows that, when preincubated in the presence of rotenone and antimycin-A, the MT Deep Red fluorescence intensity in the ROS was reduced by 4 to 5 fold, as evaluated by densitometric quantitation performed using the Image J 1.31 v. software. Quantification was performed to compare the mean relative optical density value calculated for five identical areas of each image after subtraction of the mean background value for each. Accordingly, a decrease in MT Deep Red fluorescence was observed in mitochondria [Fig. 2d]. MT Deep Red probes are known to passively diffuse across the plasma membrane and accumulate permanently in actively respiring mitochondria.²⁶ The fluorescence intensity associated with a membrane is likely to be a reflection of a protonic potential because, at least in mitochondira, the electron transport establishes a proton gradient across the mitochondrial inner membrane.

The unexpected staining of ROS prompted us to try another classical mitochondria-specific dye to verify whether the phenomenon was repeatable. We turned to JC-1, a membrane-potential carbocyanine-sensitive probe for imaging living cells. The advantage of JC-1 over rhodamines is that its color alters reversibly from green to red with increasing membrane potential. The so-called J-aggregates are favored at a higher membrane potential. Figure 3 shows the results of an experiment in which a whole retina was incubated with JC-1 by the same procedure as used in Figs. 1 and 2. CLSM images in Figs. 3b, 3c, 3d, 3e show that both the ROS and mitochondria of the various retinal cells appear fluorescent ( $10\min$ after addition of JC-1). Figure 3a shows a 3-D reconstruction of the image in Figs. 3b and 3c, which show the green and red fluorescence of JC-1, respectively, in the two different channels acquired sequentially. In the ROS, the dye underwent a change in fluorescence emission from green to red. In inactive mitochondria, the JC-1 monomer emits green fluorescence, while the aggregated form, which is prevalent when membrane potential (i.e., functional capacity) is high, emits a red fluorescence. J-aggregate fluorescence intensity was found to increase linearly with increasing mitochondrial membrane potential.^{31, 32} Compared to other dyes that give single-component fluorescence signals, JC-1 red fluorescence is not affected by mitochondrial size, shape, or density.³³ J-aggregate fluorescence was demonstrated to be independent of pH, within the physiological range, and not to be due to any metabolic conversion.³² In experiments with mitochondria, the red peak was found to be sensitive to protonophores,³² which cause a loss of membrane potential; thus, we utilized nigericin, a specific protonophore, to verify whether the ROS behaved the same as mitochondria. Nigericin was added directly to the imaged retina on the glass slide during the CLSM experiment (at a $4\text{-}\mu \mathrm{M}$ final concentration in a minimum-DMSO volume), then the observation of the retina was continued. Fluorescence in the sample was found to diminish, both in the ROS and in most of the mitochondria: Figs. 3d and 3e show that $10\min$ after the addition of nigericin, the red fluorescence was reduced by 2 to 3 fold, as shown in Fig. 3. On the other hand, green fluorescence persisted, since the sample was not processed further. The large aggregates present in Figs. 3a, 3b, 3c, likely due to a relatively high potential on the ROS tips, disappeared with the addition of nigericin. Since JC-1 does not exhibit any concentration-dependent quenching effects,³² a strong shift from red to green emission of JC-1 should correspond to a decrease in membrane potential.

Fig. 3

CLSM images of JC-1 fluorescence in a living bovine retina. All the scale bars are $30\mu \mathrm{m}$ . (A) 3-D reconstruction made on a portion of the image stack visualized in panel B. (B)–(C) Maximum projection $(z=54\mu \mathrm{m})$ of serial confocal sections on an undulate part of the retina. Green $(520\pm 20\mathrm{nm})$ and red $(580\pm 20\mathrm{nm})$ channels were acquired sequentially. (D)–(E) Maximum projection $(z=54\mu \mathrm{m})$ of serial confocal sections on an undulate part of the retina $10\min$ after the addition of nigericin. Green $(520\pm 20\mathrm{nm})$ and red $(580\pm 20\mathrm{nm})$ channels were acquired sequentially. (Color online only.)

By the technique reported herein, only the ROS and mitochondria appear to be stained by fluorescent mitochondrial probes in a whole living retina. The same results were obtained when the whole eye semicup was mounted onto the coverslip (data not shown). No differences were observed, such as those for the fluorescence emission intensity, when the incubation with both dyes was conducted in room light with respect to dim red light (data not shown).

4.

Conclusion

This paper demonstrated that the mitochondria-specific interaction of the cationic dyes utilized in this study depended on the presence of a high transmembrane potential, like the proton gradient $\left[\Delta \mu \left({\mathrm{H}}^{+}\right)\right]$ maintained by functional mitochondria across the inner membrane. In our experimental conditions, dissipation of membrane potential by treatment with mitochondrial poisons and uncouplers eliminated the selective association of MT Deep Red and JC-1 with both the mitochondria and the ROS (Figs. 2 and 3). Further studies may establish whether this means that the tip of the ROS possesses a transmembrane potential of the kind and order of magnitude as those of mitochondria (about $190\mathrm{mV}$ ) and an electron transport. For our CLSM measurements, the eye semicups were incubated with the probes in the dark, then retinas were withdrawn and mounted onto the glass slide in room light. In these conditions, the ROS should be in a relatively hyperpolarized state. If the high transmembrane potential to which JC-1 is sensitive resides on the ROS plasmamembrane, the whole ROS would be stained, which seems not to be the case. On the other hand, it may be supposed that the high potential resides in the disk membranes. Several previous results are consistent with the notion of ROS disks as organelles able to store and/or release protons.^{34, 35} The existence of a $\Delta \mu \left({\mathrm{H}}^{+}\right)$ across the disk membranes was previously demonstrated by light scattering experiments.^{36, 37} According to these authors, the light-scattering signal also suggested a ${\mathrm{H}}^{+}$ translocation across disks. It is tempting to assume that a $\Delta \mu \left({\mathrm{H}}^{+}\right)$ across disk membranes is responsible for the JC-1 staining (Fig. 3), because it was reverted by the protonophore nigericin.

In mitochondria, membrane potential $\left(\Delta \psi \right)$ represents a sensitive parameter of the coupling of the mitochondrial bioenergetic function. Moreover, increases in the accumulation of cationic mitochondrial probes were suggested to reflect an increase in ATP requirement in cells.³⁸ The ROS is considered to contain organelles devoid of mitochondria, the site of anaerobic metabolism, even though glycolysis, which is present in the ROS, was estimated to be insufficient to supply ATP for phototransduction.^{6, 39} The results shown in this paper suggest that the functioning of rod disks might involve the build-up of a proton potential. This would be consistent with our recent results from a proteomic analysis of purified disks, where the proteins of the four complexes of the electron transport chain were found to be expressed.¹² Moreover, those results showed that ${\mathrm{F}}_1{\mathrm{F}}_{\mathrm{o}}$ -ATP synthase is located and catalytically active in the ROS disk membranes,¹² suggesting the presence of an aerobic glucose metabolism in disk membranes that could represent a mechanism for an ATP supply in phototransduction. Further studies may clarify the physiological reason for the ROS behavior when exposed ex vivo to classical vital mitochondrial dyes.