An AO multiphoton microscope has been used to visualize and characterize the structure of non-stained chicken retinas. The instrument provides TPEF images corresponding to different retinal layers, from the retinal nerve fiber area to the photoreceptor mosaic. Moreover the quality of the images together with the intrinsic optical sectioning capabilities allows the 3-dimensional reconstruction of the retinal structures. These volume renderings might help to better visualize the areas of interest and to provide additional information on the spatial distribution of retinal cells.
The retina is a light sensitive and transparent tissue. Before striking the photoreceptors, the light passes through all neural layers. In standard experiments using bright-field microscopy for the analysis of retinal structures, different staining procedures and well-defined markers are used to isolate the signal from the different cells. These markers have particular responses to the different parts of the (broad-band) spectrum of the illumination source, what is essential for the visualization and identification of the features. The different retinal layers cannot effectively be imaged without these markers, as clearly shown in . In this sense multiphoton microscopy presents noticeable advantages. A unique wavelength is used to image the entire retina structure. Moreover, it is worth to take into account that the measured signal comes from the sample’s autofluorescence (endogenous fluorescence) arising from specific molecules. This allows visualizing the different retinal cells without the need of histological markers.
To test if retinal cells provided SHG signal, the TPEF filter used in this work was changed by a narrow-band filter (380 ± 10 nm). Under this experimental condition no signal was detected, what confirms that autofluorescence is the only nonlinear signal provided by the retinal structures.
To our knowledge, this is the first time that TPEF images from the chicken retina are reported. Almost every retinal layer provides TPEF signal and although the different structures are lucidly distinguished, the photoreceptor oil droplets provided the strongest signal. Starting from the inner structure, the nerve fiber layer is observed. Although the origin of the signal from these fibers is not completely understood, they are also visible in ex-vivo human retinas [20
]. Underlying this layer, the GCs were also detectable under nonlinear microscopy imaging. These cells are hardly visible with conventional microscopy unless staining substances are used. The autofluorescence signal from the GCs layer originates from the cytoplasm of the cells, while the dark part corresponds to the cell nucleus, which exhibits no signal (see for instance ). Two main sources are responsible for the endogenous fluorescence of the GCs: the mytochondrial oxidized flavin proteins, such as the yellow emitting flavin adenine dinucleotide (FAD) [32
] and the mitochondrial reduced pyridine nucleotides NAD(P)H [33
The inner plexiform layer provides a weak TPEF signal (). For the inner nuclear layer the autofluorescence is located within the cytoplasm of amacrine and horizontal cells (see and ). At the outer plexiform layer, mainly composed of synapses, the nonlinear signal is weakly observed (). The cytoplasm of the cells within the outer nuclear layer was also easily observed ( and ). The photoreceptor mosaic was the area showing the strongest TPEF signal. The photoreceptors’ inner segment (closer to the nuclear layer) contains organelles and the cells’ nuclei. The nonlinear signal is due to mitochondrial NAD(H)P and FAD [35
]. The outer segment (closer to the choroids) contains light-absorbing materials and the signal derives from all-trans
-retinol, produced during the visual cycle.
Studies exploring the depth-resolved retinal structure using multiphoton microscopy are scarce in the literature. Some authors have shown TPEF images of the retinal pigment epithelium photoreceptor mosaic in humans [36
] and mice [35
]. TPEF images of GCs in porcine eyes [39
] have also been displayed. Recently, we have studied the TPEF sources across the human retina, providing images of individual cells in a reliable and efficient manner [20
]. We show the different layers of the chicken retina with enhanced contrast, imaged through backscattered TPEF microscopy without noticeable photodamage. An integrated morphological study can be performed as a function of both depth position and eccentricity, which elucidates the retinal structure. Due to the high resolution of the different retinal layers, it has been shown how the densities of photoreceptors and GCs change with retinal eccentricity. Changes in the density of retinal cells with eccentricity are important since this parameter limits the ocular resolving power at different retinal locations. Moreover the knowledge of these densities might help to understand and quantify the loss of cells as a consequence of retinal pathologies and the development of ocular refractive errors.
Chicken (and birds in general) have among the most complex retinas of all vertebrates [40
] and the analysis is difficult due to the extraordinary high density of small cells. Apart from rods and cones, the adult morphology of the chicken photoreceptor mosaic presents two types of cones and three types of double cones [41
]. The counting procedure here performed included all types of photoreceptors and no distinction in terms of single or double-cones was made. For the samples here studied photoreceptor densities ranged from 20000 (area centralis) to 7700 cell/mm2
(peripherally), with a mean of 13800 ± 4300 cones/mm2
. The highest density was located at the central retina and a gradual and significant decrease towards the periphery was present. This agrees with previous literature findings [31
In general, mammals also show a decrease in the density of photoreceptors and ganglion cells towards the peripheral retina [43
]. The peak density of photoreceptors was also found at the central retina for different birds, such as the duck [48
] or the crow [49
], although with densities well different: 33573 and 92109 cell/mm2
, respectively. A much lower density was found for the pigeon: 10021 cell/mm2
]. Kram and associates [42
] have recently reported that the density of cones in chickens decreases with increasing retinal eccentricity, although they did not provide data. They only provided numerical results on cone density at the mid-peripheral retina in four quadrants (17585 cones/mm2
for all types of single cones). We found an average density of 15278 ± 1444 photoreceptors/mm2
for areas between 40° and 60°, which is consistent with the previous result.
From the photoreceptor TPEF images we have computed the anatomical resolution as a function of the retinal eccentricity. We estimated the MARP to be between 6 and 7 c/deg at the central retina. This reduced to about 4 c/deg for the peripheral retina. Different authors have reported data on this parameter in birds. Values ranged between 140 c/deg for the eagle [24
], 18 c/deg in pigeon [51
], 6 c/deg for the quail [52
] and about 4 c/deg in chicken [53
]. This variability is also present when computing the behavioral spatial resolution in chicks, ranging between 1.5 c/deg reported by Over and Moore [54
] and 6.0-8.6 c/deg measured by Schmid and Wildsoet [55
], and Diedrich and Schaeffel [53
]. Our estimates are similar to Ghim and Hodos’ data [52
] and higher than those reported by Diedrich and Schaeffel [53
]. No data on MARP for different retinal eccentricities have been found in the literature to establish a comparison.
The density of GCs has been estimated in whole mounted retinas using different methods (see [56
] for details). Changes in the distribution of GCs across the different parts of the chicken retina [57
] were early reported. In particular, Ehrlich reported a reduction from 24000 (area centralis) to 4000 (edge) cells/mm2
]. Later Chen and Naito estimated a decrease from 13500 to 4300 cells/mm2
(average density: 8600 cells/mm2
]. Other studies centered on the changes in GC density in the developing chick retina [60
] or in the potential to generate new neurons [61
]. A high-density region at the central retinal area was also found for aerial birds such as pigeons and quails [62
]. Experiments in crows showed that the density of GCs diminished nearly concentrically from the central area towards the retinal periphery [49
Unlike previous studies on chicken retinas, our multiphoton imaging technique was also able to visualize the GC layer with enough contrast to compute the density of cells without the need of histochemical markers. For our specimens the density of GCs ranged between 11400 (at the central retina) to 4000 cell/mm2
(peripheral retina), with an average of 8100 cells/mm2
. The decrease with retinal eccentricity was also significant. These values agree with those estimated in [56
]. In particular the information obtained with the present technique might be used to understand changes in visual quality and ratios among different types of retinal cells and analysis of generation and loss of retinal cells during development among others. Moreover, since the loss of GCs is directly involved in glaucoma development [64
], their quantification might be useful to understand the origin and progress of this pathology.
In conclusion, it has been proven that multiphoton microscopy is a powerful tool to image the multilayered structure of retinal tissues in animal models (such as chicken presented herein) which provides complementary information. In particular, since the density of different retinal cells can be computed, changes in retinal organization can be tracked and comparisons between myopic and emmetropic eyes might be carried out. In that sense, the still-open question on whether the ocular enlargement with myopia leads to a stretching of the tissue or new cells are generated in order to maintain a constant density of cells (number per unit area), might be unveiled by using these nonlinear imaging techniques.