In vivo angiography of M. unguiculatus using cSLO (, ) showed a strikingly different pattern of the retinal blood vessels than in mice (). The major blood vessels supplying the inner retina divide into several main branches shortly after entering the eye at the optic disc, and then subdivide into a network of capillaries across the retinal surface obviously avoiding a distinct region of the retina. A similar horizontal H-shape of major vessels was also found in G. perpallidus and P. campbelli. The vessels ramify and narrow into a bed of capillaries as they approach the horizontal sparse vascular band, and except for one artery and one vein only small capillaries invade the region in-between the major vessels. This horizontal sparse vascular band visible in in vivo angiography is located dorsal to the optic disc in all three rodent species (). It appears to correspond to a visual streak, very similar to a specialized retinal region like an area centralis or human fovea.
In vivo morphological analysis and retinal layering with cSLO angiography and SD-OCT imaging.
In cSLO angiography (), both dyes used for in vivo
angiography, fluorescein (FLA) and indocyanine green (ICG), are present in the retinal as well as in the choroidal circulation. Fluorescein angiography provides the most detailed images of retinal capillaries, whereas ICG images depict both retinal and choroidal structures. The degree of visibility of deeper-lying fundus structures depends on the retinal layering 
. Usually, with fluorescein angiography only the retinal vasculature is visible. However, in the area of the visual streak also choroidal vessels were clearly discernible (). This strongly suggested a different retinal layering within the visual streak region. We assessed this question by in vivo
analysis using SD-OCT imaging. In vivo
cross-sectional imaging was performed using a commercially available third generation Spectralis HRA+OCT (Heidelberg Engineering, Heidelberg, Germany), which provides high-resolution depth profiles of the retina based on reflectivity of light. This imaging method allows to reliably identify all retinal layers in vivo
with a resolution comparable to that in histological sections 
. A significant increase in retinal thickness was found in the visual streak region in comparison to surrounding parts of the retina (). Virtual cross-sections displayed identical laminar organization in the inner retina, but in all three species, SD-OCT revealed a clear increase of thickness in the outer retina, particularly in the layer corresponding to the photoreceptor outer segments 
In vitro wholemount stainings allowed visualization and assessment of the three vascular layers in the entire visual streak region and thus complemented in vivo cSLO imaging. Collagen IV staining revealed that the specialized vascular patterning was restricted to the superficial vascular plexus within the streak area (). G. perpallidus showed a reduced capillary branching in this horizontal streak area. The capillaries traversed the streak area in a vertical pattern. The superficial vessels of M. unguiculatus arborized in the horizontal streak region similarly to G. perpallidus. In both G. perpallidus and M. unguiculatus, this specialized vascular patterning was present in a horizontal band that spanned the entire retina. The retinal vasculature of P. campbelli showed only slight modifications of the superficial vessels in the streak region, which did not cross the entire retina.
Detailed analysis of the retinal blood vessel distribution in vivo using cSLO and in vitro using collagen IV staining.
The course of the capillary network is very similar in magnifications of both in vivo cSLO FLA angiography and in vitro collagen IV stained retinal wholemounts (). For further morphological analysis, semithin transverse sections were prepared from Epon-embedded eyes of all three species. Likewise, in these histological sections the streak region could be identified by a thickening of the retina (). Direct comparison of the retinal layering within the streak region () to areas outside () revealed that particularly the layers of the photoreceptor outer segments were substantially enlarged, corroborating the observations from SD-OCT imaging. Electron microscopy further confirmed the significant elongation of the photoreceptor outer segments within the streak region (). Their elongation was roughly estimated to be 1.5 fold. Interestingly, this specialized retinal region generated some difficulties for embedding and cutting. As shown in the longer outer segments had a slanted orientation in the electron micrographs.
Detailed retinal analysis in vitro.
Nissl-stained wholemounts revealed a horizontal visual streak of increased neuron density in the ganglion cell layer (GCL). Among the three species, G. perpallidus showed the lowest overall neuron density in the GCL, and a visual streak was very noticeable. Within the visual streak region, a markedly higher neuron density could be detected ( bottom). The increase in neuron density was sharper at the dorsal edge of the streak, ventrally the transition was more gradual. M. unguiculatus and P. campbelli had overall higher GCL neuron densities across the retina. M. unguiculatus also possessed a distinct GCL visual streak with a sharp dorsal edge and a less well-defined ventral edge, whereas P. campbelli only showed a very moderate density increase with shallow dorsal and ventral gradients in the streak region.
The labeling of retinal wholemounts for cone opsins revealed inter-species differences. Across the whole retina, G. perpallidus
had the lowest cone densities among the three species. Most of the cones expressed the middle-to-longwave sensitive (MWS) opsin, and only a few percent the shortwave sensitive (SWS) opsin. In large patches across the retina, SWS opsin was completely absent. All SWS opsin-expressing cones co-expressed the MWS opsin, which is a feature observed in a number of rodents 
. Cone density had a shallow peak in the streak region, with a distinct drop towards dorsal periphery (). M. unguiculatus
had overall higher cone densities. MWS and SWS opsin were expressed across the retina, with MWS cones forming the majority. However, a large proportion of the MWS cones co-expressed small levels of SWS opsin. Highest cone densities were found in ventral retina, and there were no cone specializations associated with the dorsal streak. P. campbelli
also had high cone densities with maxima in ventral retina. Peculiarly, SWS opsin expression was restricted to the ventral two thirds of the retina. At the streak, SWS opsin expression abruptly ended and dorsal of the streak only MWS cones were present (). Ventral to the streak, many cones co-expressed both opsins.
Immunofluorescent stainings of retinal sections revealed distinct differences in cone cell distribution and vessel density (). Isolectin IB4-FITC stained retinal sections of G. perpallidus revealed the sparse superficial vascular network within the streak region (). In comparison to dorsal periphery, peanut agglutinin stained retinae of P. campbelli illustrate increased cone densities within the streak region () corresponding to . More specifically, SWS opsin staining of M. unguiculatus retinae revealed enhanced SWS cone density within the streak area () compared to the dorsal periphery. Both rods and MWS cones were evenly distributed across the retina as shown by rod transducin (GNAT1) staining in M. unguiculatus (), rod opsin staining (), and MWS opsin staining in G. perpallidus (). Remarkably, all the antibodies used for immunofluorescent stainings () stained the same antigens as in mice.
Immunofluorescence labeling of retinal sections revealed distinct differences in cone cell distribution and vessel density.