GROUND SQUIRREL – A COOL MODEL FOR A BRIGHT VISION

Abstract

The great evolutionary biologist, Theodosius Dobzhansky, once said, “Nothing in biology makes sense except in the light of evolution.” Vision, no doubt, is a poster child for the work of evolution. If it has not already been said, I would humbly add that “Nothing in biology makes sense except in the context of metabolism.” Marrying these two thoughts together, when one chooses an animal model for vision research, the ground squirrel jumps out immediately for its unique cone dominant retina, which has evolved for its diurnal lifestyle, and for hibernation—an adaptation to unique metabolic challenges encountered during its winter sojourn.

1. Introduction

Ground squirrels are a group of rodents that belong to the squirrel family (Sciuridae). They are diverse both in their body size, ranging from tiny chipmunks (~100g) to giant-sized marmots (>5kg), and in their geographic habitats, ranging from the scorching Mexican desert (>40 °C, antelope ground squirrel) to the freezing Alaskan ice-land (<−40 °C, Antarctica ground squirrel). However, they all share a common name derived from their ground-dwelling rather than tree-climbing behavior. Unlike many mammals (other than primates) that are nocturnal [1], ground squirrels are diurnal day-dwellers. Primarily living in open or grassy areas, many ground squirrels developed the capability of standing upright in order to spot food or predators from the ground. They assume this vertical posture by using their hind limbs for prolonged period of time, with forepaws hanging in front of their chest or holding onto food. A forward-looking gaze may help them to stabilize such an upright posture [2]. Ground squirrels are omnivorous, they seek nuts and seeds but also hunt insets such as grasshoppers and even smaller mammals. They therefore evolved an outstanding visual system with high acuity, good color discrimination, and fast-tracking ability, characteristics that made ground squirrels popular test subjects from the early days of modern vision research [3,4]. Another unique feature of the ground squirrel is that many of them hibernate during winter to cope with the scarcity of food and cold weather. Hibernation invokes drastic metabolic changes to the ground squirrel as a whole affecting neural tissue like the retina in particular, owing to their high energy demand in normal awake conditions. This provides a unique opportunity to study the visual system and its plasticity linked with metabolism [5]. This review will encompass the history and landmark works on the ground squirrel visual system with a focus on the retina and these distinctive features.

2. Cone dominance

Many modern-day mammals are nocturnal, owing to selective pressure faced by their ancestors during the “nocturnal bottleneck” in the Mesozoic era, when they were forced to adopt a nocturnal temporal niche in order to avoid predation by large, carnivorous diurnal dinosaurs [3,1,6]. This evolutionary path had a profound impact on the visual system of mammals. Among the many adaptive changes, one of the most direct and significant is perhaps the expansion of the rod photoreceptor population to achieve higher sensitivity for better night vision. For many mammals, rods account for > 90% of their total photoreceptors. Even for primates, which are mostly diurnal, their retinas contain approximately 95% rods, except for the fovea region that is composed exclusively of cones for high acuity, color, and other daylight vision needs. Thus, ground squirrels stood out as an interesting mammalian model for cone photoreceptor-driven vision (Fig. 1). As termed by Walls [3], although the ground squirrel does not possess a specific fovea region, their retina appears to be a case of “universal macularity”. Indeed, for a long time, the ground squirrel retina had been believed to be an “all-cone” retina, until some fine electron microscopy studies in the mid-1970s identified a small population of “rod-like” photoreceptors in a handful of ground squirrel species [7–9]. The unique features of the ground squirrel rod photoreceptor and rod pathway will be discussed in further detail later. Regardless, the cone-dominated ground squirrel retina has been exploited to study different aspects of cone vision from the visual cycle to visual behavior.

Figure 1.

A ground squirrel retina section labeled with antibodies against recoverin (green), calretinin (red) and PKA (blue). Most photoreceptors are cones, asterisks mark soma of putative rods.

2.1. Cone photopigment regeneration

Ground squirrel retina has been instrumental in the research regarding the mechanism of cone photopigment regeneration. It has been long recognized that cones might have a different (or additional) mechanism for visual pigment regeneration than rods to enable faster recovery from photopigment bleaching - a process known as “dark adaptation” [10]. The rate-limiting factor for dark adaptation appears to be the availability of recycled chromophore to photoreceptors [11,12]. In line with the fact that rod pigment regeneration occurs in the retinal pigment epithelium (RPE), a majority of retinyl esters as well as the key enzyme RPE65 were found in the RPE of rod-dominant retinas [13,14]. On the contrary, for cone dominant retinas found in ground squirrels and chickens the retinyl esters were instead concentrated in the neural retina, and a different set of enzymes, presumably in Müller cells, may constitute an additional pathway for cone chromophore regeneration [15,16,reviewed in 17]. Physiological evidence came later, first from salamander retina [18], and then from transgenic mice in which cones were the only photoreceptors that could generate a light response [19]. Isolated recordings from individual cones demonstrated that the recovery of the cone response was independent of the RPE but required interaction with the retina, likely Müller cells. It would be desirable to acquire similar single cone physiological data from the ground squirrel retina in the future. Nonetheless, recent in vivo electroretinogram (ERG) recordings from ground squirrels provided interesting insight into cone pigment regeneration [20]. On one hand, an RPE65-selective inhibitor (MB-001) significantly slowed the recovery of the ERG b-wave after bleaching, indicating a critical role of RPE65 in cone pigment regeneration. On the other hand, after recovery, cone ERG responses were well maintained, confirming an RPE65-independent pathway for cone pigment regeneration. Given that the details of the cone visual cycle remain incomplete, it is agreeable that the cone dominant ground squirrel is a suitable animal model for the future biochemical and physiological studies on this topic.

2.2. Cone physiology, cone-bipolar cell synapse, and cone pathways

Visual neurophysiological studies on ground squirrel started in the 60’s. ERG, visual evoked potential (VEP), and single optic fiber recordings have been employed to probe the responses of the ground squirrel visual system to basic stimuli such as contrast, color, and directional selectivity, many of which are still being studied more than half century later [21–25]. These early works took advantage of the cone dominant feature of the ground squirrel retina to study cone vision but were not able to directly explore the physiology of cone photoreceptors due to technical limitations at the time. The development of the suction electrode technique revolutionized the research of photoreceptor physiology [26,27]. Facing the difficulty of finding cones in a sea of rods even in primate retina (apart from fovea, which has its own nuisance), Timothy Kraft [28] first turned to the cone-rich ground squirrel and characterized in great detail the features of cone photocurrents in the golden-mantled ground squirrel, some of which has been revisited in a recent comparative study of mammalian cone light responses [29]. One of the unique characteristics is the fast time to peak (20ms) of their flash response, the mechanism of which remains unclear.

In the late 90’s, taking full advantage of the cone dominance of the ground squirrel retina, Steven DeVries initiated a series of elegant electrophysiological studies on cone-bipolar synapse and cone pathways in a slice preparation, using a state of the art double-patch clamp technique. In a landmark paper [30], kainate receptors were discovered to be the ionotropic glutamate receptor expressed by OFF cone bipolar cells (Fig. 2). The desensitization of the kainate receptors prevented the cone-OFF bipolar cell synapse from saturating, thus encoding the entire dynamic range of cone input. In addition, this is also one of first studies to demonstrate that kainate receptors can mediate critical synaptic transmission in the central nervous system [31]. Subsequently, by correlating the morphology of different types of OFF cone bipolar cells with their distinct receptor types (AMPA and kainate) and kinetics, DeVries [32] proposed a scheme by which synaptic processing at the cone-OFF bipolar synapse transformed the same cone output into parallel temporal channels. This work has been further advanced by more detailed subtyping of OFF cone bipolar cells and their corresponding receptor subunit as well as auxiliary protein compositions using a combination of anatomical, molecular, and physiological approaches [33,34]. Furthermore, it has been demonstrated that such temporal processing at the cone-OFF bipolar synapse can be greatly facilitated by the unique architecture of the cone synapse. Glutamate released at the base of the synaptic ribbon in a cone terminal must diffuse through a narrow invagination jammed with seemingly chaotic postsynaptic processes, but in actuality, there is a stereotyped spatial arrangement such that different types of bipolar cell dendritic tips are exposed to different concentrations of glutamate, which works in concert with their unique glutamate receptor composition to extract distinct temporal signals from a single source of cone output [35].

Figure 2.

Dual patch clamp recording from a cone (green) and an OFF cone bipolar cell (red) in a ground squirrel retinal slice (A and B). C. Cone depolarization-triggered post synaptic excitatory current from the OFF cone bipolar cell can be blocked by CNQX (AMPA/KA receptor antagonist), but not GYKI 53655 (AMPA receptor specific antagonist). Modified from [30].

These works and others [36] ascended ground squirrels to the short list of animals of which a near complete census of retinal bipolar neurons have been mapped out. Compared to mouse and primate retina, the ground squirrel retina has, naturally, a much larger proportion of cone bipolar cells owing to their cone dominance, while sharing a similar general layout. Some unusual but shared features include specific bipolar cell type(s) that express Na⁺ channels, which breaks the general rule of bipolar cells being non-spiking neurons that signal by graded potentials [37–39]. One outstanding difference in the ground squirrel retina appears to be the existence of a type of OFF cone bipolar cell that expresses AMPA type receptors (cb2) that might be missing from mouse and primate retina [32,40,41,but see 42,43]. This particular bipolar pathway appears to be critical for transient, high frequency signaling in the ground squirrel retina, as it possesses several unconventional features. First, cb2 bipolar dendritic tips occupy a semi-invaginating position close to the glutamate release site for fast, high concentration glutamate access. Second, cb2 bipolar cells express AMPA receptors to generate a transient depolarization at light off, akin to the effect of Na⁺ channels expressed in rat CB3a [37] and primate DB3a [39], enhancing transient signaling to the downstream ganglion cells Third, the AMPA receptors on cb2 bipolar cells are prone to saturation, such that output signals from a depressed cone ribbon synapse can still be transmitted under high temporal frequency stimulation [44]. This unique cone-bipolar pathway, in combination with the fast cone response mentioned above [28], highlights a speedy visual signal transmission system that is likely suited for ground squirrel’s diurnal and anti-predatory vigilant behavior.

2.3. Cone ribbon synapse

The ability to perform patch clamp recordings simultaneously from a presynaptic cone and a connected postsynaptic bipolar cell in the ground squirrel slice preparation also provides an opportunity to probe cone ribbon synapse features such as vesicle release properties. In addition, both capacitance measurements [44] and glutamate transporter (Cl⁻) current measurements [45] can be used to monitor vesicle release from the cone synapse. More intriguingly, cone synaptic ribbons undergo drastic structural changes during hibernation [46,47,5], providing an opportunity to study the functional consequence of such ribbon structural modifications. Indeed, in line with results from the mouse rod bipolar cell ribbon synapse, data on the cone ribbon synapse from hibernating ground squirrel retinas indicated that spontaneous released vesicles from ribbon synapses may be influenced by the size of the ribbon and are likely separated from the readily releasable pool of vesicles [47]. More hibernation related vision research will be discussed in section 3.

2.4. Color vision

Naturally, because of their cone dominant feature and diurnal behavior, color vision became perhaps the first topic of vision research in the ground squirrel. Inspired by the depiction of the ground squirrel retina [3], and later by conversing with Gordon Walls, Frederick Crescitelli started working on color vision of ground squirrels in the early 60’s, when vision research tools were fairly primitive at the time [22]. As described by Crescitelli and Pollack [48], “the study of color vision in animals other than man is, at best, a troublesome and uncertain occupation”. Nonetheless, these pioneers soldiered on and produced an impressive body of work on this subject.

An early behavioral study demonstrated that under conditions of equal luminosity, antelope ground squirrels were able to discriminate 460 nm light illumination against a background of between 500 to 600 nm [49]. Although microspectrometer measurements at the time only identified one visual pigment which peaked at 523 nm, subsequent tungsten electrode recordings of optic nerve fibers in the Mexican ground squirrels clearly demonstrated color opponent responses from retinal ganglion cells [23–25]. In fact, about one quarter of the recorded optic nerve fibers were color opponent, among which green ON/blue OFF and blue ON/green OFF accounted for about half each. In addition, center-surround receptive field properties of these color opponent cells have also been described in detail [25]. Subsequently, behavioral studies showed that all five different species of ground squirrel have very similar dichromatic color vision and visual sensitivity [50], and their cone pigments peak at about 440 and 525 nm. These initial observations were not too far off from the more accurate ERG flicker photometry measurement [51], as well as single cell measurements performed a decade later [435 and 520 nm, [28]]. One interesting finding among the ERG studies of the ground squirrel color vision during the late 80’s is that the b-wave mediated by S-cone input was more than 20 ms slower than that generated by M-cones, while both cone types showed no difference in flicker frequency responses [52]. This indicates a disparity in post-receptor processing of S- vs. M-cones, for which the authors attributed to a possible difference in horizontal feedback. A recent study in the primate retina, however, did find the S-cone light response to be slower than L- and M-cones, thus providing an alternative explanation [53]. Nonetheless, the S-cone pathway seems to be more sluggish for additional reasons that will be discussed below.

Despite the early physiological demonstrations of color responses in the ground squirrel retina, the anatomical description of cone topography was not available until the late 90’s when rod and cone distributions in the California ground squirrel were reported [54]. Each cone in the ground squirrel retina expresses either S- or M-opsins. S-cones account for roughly 7% of the cone population, semi-regularly distributed in a forest of M-cones. This confirmed the behavioral and physiological studies that the ground squirrel, like most of mammals, are dichromats.

Because of their cone-dominance, neighboring cone pedicles are in close contact with each other, making it easier to form gap junctions [8], similar to what had been observed in primate fovea cones [55]. The first direct electrophysiological evidence of mammalian cone gap junctions was acquired by dual patch clamp recording of neighboring cones in the ground squirrel retina [56]. Based on the conductance measured between cones, the neural blur caused by electrical coupling was calculated to be smaller than the optical blur of the eye. Therefore, without much degradation to spatial resolution, gap junction between cones can in fact enhance the signal to noise ratio by over 70% through the effect of averaging of uncorrelated noise. However, electrical coupling between cones of different spectral types could indeed compromise color processing, especially for S-cones, that due to their scarcity are always surrounded by an array of M-cones. Interestingly, it was discovered in the ground squirrel retina that S-cones were in fact isolated from the network of M-cones, thus preserving their spectral tuning by not forming gap junctions with neighboring M-cones [57], similar to S-cone isolation discovered in the primate [58].

The ability to identify S-cones (< 10% of cones) in live ground squirrel retinal slices [57,59] opened an opportunity to map the downstream connecting neurons. By systematically pair-recording S-cone and different types of bipolar cells, together with post-recording anatomical examination, bipolar cells can be classified into S-cone (only) bipolar cells, S-cone avoiding bipolar cells, and non-selective bipolar cells, outlining parallel color and luminance pathways [60]. Of note, all the S-cone bipolar cells appeared to be ON type, raising the question of where the blue OFF signal might come from. This is not to be confused with the midget OFF bipolar cells that contacts S-cones in the primate retina since there is no midget system in non-primate mammals [for review on this subject see [61]]. One possibility is that an amacrine cell may invert the signals from an S-cone bipolar cell and inhibit the downstream cells, akin to the AII amacrine cell in the rod pathway. Indeed, using S-cone isolating (silent substitution) stimuli, we surveyed copious amounts of amacrine cells in the ground squirrel retina and identified a specific type of glycinergic amacrine cell that was tuned to S-cone input through S-cone bipolar cells (Fig. 3A; [62]). Meanwhile, it has been demonstrated, using multi-electrode recording of chromatic responses from retinal ganglion cells (RGCs) in the ground squirrel retina, that a mosaic of blue-OFF/green-ON RGCs lost their blue-OFF response when the ON pathway was blocked (Fig. 3B; [63]). Similarly, antagonizing glycine receptors also eliminated the blue-OFF response, in good agreement with the notion that blue-OFF signals originate from a glycinergic amacrine cell inverting blue-ON signals. This additional synapse in the S-cone signaling pathway will certainly make it more sluggish as discussed above. Thus, the ground squirrel has proven to be a good model for the study of color pathways in the retina, owing to its cone dominant retina and separate S- and M-cones [61].

Figure 3.

A1) A dye-injected S-cone bipolar cell (magenta) in a ground squirrel retinal slice contacts an S-cone (cyan, S-opsin antibody). Scale bar = 10 μm. A2) Blocking S-cone to S-cone bipolar cell synapse with mGluR6 receptor agonist (L-AP4) or blocking S-cone bipolar cell to S-cone amacrine cell synapse with AMPA/Kainate glutamate receptor antagonist (NBQX) eliminates S-cone amacrine cell light response to S-cone isolating stimulus. A3) Morphology of an S-cone amacrine cell. B) Spike trigger average of four types of ganglion cells simultaneously recorded in the same piece of ground squirrel retina using a multi-electrode array. Blocking S-cone bipolar cell input with a combination of mGluR6 receptor agonist (L-AP4) and antagonist (LY341495) eliminates both the blue-OFF and green-ON responses in blue-OFF/green-ON ganglion cells, while blocking S-cone amacrine cell output with glycine receptor antagonist (Strychnine) eliminates only the blue-OFF response in blue-OFF/green-ON ganglion cells. A is modified from [60] and B is modified from [61].

2.5. Rods, rod synapse, and rod pathways

The minute population of rods in the ground squirrel retina had not been appreciated until West and Dowling [7] reported light and electron microscopic descriptions of “rod-like” receptors in two species of ground squirrel and prairie dog retinas. Similar results from several other ground squirrel species were reported around that time [8,9]. Compared to cones, these “rods” had longer outer segments with disks that were discontinuous with the cell membrane, skinnier inner segments with less densely packed mitochondria in the ellipsoid region, and round nuclei that situated close to outer plexiform layer (OPL). Aside from these common characteristics shared with rods of rod dominant mammals, ground squirrel rods have some distinguishing features, the most significant of which is their synaptic terminals. Instead of having a single large ribbon, it often contains multiple small ribbons, thus making more synapses than a typical mammalian rod does with processes of horizontal cells, bipolar cells, and even neighboring cone teleodendrites [7]. A complete topographic mapping of rods in the ground squirrel retina was described more than two decades later [54]. In California ground squirrel retina, rods account for ~15% of the total photoreceptors, more than originally estimated by West and Dowling [7], albeit in a different ground squirrel species. One possible reason for the underestimation in earlier studies may be the very uneven distribution of rods, from <1,000/mm² in the visual streak area, to more than 13,000/mm² in the peripheral inferior retina. Interestingly, in the visual streak area (about 2 mm inferior to the linear optic nerve head), cones reach their density peak at ~50,000/mm². Thus, with a cone to rod ratio of 50:1, it is similar to that of the human fovea region, ~100 μm from the center [64].

Although the anatomical data of rods in the ground squirrel retina has been well documented, the physiological evidence remains obscure. Early ERG work demonstrated the occurrence of a spectral sensitivity shift (Purkinje shift), the discrimination of a wavelength typical of rhodopsin from a longer wavelength cone opsin, suggesting functional rod photoreceptors in the ground squirrel retina [8]. However, it was noted that not all the animals tested demonstrated the Purkinje shift, suggesting variable rod function between individual animals. Jacobs et. al revisited this subject with a larger sample of California ground squirrels, an additional species (golden mantled ground squirrel), and behavioral tests [65]. The results convincingly showed that about 1/3 of the population tested did not have rhodopsin-mediated scotopic function even though they surprisingly possessed a very similar population of rods! This remains a mystery till this day, but I would like to entertain one possible explanation that was not considered at the time. In examining the genome of another cone dominant mammal, the Chinese tree shrew, it was discovered that rhodopsin gene had a faster evolutionary rate in the tree shrew lineage, presumably due to less selective pressure for night vision under a diurnal lifestyle [66]. In fact, they identified a mutation in rhodopsin that may render it dysfunctional. Thus, similar relaxation of selective pressure might have occurred in the ground squirrel lineage where some individuals within the population bear a rhodopsin (or other genes critical for rod vision) mutation that compromises their rod vision.

Although sparse, rods can be easily recognized in a ground squirrel retinal slice, allowing in-depth electrophysiological studies of rod synapse and connectivity. Using a dual patch clamp recording technique, we recorded rod-cone gap junctional conductance in the ground squirrel retina slices and found it to be comparable to that between cones [67]. Electrical coupling between a rod and a neighboring S-cone has also been observed (unpublished data). A survey of synaptic connection between rods and cone bipolar cells identified the cb2 (AMPA receptor expressing) bipolar cell receives direct rod synaptic input [67]. This was the first synaptic recording of a so-called “tertiary” rod pathway [68]. Such a direct synaptic connection with OFF cone bipolar cells enabled interrogation of features of the rod ribbon synapse, because the detection of vesicle (glutamate) release is not filtered by the slow, metabotropic glutamate receptors on rod bipolar cell - the conventional synaptic partner of rods. It was surprising to discover that vesicle release from rod ribbon synapse was as fast as that of the cone ribbon synapse, offering a speedy, alternative rod signaling pathway.

3. Hibernation

To survive a cold winter with limited food availability ground squirrels opt to hibernate underground till springtime. They often cease food and water intake and radically lower their metabolic rate (to <5%), achieved in part, by lowering their core body temperature to just 2–3° above the ambient temperature. For the arctic ground squirrel their core body temperature can amazingly drop below the freezing point of water [69]. During hibernation, ground squirrels reduce their heart rate to ~20 beats per minute (bpm) and breathe once every 20 mins or so. Such dramatic metabolic change clearly will have a significant impact on every organ system, including the retina which is normally quite energy-demanding.

The first study related to the retina of the hibernating ground squirrel was conducted by Toichiro Kuwabara [70]. This EM study focused on the interface of RPE and the photoreceptor outer segment, but also presented data of cone synaptic terminals. The major findings were a significant loss of outer segments and the disappearance of the synaptic ribbon - both recovering structural integrity after hibernation. Remé & Young [46] performed more extensive experiments and observed shortened, but not absent outer segments, which was confirmed by a recent study with modern imaging techniques [71]. However, the change in the synaptic ribbons at the cone pedicles observed by Kuwabara was verified. This unique hibernation-induced synaptic change has been exploited to study synaptic ribbon function in the context of reversible adult plasticity [5,47]. We further probed the function of the remaining ribbon during hibernation, as well as the functional consequence of losing the major part of the synaptic ribbon to provide a clue to their normal function. We discovered that they are critical for vesicle replenishment, thus an important mechanism for enhancing high temporal frequency signaling (unpublished data). Another interesting hibernation related change reported by Remé & Young [46] is that the mitochondria in cone photoreceptors appeared to undergo reversible modification in their size and numbers. This was also confirmed recently with an EM reconstruction and imaging approach [72].

At a systems biology level, transcriptomic and metabolomic analysis of ground squirrel retinas from awake and hibernating animals have generated interesting sets of data for further investigations on molecular adaptation of hibernation [73]. For example, changes in tricarboxylic acid (TCA) cycle and some of the metabolites may be critical signaling molecules for metabolic adaptation [For review see [74]]. In addition, computational molecular phenotyping (CMP) [75] has been applied to ground squirrel retina, exploring the molecular changes and metabolic states in awake and hibernating retinal tissues at the EM level [5].

4. Disease and injury models

Because of their cone dominant feature and their ability to overcome metabolic stress during hibernation, the ground squirrel ought to be an ideal model animal for retinal diseases. There have been studies on drug induced cone degeneration [76,77], as well as a series of studies on retinal detachment by Steven Fisher’s group [78–81]. For an overview of the literature on this topic, I refer the readers to an excellent recent review [5]. A recent study demonstrated how S-cone bipolar cells in the ground squirrel retina can rewire with the surviving S-cones after laser induced deafferentation of their presynaptic S-cones [82]. What has been missing, but could potentially be very interesting is a model of diabetic retinopathy in the ground squirrel, since hibernating ground squirrels show reversible insulin resistance [83].

5. Beyond the retina

To match the cone dominant input, there are over one million RGCs as output projecting to various subcortical visual areas [84]. There is a plethora of anatomical and physiological studies on ground squirrel retinal projections, subcortical and cortical structures, which are perhaps underappreciated [85–97]. I will simply highlight some outstanding findings from the most recent anterograde tracing study facilitated by intraocular injection of cholera toxin B (CTB) in the California ground squirrels [95]. Retinofugal projections appear to be bilateral, but predominantly contralateral. This determination corrected some previous reports of purely contralateral projections, likely reflecting improved resolution of minor axon fibers and terminals that project to the ipsilateral targets. The suprachiasmatic nucleus (SCN) is one of those targets with heavy innervation, chiefly contralateral, from the retina. The dorsal lateral geniculate (dLGN), a laminar structure comparable to the primate dLGN, comprises the main visual area that contains alternating inputs from ipsi- and contralateral eyes. Additionally, the ground squirrel has a large superior colliculus (SC), the most densely innervated subcortical region, with noticeable columnar organization of retinal inputs. Functional significance of such organization has long been explored [86]. Cells within the column tend to have similar directional and orientational tuning. Similar column structure in SC remain controversial and are active research subjects in mice. Recently, a small population of GABAergic RGCs projecting to SCN has been reported in mice to reduce the sensitivity of SCN neurons to fluctuating light input [98]. Interestingly, similar GABAergic RGCs have been reported in the ground squirrel [99]. Instead of the SCN, they innervate the SC, suggesting that such GABAergic RGCs may be more common than one would expect. The visual cortex of the ground squirrel is about twice as large as that of the rat and contains complex substructures within the primary visual cortex, as well as a large number of extrastriate visual areas. Overall, the ground squirrels have highly developed geniculostriate and tectofugal systems to match their cone dominant retina for high visual performance in daylight.

5. Conclusions and future research

As a non-conventional animal model, the ground squirrel has a lot to offer for vision research. Cone dominance is in no doubt the most attractive feature, for its similarity with the human fovea/macular region that is indispensable to our sight. One of the urgent unanswered questions has to do with the development of such high cone density, which can be extremely useful for research of cone regeneration. To this end, we have conducted both bulk and single cell RNAseq to profile the development of ground squirrel retina and cone photoreceptors and delineated the transcriptomic timeline for cone development (unpublished data). In addition, we generated induced pluripotent stem cells (iPSCs) from the ground squirrel (Ictidomys tridecemlineatus) [100], which allow us to differentiate them to different cells and tissues, including retinal organoids (unpublished data). This will facilitate the study of ground squirrel cone development and rod/cone ratio in vitro. The cone dominant feature of the ground squirrel retina greatly facilitated the research of cone synapse and cone pathways. However, many questions still remain. For example, retinal pathways that encode color have not been completely mapped out. New techniques such as AAV mediated genetic manipulation may bridge the gap between a non-conventional animal model like the ground squirrel and more genetically engineerable model species. Hibernation is another fascinating feature of which we have only scratched the surface. Uncovering the metabolic adaptation achieved during hibernation will certainly improve our understanding of retinal injury and disease, many of which, if not all, are closely associated with metabolic stress.