Eye features and retinal photoreceptors of the nocturnal aardvark (Orycteropus afer, Tubulidentata)

Mammal, vision, tapetum lucidum, cone photoreceptor, rod photoreceptor, opsin 26 coexpression, thyroid hormones 27 28 29 *Corresponding author: 30 [email protected] 31 32 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 2 33

34 The nocturnal aardvark Orycteropus afer is the only extant species in the mammalian order 35 Tubulidentata. Previous studies have claimed that it has an all-rod retina. In the retina of one 36 aardvark, we found rod densities ranging from 124,000/mm² in peripheral retina to 37 214,000/mm² in central retina; the retina of another aardvark had 182,000 – 245,000 38 rods/mm². This is moderate in comparison to other nocturnal mammals. With opsin 39 immunolabelling we found that the aardvark also has a small population of cone 40 photoreceptors. Cone densities ranged from 300 to 1,300/mm² in one animal, and from 1,100 41 to 1,600/mm² in the other animal, with large local variations and no large central-peripheral 42 density gradient. Overall, cones comprised 0.25-0.9% of the photoreceptors. Both typical 43 mammalian cone opsins, longwave-sensitive (L) and shortwave-sensitive (S), were present. 44 However, there was colocalization of the two opsins in many cones across the retina (35 – 45 96% dual pigment cones). Pure L cones and S cones formed smaller populations. This 46 probably results in poor colour discrimination. Thyroid hormones, important regulators of 47 cone opsin expression, showed normal blood serum levels. The relatively low rod density and 48 hence a relatively thin retina may be related to the fact that the aardvark retina is avascular 49 and its oxygen and nutrient supply have to come from the choriocapillaris by diffusion. In 50 contrast to some previous studies, we found that the aardvark eye has a reflective tapetum 51 lucidum with features of a choroidal tapetum fibrosum, in front of which the retinal pigment 52 epithelium is unpigmented. The discussion considers these findings from a comparative 53 perspective. 54 55 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 3 56

57 The aardvark (Orycteropus afer) is the only extant species of the order Tubulidentata in the 58 supraordinal mammalian clade Afrotheria. It is a medium-sized, pig-like mammal native to 59 sub-Saharan Africa (Fig 1). The aardvark is a nocturnal, burrowing species that mostly feeds 60 on ants and termites (for an overview, see, e.g., [1,2]). The strong legs with sharp claws are 61 used to dig out termite and ant nests, as well as abode burrows; the genus name Orycteropus 62 means ‘burrowing foot.’ The aardvark is considered a ‘living fossil,’ because Orycteropus 63 fossils from about 20 million years ago show nearly identical morphological features to those 64 of living aardvarks [2]. It is assumed that aardvarks have an acute sense of smell and hearing, 65 but poor eyesight. However, in contrast to the abundant literature available on the eyes and 66 retinae of many other mammals, the only substantial study of the aardvark retina known to us 67 is that of Victor Franz, published in 1909 [3]. Franz obtained the two eyes of one animal 68 hunted at a zoological expedition and examined them in detail macroscopically and 69 microscopically. The study contains much valuable information, but also some apparent 70 errors, most likely due to the histological methods available at the time. Franz [3] reports a 71 complete absence of cone photoreceptors in the retina, which appears doubtful in the light of 72 more recent research demonstrating cones in nearly all mammals studied to date (reviews: 73 [4,5]). Among the few exceptions without functional cones are some deep-diving whales [6], 74 the subterranean golden moles [7], and Xenarthra [8]. Franz [3] also states that the aardvark 75 has no reflective tapetum lucidum. In contrast, a recent study of the orbital structures and eye 76 tunics of young and adult aardvarks reports the presence of a tapetum lucidum [9]. Nocturnal 77 photographs of aardvarks show a strong eyeshine or ‘glow’ (Fig 1). However, this could also 78 be a reflection off the fundus like the ‘red eye effect’ seen, e.g., in human eyes in flash 79 photographs. When we obtained the relatively well-preserved eyes of two aardvark 80 individuals, we studied them with currently available histological approaches to resolve the 81 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 4 above discrepancies and to add new observations on aardvark retinal anatomy. The present 82 paper reports on general eye features and the photoreceptors. A separate paper shall describe 83 aardvark retinal bipolar cells, horizontal cells, amacrine cells and ganglion cells. 84 85 86 Fig 1. Aardvarks at day and night, note the laterally positioned eyes. (A) Aardvark 1, the 87 post mortem donor of the studied eye, at daytime. (B, C) Another aardvark, flashlight 88 photographs taken at night. When the head is viewed horizontally, there is a bright whitish 89 eyeshine indicating a tapetum lucidum; when viewed from above, the weaker eyeshine 90 appears orange to red. For details see text. Image sources: (A) Frankfurt Zoo; (B, C) Christina 91 Geiger. 92 93 94

95 Tissue 96 This study has used tissue from aardvarks (Orycteropus afer) that was obtained from zoo 97 animals that had died of natural causes. The aardvark IUCN status is ‘least concern’ and no 98 ethical approval was required for use of the tissue. The right eye of a nearly 25 years old male 99 aardvark was obtained when the animal died of old age in the Zoo of Frankfurt am Main, 100 Germany. The animal is termed “aardvark 1” here (Fig 1A). It had an age-related cataract and 101 a mild chronic Uveitis anterior, but the retina appeared macroscopically normal. One day post 102 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 5 mortem, the eye was enucleated during autopsy, punctured behind the cornea for better 103 fixative penetration, and immersion-fixed in 4% formalin for 24 h at 4°C. When punctured, 104 the eye lost some liquefied vitreous and aqueous humor, leading to a collapse of the cornea 105 and a slight deformation of the eyeball during fixation. After fixation the external eye 106 dimensions were recorded, the eye was cut open behind the cornea, and the posterior eyecup 107 with attached retina was washed in 0.01M phosphate buffered saline (PBS, pH 7.4). The 108 retina was carefully dissected from the eyecup after having marked its orientation. It was 109 cryoprotected by successive immersion in 10%, 20% and 30% (w/v) sucrose in phosphate 110 buffer (PB, pH 7.4) containing 0.05% sodium azide, and stored frozen at -20°C until further 111 processing. The eyecup was stored in PBS with 0.05% sodium azide at 4°C. 112 113 The eyes of a nearly 6 years old female aardvark were obtained when the animal died of 114 perinatal complications in the Zoological Garden of Wrocław (Poland). The animal is termed 115 “aardvark 2” here, it was genetically unrelated to aardvark 1. The eyes were enucleated 116 immediately post mortem and immersion-fixed in 4% buffered formaldehyde solution, they 117 have been used in a previous study of aardvark eye and orbital features [9]. The right eye was 118 kept in the fixative for 1 week and then embedded in paraffin for sectioning. The left eye was 119 permanently stored in the fixative, and only small pieces of the retina were available for the 120 present study. Probably due to the long fixation time of this eye (ca. 7 years), some of the 121 antibodies used here did not work (see below and Results). 122 123 For frozen vertical sections of the retina (i.e., perpendicular to the retinal layers), pieces of 124 retina were transferred from 30% sucrose to tissue freezing medium (Leica Biosystems, 125 Wetzlar, Germany), frozen, sectioned at 16 μm thickness with a cryostat (Leica CM 3050 S, 126 Wetzlar, Germany), and collected on Superfrost Plus slides (Menzel Gläser, Braunschweig, 127 Germany). 128 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 6 129 For electron microscopy, small pieces of the sclera and presumed tapetum lucidum were 130 stained as previously described [10]. Briefly, the samples were stained in a solution 131 containing 1% osmium tetroxide, 1.5% potassium ferrocyanide, and 0.15 M cacodylate 132 buffer. The osmium stain was amplified with 1% thiocarbohydrazide and 2% osmium 133 tetroxide. The tissue was then stained with 2% aqueous uranyl acetate and lead aspartate. The 134 tissue was dehydrated through an 70%–100% ethanol series, transferred to propylene oxide, 135 infiltrated with 50%/50% propylene oxide/epon medium hard formulation (EMbed 812, 136 Electron Microscopy Sciences; [11]), and then 100% epon medium hard. The epon medium 137 hard infiltrated tissue was transferred into multi-well embedding molds (Electron Microscopy 138 Sciences) and hardened at 60°C. For scanning electron microscopy (SEM), a few serial 139 sections of 50 nm were taken with a Diatome ultra diamond knife and collected on glow 140 discharged silicon wafers and dried on a heating plate at 50 °C until the water was fully 141 evaporated. The wafers were mounted with silver paint (Plano) on a sample holder and 142 images were taken with a Supra55 (Leica) SEM. For transmission electron microscopy 143 (TEM), 50-nm-thick sections were cut with an Ultra diamond knife and transferred on carbon-144 coated copper grids with a hole size of 35/10 nm (S35/10, Quantifoil, Electron Microscopy 145 Sciences). Images were recorded with an analytical electron microscope (JEM-2200FS, Jeol) 146 at an energy of 200 keV with a CMOS camera (TEM-CAM F416, TVIPS). 147 148 For comparison of the choroid, vertical cryo-sections of the formalin-fixed eye of a captive 149 adult African elephant (Loxodonta africana) from a German zoo was used. In agreement with 150 CITES regulations the eye was collected during necropsy for pathological investigations 151 performed by the Institute for Zoo and Wildlife Research, Berlin (IZW) after the animal had 152 to be euthanized because of severe disease unresponsive to treatment. 153 154 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 7 Immunohistochemistry 155 Immunohistochemistry was performed on vertical sections of the retina, as well as on 156 unsectioned retinal pieces from various regions to assess cell populations in flat view. 157 Immunolabelling followed standard protocols. Briefly, sections on the slide and free-floating 158 pieces were preincubated for 1 h in PB with 0.5% Triton X-100 and 10% normal donkey 159 serum (NDS). Incubation in the primary antibody/antiserum solution, made up in PB with 3% 160 NDS and 0.5% Triton X-100, was overnight at room temperature for sections, and 3 days at 161 room temperature or 4 days at 4°C for unsectioned pieces. Multiple immunofluorescence 162 labelling for simultaneous visualization of several antigens was performed by incubation in a 163 mixture of the antisera. Table 1 lists all primary antibodies used. The cone opsin antisera 164 JH492, JH455 and sc-14363 have been used in several previous studies to reliably label the 165 respective opsins in a range of mammals [12-16]. 166 167 Table 1. Primary antibodies and cell markers used 168 Antigen / marker Immunogen / target structure Antibody host species, catalog #, RRID Dilution Source Rod opsin N-terminal region of bovine rhodopsin1 Mouse monoclonal, Name: rho4D2, RRID: AB_2315273 1:1000 Gift of R. S. Molday, University of British Columbia Life Sciences Centre, Vancouver, Canada L cone opsin C-terminal 38 amino acids of human red cone opsin2 Rabbit polyclonal, Name: JH 492, RRID: AB_2315259 1:2,000 Gift of J. Nathans, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA S cone opsin C-terminal 42 aa of human blue cone opsin2 Rabbit polyclonal, Name: JH 455, RRID: AB_2313807 1:5,000 Gift of J. Nathans, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA S cone opsin 20 aa peptide mapping near N- terminus of human blue cone opsin3 Goat polyclonal, Cat# sc-14363, RRID: AB_2158332 1:500 Santa Cruz Biotechnology CtBP2 (C- Terminal Mouse CtBP2, aa. 361-445 Mouse monoclonal, Cat# 612044, 1:5000 BD Biosciences .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 8 Binding Protein-2) RRID: AB_399431 CtBP2 Rat CtBP2, aa 431-445 Rabbit polyclonal, Cat# 193 003, RRID: AB_2086768 1:5000 Synaptic Systems GS (Glutamine synthetase) Müller glia Mouse monoclonal, Cat# 610517, RRID: AB_397879 1:500 BD Bioscience GFAP Glial fibrillary acidic protein Mouse monoclonal, Cat# G3893, RRID: AB_477010 1:500 Sigma-Aldrich H3K4me3 Euchromatin Rabbit polyclonal, Cat# ab8580, RRID: AB_306649 1:500 abcam H4K20me3 Heterochromatin Mouse monoclonal, CMA423 1:500 Generated in Hiroshi Kimura’ lab, Tokyo University Lamins A/C Inner nuclear membrane protein Mouse serum Undiluted Gift of Harald Herrmann, German Cancer Research Center LBR Inner nuclear membrane protein Guinea pig serum 1:50 Generated in Harald Herrmann’s lab, German Cancer Research Center PNA-647 (Peanut agglutinin) General cone marker Cat# L-32460 1:100 Molecular Probes NeuN Synthetic peptide NeuN Rabbit monoclonal (EPR12763), Cat# ab177487, RRID: AB_2532109 1:100 abcam NeuroTrace (Ex 530 / Em 615) Fluorescent Nissl stain Cat# N-21482 1:100 Molecular Probes Isolectin B4, biotinylated Blood vessel marker Cat# B-1205 1:50 Vector Laboratories 1Ref. [86], 2Ref. [87], 3Ref. [12]. 169 170 Binding sites of the primary antibodies were visualized by indirect immunofluorescence, with 171 a 1.0-1.5 h incubation of the tissue in the secondary antiserum, or in a mixture of appropriate 172 secondary antisera in the case of several primary antibodies. We used secondary antisera 173 conjugated to Alexa 488, Alexa 647, Cy3 and Cy5 in appropriate combinations. Omission of 174 the primary antibodies from the incubation solution resulted in no staining. In addition to 175 antisera, we used the fluorescent markers peanut agglutinin (PNA) and NeuroTrace for certain 176 cell types (Table 1). After immunolabelling, sections were incubated in a solution of 4,6-177 diamidino-2-phenylindole (DAPI) as a fluorescent nuclear stain to reveal the general retinal 178 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 9 layering. Choroidal blood vessels were labeled by biotinylated isolectin B4 (Table 1; 179 incubation was 1 h for sections, 2 h for unsectioned pieces) and visualized by a subsequent 180 1.0-1.5 h incubation in streptavidin coupled to Alexa 549 or 488. Tissue was coverslipped 181 with an aqueous mounting medium (AquaPoly/Mount, Polysciences Inc., Warrington, PA, 182 USA; or Dako Fluorescence Mounting Medium S3032, Dako North America Inc., 183 Carpinteria, CA, USA). 184 185 In retinal pieces from the long-fixed eye of aardvark 2, the S cone opsin antiserum sc14363 186 did not work, hence assessment of the cone opsin pattern was done by sequential double-187 labeling with the two rabbit antisera JH492 against the L cone opsin and JH455 against the S 188 cone opsin as follows. One retinal piece was first incubated in a JH492 solution and then in a 189 donkey-anti-rabbit antiserum conjugated to Alexa 488. Then the piece was incubated in a 190 JH455 solution and finally in a donkey-anti-rabbit antiserum conjugated to Cy3. This 191 secondary antiserum bound to both primary antisera, hence all cones were labeled by Cy3. 192 The cones also labeled by Alexa 488 were those that contained L cone opsin, and cones only 193 labeled by Cy3 were pure S cones. In a neighboring piece of retina, the order of labeling was 194 reversed: First incubation in the JH455 solution and visualization with the donkey-anti-rabbit 195 antiserum conjugated to Alexa 488, then incubation in the JH492 solution and visualization 196 with the donkey-anti-rabbit antiserum conjugated to Cy3. Again, all cones were labeled by 197 Cy3, but here the cones also labeled by Alexa 488 were those that contained S cone opsin, and 198 those only labeled by Cy3 were pure L cones. Combining the data from the two pieces 199 provided the total cone density, the percentages of pure L and S cones, and the proportion of 200 cones containing L and S opsin, from which the percentage of dual pigment cones could be 201 calculated for that retinal region. 202 203 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 10 For assessment of the retinal vascularization, a retinal piece of 5.5 x 4.0 mm size containing 204 the optic nerve head was stained with a 3,3’-diaminobenzidine (DAB) reaction to selectively 205 visualize the endogenous peroxidase in the vasculature. The retinal piece was washed in 206 0.05% Tris buffer (TRIS, pH 7.6), then incubated for 20 min in a solution of 0.05% DAB in 207 TRIS, after which hydrogen peroxide was added to the incubation solution at a final 208 concentration of 0.01%, and the incubation was continued for 10 min until the peroxidase 209 reaction had fully developed. The reaction was stopped by several washes in TRIS and then 210 PB. The retinal piece was flat-mounted on a slide and coverslipped with AquaPoly/Mount. 211 For assessment of the retinal pigment epithelium (RPE) after removal of the retina, pieces of 212 the thin RPE layer from the central and peripheral fundus were gently removed from the 213 underlying tissue, flat-mounted on a slide and coverslipped with AquaPoly/Mount. 214 215 Imaging and analysis 216 The RPE and the DAB-labelled vasculature of the optic nerve head were analyzed with a 217 Zeiss Axioplan 2 microscope by differential interference contrast. Micrographs were taken 218 with a CCD camera and the Axiovision LE software (Carl Zeiss Vision, Germany). The 219 immunofluorescence-labelled sections and retinal pieces were analyzed with a laser scanning 220 microscope (LSM) Olympus FluoView 1000 using the FV 1.7 software (Olympus), or with a 221 Leica TCS SP5 or a Leica TCS SP8 confocal microscope. LSM images and z-stack 222 projections were examined with ImageJ (https://imagej.net); cells were counted using the cell 223 counter plugin. Images for illustration were adjusted for brightness and contrast using Adobe 224 Photoshop. Irrespective of the fluorescent dye used to visualize a label, labels are shown in 225 the RGB channels that are most suitable to illustrate label combinations. For the benefit of 226 red/green-blind readers, combinations of magenta and green are preferred over red and green. 227 228 Thyroid hormone 229 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 11 Thyroid hormone (TH), via its receptor TRβ2, is an important regulator of cone spectral 230 identity by repressing S opsin and activating L opsin in developing and adult retina (see 231 Discussion). Because of the L and S opsin co-expression in a large proportion of the aardvark 232 cones, we were interested to know whether the serum TH levels in aardvark differ from those 233 in other mammals. Frankfurt Zoo, during medical check-ups, had obtained five blood counts 234 of aardvark 1 over the last three years of his life, and one blood count of his 21 years old son 235 (here termed “aardvark 3”), all including serum TH levels. The blood analysis was done by 236 the commercial veterinary clinical diagnostics laboratory LABOKLIN (Bad Kissingen, 237 Germany). 238 239

240 Most findings reported here came from aardvark 1. The retina of aardvark 2 was used for 241 comparison of the photoreceptor findings. 242 243 General eye features 244 The eye of aardvark 1 had an equatorial diameter of ca. 23.0 mm and an axial length of ca. 245 21.2 mm. The axial length probably is an underestimate because the cornea had collapsed and 246 its original curvature was estimated (Fig 2A). The cornea was elliptical with a naso-temporal 247 diameter of 18.8 mm and a dorso-ventral diameter of 15.5 mm (mean 17.1 mm); the ratio of 248 mean corneal diameter to eye equatorial diameter was 0.75, and the ratio of mean corneal 249 diameter to eye axial length was 0.81. The lens diameter was 13.3 mm and the lens thickness 250 9.5 mm (Fig 2B). The curvature was stronger at the posterior than at the anterior side of the 251 lens (not illustrated). The ratio of lens diameter to eye equatorial diameter was 0.58, the ratio 252 of lens thickness to eye axial length was 0.45. The poor pigmentation of the choroid and 253 sclera in the present eye confirms the observations by Franz [3]. 254 255 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 12 256 Fig 2. General eye features (aardvark 1). (A) Intact eye, frontal view with attached rectus 257 muscles. The cornea had collapsed when the eye was punctured for better fixative penetration 258 to the retina. (B) Anterior part of the opened eye from the vitreous side, showing the lens and 259 ciliary body. (C) Opened eyecup showing the fundus with the retina in situ. There is a light 260 horizontal band where the retinal pigment epithelium (RPE) is weakly pigmented or 261 unpigmented. The transition to the pigmented peripheral RPE is gradual at the dorsal side and 262 with a rather sharp boundary at the ventral side. (D) Eyecup after removal of the retina. Some 263 RPE also came off during the preparation, showing an unpigmented, whitish-yellow choroid. 264 The optic disc (OD) is located ventral to the light horizontal band of (C). (E) Optic disc in the 265 isolated retina, DAB-reacted for blood vessels. There are only a few capillaries present in the 266 optic disc (arrow heads), and no blood vessels exit it to supply the surrounding retina. Around 267 the optic disc there is an accumulation of pigment. (F, G) Light microscopic images of flat-268 mounted RPE pieces from peripheral (F) and central fundus (G). In the periphery, all RPE 269 cells contain densely packed melanin granules (F); centrally, only very few RPE cells are rich 270 in melanin granules, the vast majority of RPE cells contains little or no melanin (G). Eye 271 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 13 dimensions in (A, C, D) can be determined from the millimeter graph paper in (C). The scale 272 bar in (F) applies to (F, G). d, dorsal; n, nasal; t, temporal, v, ventral. 273 274 The fundus in the opened eyecup showed a bright yellowish band extending from the 275 temporal to the nasal periphery. This bright band had the appearance of a tapetum lucidum, its 276 dorso-ventral width was larger in the temporal than the nasal fundus. Its relatively sharp 277 ventral boundary ran along the horizontal midline of the fundus, its dorsal boundary showed a 278 more gradual transition to the pigmented part of the fundus (Fig 2C, D). The ventral half and 279 the dorsal periphery of the fundus were covered by brown-black retinal pigment epithelium 280 (RPE). The optic nerve head (optic disc, OD) was located centrally on the temporal-nasal eye 281 axis and ventral to the geometric center of the eyecup, about one OD diameter below the 282 ventral boundary of the bright fundus band (Fig. 2D). When the retina was removed, the 283 remaining thin RPE layer consisted of RPE cells (melanocytes) that contained a high density 284 of melanin granules in the dark-appearing parts of the fundus (Figs 2D, F, 3A). In the central 285 bright-appearing band, the large majority of RPE cells contained very few or no melanin 286 granules, and only some single cells or small cell clusters contained ample melanin (Figs 2G, 287 3A). The absence of pigmentation in a horizontal band of the central fundus and the 288 associated tapetum-like reflection explain the eyeshine differences seen at different angles 289 (Fig 1B, C). When the eye is seen horizontally, the strong reflectivity of this band produces a 290 bright whitish to yellowish eyeshine. When the eye is seen from above, the lower reflectivity 291 of the more pigmented ventral fundus produces a fainter orange to red eyeshine that resembles 292 the ‘red eye effect’ seen in flash photographs of human eyes with their pigmented fundus. 293 294 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 14 295 Fig 3. Choroid and tapetum lucidum (aardvark 1). (A) Higher power view of the central 296 and ventral midperipheral fundus of the aardvark eye (c.f. Fig. 1). Below the partly removed 297 RPE layer, the whitish unpigmented choroid with its red blood vessels is visible. (B) SEM 298 micrograph of subcellular structures from the RPE-facing side of a transverse choroid section, 299 showing collagen fibrils arranged in parallel bundles with different orientations that indicate a 300 tapetum lucidum. In the upper image part, the fibrils are longitudinally sectioned; in the 301 bottom image part, they are cross-sectioned with round to oval profiles. (C) At higher 302 magnification, the fibrils show the typical cross-striation of native collagen (TEM image). (D) 303 In cross-section, the fibrils show a shell and core of higher electron density (SEM image). (E) 304 Differential interference contrast (DIC) image of a vertical cryo-section of the aardvark 305 choroid (Ch) and tapetum (Tap). The poorly pigmented choroid shows cross-sections of blood 306 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 15 vessels of various calibers, the tapetum shows a horizontal striation indicating tapetal laminae. 307 (F) DIC image of a vertical cryo-section of the choroid and tapetum of an African elephant 308 for comparison. The choroid is strongly pigmented, and the tapetum is thicker and more 309 conspicuously striated than in the aardvark. (G) Aardvark vertical cryo-section of the choroid 310 and tapetum with blood vessel labelling by isolectin (red). The choroid part contains vessels 311 of larger and smaller caliber, the tapetum layer is relatively thin, and the choriocapillaris at the 312 border to the RPE is densely filled with capillaries. (H) African elephant vertical cryo-section 313 of the choroid and tapetum with blood vessel labelling by isolectin (red) for comparison. The 314 image shows a vertical choroidal blood vessel supplying the CC capillaries. For the sections 315 of (E-H), the choroid has been removed from the sclera, so the sections do not show the full 316 thickness of the choroid. (I) Flat view of the aardvark choriocapillaris, labelled by isolectin 317 (red) and showing the dense capillary net. The scale bar in (E) applies to (E, F), the scale bar 318 in (G) applies to (G-I). 319 320 Choroid and tapetum lucidum 321 Once the retina was removed, the thin RPE layer readily detached from the choroid in shreds 322 (Figs 2D, 3A). The strongly vascularized choroid was unpigmented and appeared whitish with 323 a mother-of-pearl-like reflection throughout the fundus; in the fundus regions with pigmented 324 RPE, this choroidal reflection was concealed (Fig 3A). Electron microscopy of transverse 325 choroid sections showed that at the RPE-facing side of the choroid, there were fibrillar 326 structures arranged in parallel, with different orientations in neighbouring domains (Fig 3B). 327 In some domains the fibrils were densely packed, in others they were less dense or sparse. 328 Fibril diameters seen in cross-sections were 120-310 nm, being 150-250 nm in most cases. 329 Observed fibril lengths were up to about 7 µm. In TEM images, the fibrils showed the typical 330 cross-striation of native collagen, indicating a choroidal tapetum lucidum fibrosum (Fig 3C). 331 However, in SEM images, cross-sections of the fibrils showed a substructure with a more 332 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 16 electron-dense shell and core, which is more characteristic of the rodlets of a tapetum 333 cellulosum (Fig 3D). This is addressed in the Discussion. Overall, the presumed tapetum layer 334 is thinner and less conspicuously striated than, e.g., in the African elephant, like the aardvark 335 a member of the Afrotheria with an avascular retina (Fig 3E, F). The boundary of the choroid 336 and tapetum to the RPE is formed by the choriocapillaris, a dense capillary net that was 337 labelled by the endothelial cell marker isolectin B4 in vertical sections of both aardvark (Fig 338 3G) and African elephant (Fig 3H). The dense mesh of the aardvark choriocapillaris is 339 particularly obvious in flat view (Fig 3I). 340 341 General Retina Features 342 We confirm that the aardvark retina is avascular. In the fundus, there were no obvious blood 343 vessels emerging from the optic disc or extending across the retina. Staining of blood vessels 344 with DAB in a piece of central retina containing the optic disc revealed a few small capillaries 345 within the optic disc and confirmed that there were no blood vessels extending outside the 346 optic disc and into the retina (Fig 2E). 347 348 In the vertical sections, retinal thickness ranged from ca. 120 µm to ca. 180 µm. This is an 349 estimate, given the fact that the sections may not be exactly vertical and that the length of 350 photoreceptor outer segments may not be fully preserved. The layering of the aardvark retina 351 conformed to the typical mammalian pattern (Fig 4). The outer nuclear layer (ONL) with the 352 photoreceptor somata was the thickest layer, indicating a dominance of rod photoreceptors, 353 which is the situation seen in most mammals. The ONL had six to nine soma tiers in more 354 central retina and five to seven tiers in more peripheral retina. The inner nuclear layer (INL) 355 had three to four soma tiers in central retina and two to three soma tiers in peripheral retina. 356 The ganglion cell layer (GCL) was sparsely populated by somata. The narrower outer 357 plexiform layer (OPL) and broader inner plexiform layer (IPL) separated the soma layers. 358 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 17 359 360 Fig 4. General retinal features (aardvark 1). (A) Overview of a vertical retinal section 361 labelled with the fluorescent Nissl stain NeuroTrace, revealing the retinal layering. Cell 362 bodies in the outer nuclear layer (ONL) and inner nuclear layer (INL) are stacked in several 363 tiers. The photoreceptor outer and inner segments (OS+IS) are also labelled. The ganglion cell 364 layer (GCL) is sparsely populated by cells of various soma sizes. A large soma of a putative 365 alpha cell is marked by an arrowhead. (B, C) Double labelling with an antibody against the 366 neuronal marker NeuN and with NeuroTrace. NeuN only labels a few presumed amacrine cell 367 somata in the INL and some somata in the GCL (B). The NeuroTrace counterstain shows the 368 position of the NeuN somata in the layers (C). (D, E) Immunolabelling for glutamine 369 synthetase shows the Müller cells forming the retinal glia scaffold (D). Counterstaining with 370 DAPI (E) shows that the Müller cells have their somata in the INL and vertically extend their 371 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 18 processes from the inner limiting membrane (ILM) formed by their endfeet to the outer 372 limiting membrane (OLM). The images are maximum intensity projections of confocal image 373 stacks. OPL, outer plexiform layer; IPL, inner plexiform layer. Scale bars are 100 µm, scale 374 bar in (C) applies to (B-E). 375 376 Müller cells, the scaffolding radial macroglia of the retina, were specifically labelled by an 377 antibody against glutamine synthetase (Fig 4D, E). They were present at a high density and 378 had the mammalian-typical morphology. Their somata were located in the INL. Their inward 379 processes traversed the IPL and terminated in the Müller cell endfeet that ended at the inner 380 limiting membrane (ILM), separating the retina from the vitreous. In some of the cells these 381 processes bifurcated and formed two endfeet. Their outward processes encircled the 382 photoreceptor somata in the ONL and ended at the outer limiting membrane (OLM) that lies 383 between the ONL and the photoreceptor inner segments. An antibody against glial fibrillary 384 acidic protein (GFAP) did not label any structures (not illustrated). Hence, there are no 385 astrocytes in the aardvark retina, as mammalian astrocytes specifically express GFAP [17]. 386 This is in line with the absence of retinal blood vessels (see Discussion). Furthermore, the 387 absence of GFAP label in the Müller cells indicates that the studied retina was healthy. The 388 Müller cells of healthy retinae have very low or no GFAP expression, but show reactive 389 gliosis with dramatically upregulated GFAP expression to practically all forms of retinal 390 stress, i.e. to various retinal diseases and injuries (reviews: [18,19]. 391 392 Rod photoreceptors 393 In the retina of aardvark 1, rod photoreceptors were identified by labelling with an antibody to 394 rod opsin (Fig 5A, B), and by their characteristic nuclear morphology (Fig 5C, D). Their outer 395 segments showed the most intense rod opsin label and formed a densely packed layer at the 396 outer retinal surface (Fig 5A). The used antibody rho4D2 also, but less intensely, labelled the 397 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 19 other parts of the rod cells, particularly the soma cytoplasm in the ONL and the axonal ending 398 in the OPL (Fig 5B), which is common in mammals. The large majority of the photoreceptor 399 somata in the ONL showed rho4D2 label and hence are rod somata. This fits the very low 400 aardvark cone densities (see below). 401 402 403 Fig 5. Rod photoreceptors (aardvark 1). (A, B) Vertical retinal section, rod opsin label 404 (green). (A) The rod outer segments are strongly labelled, DAPI counterstaining (blue) shows 405 the retinal nuclear layers. (B) Overexposure of the same field shows less strong opsin label in 406 the rod somata in the outer nuclear layer (ONL) and the rod axonal spherules in the outer 407 plexiform layer (OPL). Clearly, the vast majority of ONL somata belong to rods. (C-G) DAPI 408 nuclear staining of various retinal neurons to reveal their heterochromatin arrangement. (C) 409 Overview of a DAPI stained section, nuclei in the ONL (top) are more intensely stained than 410 those in the INL (middle) and GCL (bottom), confirming the appearance seen in (A). (D) The 411 rod nuclei show a “semi-inverted” nuclear architecture with most of the heterochromatin 412 clustered in the nuclear centre, often in two aggregates, but with some extensions towards the 413 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 20 nuclear periphery. (E) In contrast, the cone nuclei (arrowhead) have several smaller 414 heterochromatin clusters localized towards the nuclear periphery (conventional nuclear 415 architecture). The same conventional heterochromatin arrangement is seen in cells of the INL 416 (F) and in retinal ganglion cells (G). (H) In the rods, euchromatin (immunolabelled by anti-417 H3K4me3, green) is located mostly in the nuclear periphery and in the gaps between the 418 heterochromatin clusters (DAPI, red). (I) In the rod nuclei, heterochromatin (immunolabelled 419 by anti-H4K20me3, green) colocalizes with the DAPI staining (red), the merge of the labels 420 appears yellow. (J-L) Immunolabelling for lamin A/C (red) and LBR (green), counterstained 421 with DAPI (blue). The aardvark retina shows a presence of lamin A/C in the neuronal nuclei 422 in all layers (J, K). As a positive control for labeling with the anti-LBR antibody, the nucleus 423 of a microglial cell (arrowhead) expressing LBR but not lamin A/C is shown in (K). LBR 424 label is only present in microglial cells, not in any neurons. (L) A rod nucleus (left) and a 425 cone nucleus (right, arrowhead) with labelled lamin A/C. For layer abbreviations, see Fig 3. 426 Scale bar in (B) applies to (A, B). 427 428 In the vast majority of eukaryotic cells, the euchromatin is located in the centre of the nucleus 429 and the heterochromatin in the nuclear periphery. In contrast, the rod nuclei of nocturnal 430 mammals have a unique inverted chromatin architecture with heterochromatin aggregated in 431 the nuclear centre and euchromatin arranged at the nuclear periphery. This unusual chromatin 432 arrangement evolved as an adaptation to night vision because it reduces light scattering in the 433 thick retinae of nocturnal species, enhancing their ability to detect low intensity light [20-22]. 434 In the nocturnal aardvark, the heterochromatin in the rods was clustered in two large central 435 granules, thus the nuclei seem to be inverted (Fig 5D, H, I). At the same time, the internal 436 heterochromatin exhibited several protrusions towards the nuclear envelope (Fig 5D), making 437 these nuclei semi-inverted. All other retinal cells, including the cones, had the conventional 438 nuclear architecture with heterochromatin attached to the nuclear periphery or nucleolus (Fig 439 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 21 5E-G). The nuclear envelope protein lamin A/C was present in all retinal nuclei including the 440 rod nuclei, whereas the lamin B receptor (LBR) of the nuclear envelope was only present in 441 retinal microglial cells, not in any retinal neurons (Fig 5J-L). The functional implications of 442 the nuclear inversion and the role of nuclear envelope proteins in this process are addressed in 443 the Discussion. 444 445 In the retina of aardvark 1, photoreceptor densities (and hence rod densities) were estimated 446 from counts of ONL nuclei in DAPI-stained vertical sections at 16 positions from the visual 447 streak region, midperipheral and peripheral retina. The observed range was about 124,000 – 448 214,000 photoreceptors (rods)/mm², with densities decreasing from central to peripheral 449 retina. In the retina of aardvark 2, photoreceptor densities were estimated from counts of ONL 450 somata at seven positions in vertical paraffin sections from central and midperipheral retina. 451 Here the range was about 182,000 – 245,000 photoreceptors (rods)/mm². The ONL had five to 452 nine tiers of photoreceptor somata like the ONL of aardvark 1. It is possible that aardvark 2 453 had higher rod densities than aardvark 1, but the larger shrinkage of paraffin-embedded tissue 454 may also have artefactually increased the cell density. 455 456 Cone photoreceptors 457 The cone photoreceptors were identified by labelling with antisera to the longwave-sensitive 458 (L) cone opsin and the shortwave-sensitive (S) cone opsin in retinal sections (Fig 6A, B) and 459 in flat-mounted retinal pieces from various retinal regions (Fig 6C, S1 Fig, S2 Fig). The two S 460 opsin antisera sc14363 (directed against an N-terminus epitope) and JH455 (directed against a 461 C-terminus epitope) showed complete colocalization of labelling (not illustrated). This is 462 evidence that a full-length, functional S opsin is present. Interestingly, a very large proportion 463 of aardvark cones throughout the retina showed co-expression of the L and S opsin. 464 465 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 22 466 Fig 6. Cone photoreceptors (aardvark 1). (A, B) Vertical retinal sections double-467 immunolabelled for S cone opsin (A1, B1, red) and L cone opsin (A2, B2, green). In the 468 merged images, DAPI counterstaining in blue shows the retinal nuclear layers (A3, B3). Most 469 cones co-express S and L opsin, arrowheads point to pure S cones. In many cones of the 470 region shown in (A), both the S opsin and the L opsin label extend throughout the cone from 471 the OS to the cone pedicle in the OPL; in the region shown in (B), the L opsin label is 472 restricted to the OS. (C) Double immunolabelled cones in a flat-mounted piece from ventral 473 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 23 midperipheral retina. S opsin label (C1, magenta) and L opsin label (C2, green) are 474 colocalized in most cones (C3). Three of the pure S cones are indicated by arrowheads, pure L 475 cones are not present in this field. The aureole seen around most cones is an artefact; as the 476 cones were photographed from the vitreous side of the retina, there is light scatter at many 477 cellular structures. (D) Vertical section double-immunolabelled for S opsin (D1, in red, 478 showing the S cone pedicles in the OPL) and the synaptic ribbon marker CtBP2 (D2, in 479 green). Most of the small CtBP2 structures in the OPL are ribbons of rod spherules; the merge 480 (D3) shows that cone pedicles do not have the ribbon/CtBP2 clusters seen in other mammals. 481 As expected, many somata in the INL are also CtBP2-labelled. (E) Flat-mounted retinal piece 482 double-labelled for S opsin (E1, magenta) and CtBP2 (E2, green). The focus is on the cone 483 pedicles in the OPL. The merge (E3) confirms the absence of cone-typical ribbon/CtBP2 484 clusters at the pedicle. (A-E) are maximum intensity projections of confocal image stacks. 485 The stack in (C) starts at the level of the intensely labelled cone outer segments and ends at 486 the cone soma level. In this region, faint S opsin label extended throughout the cone, whereas 487 L opsin label was restricted to the outer segment in most of the cones. The stack in (E) is of 3 488 focal images spaced 0.5 µm apart. For layer abbreviations, see Fig. 3. Scale bar in (B3) is 100 489 µm and applies to (A, B); scale bar in (C3) is 50 µm; scale bars in (D3) and (E3) are 20 µm. 490 491 In sample fields across the retina of aardvark 1, between 35% and 96% were such dual 492 pigment cones. A substantial proportion of the cones were pure S cones without L opsin 493 expression (some marked in Fig 6A-C). In dorsal retina, between 2% and 37% of the cones 494 were pure S cones, in ventral retina, this percentage was between 22% and 65%. Hence, it 495 appears that the proportion of pure S cones is markedly higher in ventral than dorsal retina, 496 but the large variation of percentages between sampling fields also indicates a high local 497 variability. Only a small proportion of the cones were pure L cones without S opsin 498 expression. In many counting fields containing between 100 and 300 cones, pure L cones 499 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 24 were absent; in other fields, pure L cones comprised 0.3% to 7% of the cones with no obvious 500 regional trend. 501 502 In the retina of aardvark 2, the proportions were somewhat different. In several counting 503 fields from a sample region in far peripheral retina, 67-80% of the cones were co-expressing 504 L and S opsin, 15-21% were pure L cones, and 6-12% were pure S cones. In another region 505 from an unknown but probably also relatively peripheral location, 86-95% of the cones were 506 co-expressing L and S opsin, 2-6% were pure L cones, and 3-8% were pure S cones (S1 Fig). 507 508 These percentages have to be considered with some reservation. First, the relative intensity of 509 the L and S opsin label varied considerably between cones, in some cones one of the labels 510 was barely above background. This was particularly obvious in aardvark 2 (S1 Fig) but can 511 also be seen in aardvark 1 (Fig 6C). Hence, assessment of opsin co-expression has a 512 subjective component. Second, in many retinal regions, the L opsin label was restricted to the 513 cone outer segments (Fig 6B), whereas the S opsin label typically extended throughout the 514 cone including the soma, axon, and cone pedicle (Fig 6A, B, D, E). The cone opsin labelling 515 revealed some degree of post mortem outer segment damage, some outer segments were 516 elongated, others were just small stumps (see Fig 6C). Hence, S opsin-labelled cones could be 517 identified rather reliably by focusing through the outer retina to their soma level, whereas L 518 opsin expression may have been missed in cases of outer segment loss. Nevertheless, there 519 were many sampling fields where the cone outer segments were partially preserved, so the 520 above proportions of pure S and L cones are at least semi-quantitatively reliable. 521 522 The synaptic ribbon marker CtBP2 mainly revealed the single synaptic ribbons of rod 523 synaptic endings (rod spherules) in the OPL (Fig 6D, E). These synaptic ribbons often showed 524 the horseshoe shape that is typical for mammalian rod ribbons (Fig 6E2). In contrast to the 525 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 25 situation in other mammals, CtBP2 did not show the typical clustering of ribbons in the cone 526 pedicles; there were only a few or no CtBP2 puncta at the pedicles of S cones. This is 527 addressed in the Discussion. 528 529 The general cone marker peanut agglutinin (PNA) labelled the S cone pedicles in the aardvark 530 retina (S2 Fig). This was evident by the overlapping label of the pedicles by the S opsin 531 antiserum and PNA. The lateral displacement of the pedicles against the cone outer segments 532 (S2C Fig) is mostly due to slight tissue shearing during the mounting of the retinal pieces and 533 could be followed by tracing the stained S cone axons through the ONL in the image stacks. 534 The PNA label of cone pedicles varied in intensity and size; in a few S cone pedicles it was 535 absent or too faint to be detected. Surprisingly, the L cone pedicles did not show PNA 536 labelling. This was checked in several L cones found in appropriately stained retinal pieces. 537 Two pure L cones are present in the field shown in S2 Fig. It remains open whether PNA does 538 not label the L cone pedicles at all, or whether the label is below the detection threshold of our 539 staining. 540 541 For aardvark 1, cone densities were assessed in more than 70 sample fields in flat-mounted 542 pieces coming from various positions across central and peripheral retina. The sampled tissue 543 included a large piece of temporal retina that contained the presumed area centralis. Total 544 cone densities, i.e., of S cones, L cones and dual pigment cones, showed no strong central-545 peripheral gradient. Highest densities of up to 1,100 – 1,300 cones/mm² were present in a 546 region about 4 mm temporal to the optic disc. Density minima of about 300 cones/mm² were 547 present in a few fields in dorsal peripheral retina, but many other fields in dorsal, ventral, and 548 nasal periphery had densities of 400 to 1,100 cones/mm². In pieces from midperipheral retina, 549 densities ranged from 600 to 1,000 cones/mm². In the region of the bright (unpigmented) 550 horizontal fundus band described above (see Fig 2C), the cone density did not increase, fields 551 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 26 from a temporal and a nasal region of the band had 600 – 900 cones/mm². This suggests that 552 the band does not represent a retinal specialization (visual streak) at the level of the cone 553 population. With rod densities of up to 214,000/mm² in central and down to 124,000/mm² in 554 peripheral retina, the above cone densities correspond to cone percentages of 0.25 – 0.85% 555 among the photoreceptors, with a rough average of approximately 0.5% cones over most of 556 the retina. For aardvark 2, total cone densities could only be determined in the two small 557 peripheral pieces of retina described above. They ranged from 1,100 to 1,600 cones/mm². 558 These densities are somewhat higher than those of aardvark 1. With the higher rod densities 559 of 182,000 – 245,000/mm², this again amounts to cone percentages of 0.5 – 0.9% of the 560 photoreceptors. 561 562 Thyroid hormone levels 563 Given the high incidence of cone opsin co-expression in the aardvark and the important role 564 of thyroid hormone (TH) in regulating cone spectral identity, we looked at the serum TH 565 levels available for aardvark 1 (five measurements within 2.5 years) and his adult son 566 aardvark 3 (one measurement) in comparison to published serum TH levels in some 567 representative mammals (Table 2). To our knowledge, these are the first published serum TH 568 levels for aardvarks. The serum values of total thyroxine (tT4), total triiodothyronine (tT3), 569 free T4 (fT4) and the biologically active form free T3 (fT3) are very similar for aardvark 1 570 and his son, and are also similar to or higher than the respective values in other mammals. 571 This suggests that aardvarks are not hypothyroid and that the S opsin dominance is not related 572 to a lack of TH (see Discussion). 573 574 575 Table 2. Thyroid hormone serum levels in different mammals 576 (Different units in the publications have been converted to unify the table) 577 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 27 Species / Animal tT4 (µg/dl) tT3 (ng/dl) fT4 (ng/dl) fT3 (pg/ml) TSH (ng/ml) Aardvark 11 12 – >15 226.3 – 390.4 1.54 – 1.93 2.28 – 4.23 15 304.2 2.25 3.32 nd Asian elephant2 11.12 126.72 0.87 1.39 0.97 African elephant2 10.76 123.27 0.93 1.41 0.56 Manatee3 4.5 – 8.3 140 – 160 1.3 – 1.6 nd nd Cow4 3.9 – 8.8 70 – 250 1.3 – 2.4 2.3 – 7.8 nd Dog5,6 1.5 – 1.6 (1.53 – 2.25) 76 – 84 (nd) ~1.75 (0.98 – 1.57) ~1.15 (nd) nd (0.07 – 0.26) Human7 5 – 12 80 – 220 0.7 – 1.9l 2.3 – 4.1 0.006 – 0.06 578 tT4 = total thyroxine T4, tT3 = total triiodothyronine T3, fT4 = free T4, fT3 = free T3; TSH = 579 Thyroid-stimulating hormone, thyrotropin; nd = not determined. 580 1Frankfurt Zoo animals (see Methods); 2Ref. [88]; 3Ref. [89]; 4Ref. [90]; 5Ref. [91]; 6Ref. [92]; 7UCLA 581 Health Website (https://www.uclahealth.org/medical-services/surgery/endocrine-surgery/conditions-582 treated/thyroid/normal-thyroid-hormone-levels), fT3 level from Cleveland Clinic 583 (https://my.clevelandclinic.org/health/diagnostics/22425-triiodothyronine-t3) 584 585 586

587 The aardvark’s designation as a ‘living fossil’ [2] suggests that its eye and retina also may 588 show prototypic, primordial mammalian features. On the other hand, the aardvark ancestor 589 could already have been specialized on feeding on ants and termites, which speaks against a 590 ‘basic general’ mammal. Hence, we were interested to study the aardvark’s eye and retina 591 when the opportunity arose. Early mammals are assumed to have gone through a ‘nocturnal 592 bottleneck’ with associated adaptations to low-light vision (see, e.g., [23-27]). Most extant 593 nocturnal mammals possess eye and retina features reflecting such an evolutionary adaptation: 594 The eye optics with a proportionately large lens and cornea serves increased light capture. The 595 retina has a dominance of the more light-sensitive rod photoreceptors and only a small 596 minority of cone photoreceptors (review: [5]). The features of the aardvark eye and retina fit 597 this ‘nocturnal’ category. 598 599 Until 2022, the only detailed study available on the aardvark eye and retina was by Victor 600 Franz, dating back to 1909 [3]. It claimed a pure rod retina without cones, but concedes that 601 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 28 the tissue conservation may not have been sufficient to prove the absence of cones. The 602 absence of cones has since been repeated in various printed summaries and websites on 603 aardvark biology and vision ([28] page 326, [29] page 579, [1,30,31]; website examples: 604 [32,33]). Franz [3] also claimed the absence of a tapetum lucidum, but conceded in a later 605 handbook chapter that a tapetum lucidum fibrosum may be present ([34] page 1214). 606 Recently, Paszta and colleagues published a study on general features of the aardvark eye and 607 extraocular structures that included a brief description of the retina [9]. This paper similarly 608 claimed an absence of cones, but reported a tapetum lucidum. Walls [23] also listed the 609 aardvark as having ‘eye shine’ and a trace of a tapetum fibrosum (his Table VII, p. 241). The 610 present study has assessed these claims. 611 612 The external dimensions of the eye of aardvark 1 are similar to those reported previously 613 [3,9]. The corneal size and lens are large compared to total eye size, the ratios of corneal 614 diameter to eye equatorial diameter (0.75), of corneal diameter to eye axial length (0.81), of 615 lens diameter to eye equatorial diameter (0.58), and of lens thickness to eye axial length 616 (0.45) are within the range seen across nocturnal mammals [25,35]. However, these 617 parameters overlap between nocturnal, cathemeral/crepuscular, and diurnal mammals, so their 618 diagnostic value is limited (for discussion see, e.g., [25], but also [36]). 619 620 Tapetum lucidum 621 Some nocturnal, crepuscular, and arrhythmic mammals have a reflective choroidal tapetum 622 lucidum behind the retina to increase the amount of light absorbed by the photoreceptors. To 623 observers the tapetal reflection appears as ‘eye shine.’ Morphologically, the two most 624 common tapetum types are a ‘tapetum cellulosum’ where the reflecting structures termed 625 rodlets are located intracellularly (found in Carnivora), and a ‘tapetum fibrosum’ where the 626 reflecting structures termed fibrils are located extracellularly (found in Artiodactyla, Cetacea 627 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 29 and Perissodactyla) (reviews: [37-39]). Our own observation (Fig 1) as well as images and a 628 video in the internet (https://www.youtube.com/watch?v=apNr2HSrx6s), last accessed on 629 October 17, 2024; aardvark shown from minute 2:11) indicate that the aardvark may have a 630 tapetum lucidum. Hence, it is surprising that Franz [3], Matas et al. [40] and Freeman et al. 631 [41] describe the absence of a tapetum in the aardvark eye. In contrast, Walls [23] and Paszta 632 et al. [9] describe the presence of a tapetum, which they consider to be a choroidal tapetum 633 fibrosum. 634 635 Our more detailed histological observations confirm a choroidal tapetum. The ultrastructural 636 appearance of the tapetal fibrils with their cross-striation (Fig 3C) indicates that the fibrils 637 consist of collagen, which is typical for a tapetum fibrosum [38,42]. On the other hand, in 638 cross-section, the aardvark tapetal fibrils show a substructure with concentric zones of 639 different electron-density (Fig 3D). This is reminiscent of the substructure of the zinc-640 containing rodlets of the tapetum cellulosum in ferret and dog [43,44]. As our tissue fixation 641 was not optimized for electron microscopy, the substructure of the aardvark tapetal fibrils 642 may be an artefact. Unfortunately, our material did not allow us to determine whether the 643 domains with their changing fibril orientations are contained intracellularly (suggesting a 644 tapetum cellulosum) or whether they are extracellular (suggesting a tapetum fibrosum). 645 Weighing all evidence, we think that the aardvark tapetum lucidum is of the fibrosum type. It 646 would certainly seem unlikely that the aardvark has a mix of both tapetum types. In mammals 647 with a tapetum lucidum (whether of the cellulosum or fibrosum type), the RPE cells in front 648 of the tapetum are unpigmented or only minimally pigmented, such that the tapetum can in 649 fact function as a reflector (reviews: [23] page 232; [38]). This is also the case in the aardvark. 650 651 Across mammals, the optically relevant properties of tapetum fibrosum fibrils and tapetum 652 cellulosum rodlets are similar. Both have diameters of about 100-200 nm and are nearly 653 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 30 hexagonally arranged with a spacing of about one fibril/rodlet diameter, suitable for 654 constructive interference (e.g., sheep: [42]; bovine: [45]; cat: [46,47]; overviews: [38,39,48]). 655 The aardvark fibrils have comparable dimensions, but their spacing is more disordered. In 656 many domains they are loosely spaced and do not show a tight hexagonal arrangement. Also, 657 the aardvark tapetum layer is relatively thin and the tapetum lamination is not as conspicuous 658 as in other species. Overall, the aardvark appears to have a rudimentary version of a tapetum 659 lucidum. Nevertheless, it obviously shows the typical ‘eye shine.’ 660 661 Avascular retina 662 The aardvark retina is avascular, as observed by Franz [3] and Matas et al. [40], and 663 confirmed here. This also explains why we did not see any GFAP-labelled astrocytes. Retinal 664 astrocytes are neuroglia cells restricted to the optic nerve fibre layer and associated with blood 665 vessels as well as with retinal ganglion cell axons. In species with limited retinal 666 vascularization, they only occur in the vascularized regions, and in avascular retinae they are 667 completely absent (review: [17]). Avascular retinae have been found in several mammals 668 from different orders [23,49-52]. There is no convincing common explanation for why the 669 retinae of some species are avascular whereas those of most other species are vascularized to 670 various extents. Damsgaard and Country [52] conclude from their large data survey that the 671 ancestral mammal had an avascular retina, and that retinal vascularization was dynamically 672 gained and lost throughout subsequent mammalian evolution depending on species-specific 673 visual needs for retinal processing capacity, neuron numbers and hence retinal thickness. 674 Mammalian avascular retinae are generally thinner than vascularized ones, commonly below 675 150 µm (reviews: [51,52]). The reason is that they have to be supplied with oxygen by 676 diffusion from the choroidal capillary network, and the maximum oxygen diffusion distance 677 has been modeled to be about 143 µm [53], although this value has been questioned by some 678 authors [54]. At 120-180 µm, the aardvark retina is slightly thicker than other avascular 679 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 31 retinae, but it is still thinner than many vascular nocturnal retinae, particularly having a 680 thinner ONL and lower photoreceptor density (see below). 681 682 Chase [51] also stated that mammals with avascular retinae lack a tapetum lucidum and 683 assumed that this is because a tapetum would add to the tissue thickness that has to be reached 684 by choroidal oxygen. The aardvark tapetum lucidum is not in line with that assumption. The 685 elephants, the horse and the zebra are further examples of species with a tapetum lucidum and 686 an avascular retina [49,50]. In fact, the choriocapillaris, the actual release site of oxygen and 687 nutrients to the retina, is located in front of the choroidal tapetum, directly facing the RPE and 688 retina. Figure 3E-H shows this for the aardvark and elephant. The aardvark choriocapillaris 689 forms a very dense capillary mesh (Fig 3I) that obviously suffices to adequately supply the 690 retina. We conclude that Chase’s assumption of an incompatibility of a (choroidal) tapetum 691 lucidum with an avascular retina is not tenable. 692 693 Rods 694 Typically for a nocturnal mammal, the vast majority of the aardvark photoreceptors are rods; 695 the proportion of only around 0.5% cones is low even among nocturnal species (review: [5]). 696 However, the estimated rod densities of 124,000 – 214,000/mm² for aardvark 1 and of 697 182,000 – 245,000/mm² for aardvark 2 are rather low in comparison to those of many other 698 nocturnal mammals, which may range from 200,000 rods/mm² to more than 700,000 699 rods/mm² [5]. These low rod densities and the correspondingly thinner ONL most likely are 700 correlated with the avascularity of the aardvark retina. The nocturnal Microchiroptera also 701 have avascular retinae (review: [51]), and rod densities in the greater horseshoe bat average 702 about 370,000 rods/mm², complemented by about 2.5% cones [55]. The avascular retina of 703 the nocturnal to crepuscular rabbit has between 300,000 rods/mm² in central and 130,00 704 rods/mm² in peripheral retina, complemented by about 4 – 6% cones [56]. Hence, even among 705 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 32 nocturnal mammals with avascular retinae, the aardvark rod density is in the lower range, 706 suggesting a low adaptive pressure for good nocturnal vision. 707 708 Also typical for nocturnal mammals is an inverted architecture of the rod nuclei [20]. Franz 709 [3] noted the central clustering of heterochromatin in the aardvark rods and described it as 710 ‘nuclei with two opposing chromatin bodies that looked like being in mitosis.’ We here give a 711 detailed description of these rod nuclei (Fig 4). As analyzed by Solovei and colleagues 712 [20,22], the central position of the inactive heterochromatin and peripheral position of the 713 active euchromatin are assumed to be strongly disadvantageous for nuclear functions, but they 714 have an optical advantage. The densely packed heterochromatin core is highly refractive and 715 shows the physical properties of a light-focusing lens. Hence, the rod nuclei in the ONL form 716 columns of microlenses that act as ‘light guides.’ This strongly reduces light scattering in the 717 ONL, which is particularly important for nocturnal mammals with their thicker ONL and the 718 need to capture a large proportion of the few photons available at night. Obviously, inverted 719 rod nuclei are an evolutionary adaptation to improve nocturnal vision. The rods of diurnal 720 mammals have a conventional nuclear architecture, because they do not have to maximize 721 photon capture. 722 723 The aardvark has semi-inverted rod nuclei. Most likely, these are less effective focusing 724 lenses than fully inverted rod nuclei. Both the relatively low rod density compared to other 725 nocturnal mammals and the semi-inverted rod nuclei support the assumption that for the 726 aardvark with its reliance on smell and hearing, there was no strong evolutionary pressure for 727 an optimally adapted nocturnal retina, and that its night vision sensitivity is lower than that of 728 many other nocturnal mammals. 729 730 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 33 In conventional nuclei, heterochromatin is tethered to the nuclear envelope by either lamin 731 A/C or the lamin B receptor (LBR); the inverted rod nuclei of nocturnal mammals are 732 characterized by the absence of both tethers [57]. The semi-inverted architecture of the 733 aardvark rod nuclei might be explained by expression of one of these proteins. We tested this 734 with antibodies to LBR and to lamin A/C. Whereas no aardvark retinal neurons showed LBR 735 immunoreactivity, all of them showed lamin A/C immunoreactivity. The lamin A/C antibody 736 clearly marked the rod’s nuclear periphery, although admittedly weaker compared to the 737 nuclei of the INL and GCL. This is similar to the weak LBR expression in the semi-inverted 738 rod nuclei of, e.g., goat and cow [57]. 739 740 Cones 741 From today’s perspective, the claim that the aardvark retina completely lacks cones [3,9] was 742 based on histological approaches that are not suitable to detect sparse populations of cones. 743 The most reliable approach to identify cones is by immunolabelling them with antibodies to 744 cone opsins, which we have done in the present study. The immunolabelling showed the 745 presence of L and S opsin in a low-density population of aardvark cones. 746 747 The two common mammalian cone opsins are the L opsin (also termed LWS opsin, peak 748 sensitivity λmax in the green to yellow part of the spectrum) and the S opsin type 1 (SWS1 749 opsin, λmax in the blue, violet or UV part of the spectrum [58]. The opsin antisera used here 750 recognize all spectral variants of the respective opsins. Partial gene sequencing has identified 751 L and S opsin genes in the aardvark, and judging from the tuning-relevant amino acids, the 752 aardvark S opsin has been suggested to be UV-sensitive, and the L opsin to have its λmax at 753 527-533 nm (Supplementary Table S5 in [7]). UV tuning is the ancestral tuning of the 754 vertebrate SWS1 opsin and has been maintained in a number of nocturnal mammals (reviews: 755 [59-61]). Our opsin immunolabelling gives evidence that the L and S opsins are indeed 756 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 34 expressed in aardvark cones, and provides information about the population density and 757 topographical distribution of the cones. 758 759 A less normal feature of the aardvark cones is, that many of them co-express the two opsins 760 across the retina and are ‘dual pigment’ cones. An artefactual double-labelling due to cross-761 reactivity between the two opsin antisera can be excluded, because the tissue also shows pure 762 L and S cones. Many nocturnal mammals have the two opsins in separate cone populations 763 and thus the basis for dichromatic color vision. This is the canonical mammalian condition of 764 being spectrally selective (reviews: [4,5,62]). Dual pigment cones were reported in the ventral 765 retinae of the house mouse, guinea pig and rabbit [63,64]. Since then, opsin co-expression in 766 all cones across the retina has been found in the pouched mouse Saccostomus campestris and 767 the Siberian dwarf hamster Phodopus sungorus [65]. The molecular mechanisms responsible 768 for the co-expression of both opsins in some cones and its suppression in other cones is only 769 partly understood. During rat and gerbil retinal development, all cones first express S opsin, 770 and prospective L cones then successively switch to L opsin expression with an intermediate 771 phase of opsin co-expression; this suggests that S opsin expression may be the default 772 pathway when an L opsin-activating mechanism is absent or suppressed [66]. One such factor 773 is thyroid hormone (see below). 774 775 In the retina of aardvark 1, the majority of single pigment cones are pure S cones (regionally 776 varying from 2% to 65% of the cones); pure L cones are a sparse population of about 0.3 – 777 7%. In the peripheral retina of aardvark 2, there are somewhat more pure L cones (2 – 21%) 778 than pure S cones (3 – 12%). If the aardvark has colour-processing (cone-opponent) retinal 779 ganglion cells, most of their input will be from dual pigment cones. What they may be able to 780 use for colour vision are the spectrally different sensitivities of pure L and S cones vs. dual 781 pigment cones. The rods may also contribute to colour vision in mesopic light conditions 782 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 35 ([58] and references therein). However, given the very low cone density and the dominance of 783 dual pigment cones, it is likely that the aardvark at best has feeble colour vision [62]. The low 784 cone density also makes good photopic visual acuity (cone-based spatial resolution) unlikely. 785 786 Our opsin immunolabelling showed qualitatively different mixtures of the two opsins in 787 different cones, but did not allow to quantitatively access the absolute amounts of the two 788 opsins per cone. Hence, we cannot comment on the presumed dominant spectral sensitivity of 789 these dual-pigment cones. However, the much higher fraction of S opsin-containing cones in 790 comparison to most other mammals suggests that the aardvark retina has a relatively high 791 cone-based sensitivity in the shortwave (blue to UV) range. This is the case, e.g., in 792 Microchiroptera with a similar S opsin dominance [67]. It may be an evolutionary adaptation 793 to the spectral composition of twilight, which contains higher proportions of short 794 wavelengths than full daylight (see, e.g., [68,69]). Twilight is the most likely situation that 795 nocturnal animals may encounter during their active phases, and cones contribute to vision at 796 this mesopic light level. On the other hand, many nocturnal mammals facing the same twilight 797 conditions, e.g., rats [70], flying foxes [71], colugos [14], nocturnal lemurs [72], and Canidae 798 [73], have low proportions of S cones (reviews: [4,5,60]). Moreover, a substantial number of 799 nocturnal mammals are completely lacking S cones (review: [74]). That makes the hypothesis 800 of a special shortwave adaptation to twilight less plausible. 801 802 As assessed in the retina of aardvark 1, cone density changes across the retina are small, 803 ranging from a shallow maximum of up to 1,300 cones/mm² in a region of temporal retina to 804 minima of 300 cones/mm² at some peripheral locations, but up to 1,100 cones/mm² in other 805 peripheral regions, representing 0.25–0.85% of the photoreceptors. The horizontal band of 806 reduced RPE pigmentation seen in the eye’s fundus has no higher cone density than the 807 surrounding midperiphery. Thus, the cone topography does not suggest specialized regions of 808 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 36 particularly high cone-based visual performance like a prominent area centralis or visual 809 streak. Such specialisations are present in many other mammals (reviews: [75-77]), but often 810 most prominent in the retinal ganglion cells. The cone density in the small parts of the retina 811 of aardvark 2 that could be studied is slightly higher, but due to the higher rod densities there, 812 the cones also represent only 0.5–0.9% of the photoreceptors. Together with the different 813 proportions of pure L cones, pure S cones and dual pigment cones in the two individuals, it 814 appears that there is some interindividual variability in the detailed characteristics of the cone 815 population, even though the basic properties of a low cone-to-rod ratio and a dominance of 816 dual pigment cones are preserved. Alternatively, these differences between the old male 817 aardvark 1 and the younger female aardvark 2 could be sex-related or age-related. A larger 818 sample of aardvark retinae would be needed to study these aspects. 819 820 A further unusual feature of the aardvark cones is that the cone pedicles do not show the 821 typical cluster of CtBP2-positive ribbons. In other mammals, the cone pedicles have a large 822 number of presynaptic sites with ribbons (reviews: [78,79]). The tissue quality did not allow 823 an ultrastructural analysis, so we could not determine whether the aardvark cones have only 824 few ribbon synapses per pedicle, or whether there are many that do not label for CtBP2. A 825 low ribbon density may impair the signal transmission performance of the aardvark cones. 826 827 Thyroid hormone (TH) 828 In mouse early postnatal development, TH, through its receptor TRβ2, is a crucial regulator of 829 cone spectral identity by repressing S opsin and activating L opsin [80-82]. Even in adult 830 mouse and rat, pharmacological suppression of serum TH reversibly activates S opsin and 831 represses L opsin in all cones [83]. It may be that dual pigment cones are the consequence of a 832 (genetic) defect of the switch-off mechanism for S opsin expression during developmental L 833 opsin activation [84]. Hence, we were interested to know whether the aardvark has unusually 834 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 37 low TH levels that might correlate with the high proportion of dual pigment cones. The data 835 presented in Table 2 show that this is not the case, the aardvark has serum TH levels that are 836 similar to or higher than those of other mammals, particularly those of the elephants and the 837 manatee that belong to the same clade Afrotheria as the aardvark. The other listed species 838 cover a broad range of mammals, and all of them have a normal cone complement with a 839 majority of pure L cones and a ca. 10% minority of pure S cones (with the exception of the 840 manatee, for which no cone population data exist). This suggests that aardvarks are not 841 hypothyroid and that the S opsin dominance is not related to a lack of TH. Also, elevated 842 serum levels of thyroid-stimulating hormone (TSH) would be a first sign of hypothyroidism, 843 but they appear low in aardvark 1 (Table 2). However, there could be other deficits in the 844 chain of thyroid hormone action that we were unable to study here. For example, one crucial 845 component is the nuclear T3 receptor TRβ2; in TRβ2 knockout mice, all cones express S 846 opsin and none express M opsin [80]. Another component is the monocarboxylate transporter 847 8 (MCT8), a plasma membrane transporter allowing TH access to the cones; in MCT8 848 knockout mice, cone opsin expression resembles that in hypothyroid or TRβ2 knockout mice 849 [85]. 850 851

852 The aardvark eye and retina have the typical features seen in nocturnal mammals: light-853 sensitive optics with a large lens and a large cornea, a reflective tapetum lucidum, and a rod 854 dominance with a very low cone density. Three unusual features are, (i) that the retina is 855 avascular with the corollary of being thinner and hence having a lower rod density than many 856 other nocturnal mammals, (ii) that the rod nuclei are only semi-inverted and not fully inverted 857 as in other nocturnal mammals, (iii) that there is opsin co-expression in a large proportion of 858 the cones. Hence, nocturnal visual sensitivity probably is lower than in many other nocturnal 859 mammals, and cone-based visual acuity and colour vision certainly are poor. 860 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 38 861 Regarding the designation of the aardvark as a living fossil [2] and the question whether this 862 entails a primordial, prototypical mammalian retina, one can state that the massive presence of 863 dual pigment cones argues against a primordial character of the aardvark retina. The most 864 common pattern in non-mammalian vertebrate retinae, as in mammalian retinae, are single 865 opsin cones. The most parsimonious assumption is that the last synapsid ancestor of the 866 mammals, and hence also the first mammal, had single opsin cones. Therefore, dual pigment 867 cones are a derived feature, and the aardvark retina is not a primordial mammalian retina as 868 far as the cones are concerned. 869 870 871 Acknowledgments 872 We thank Robert Molday and Jeremy Nathans for kindly providing antibodies. The technical 873 assistance of Alena Konoplew, Elke Laedtke and Carola Tröger is gratefully acknowledged. 874 We also thank Radosław Ratajszczak, Wojciech Paszta and Krzysztof Zagórski (Wrocław 875 Zoo) for their help in collecting research material and for providing information on the animal 876 from which the eyes were taken. 877 878 879

880 1. Shoshani J, Goldman CA, Thewissen JGM. Orycteropus afer. Mammalian Species. 881 1988; 300:1–8. 882 2. Shoshani J. Tubulidentata (Aardvarks). In: Encyclopedia of Life Sciences. John Wiley 883 & Sons Ltd, Chichester; 2001. http://www.els.net. doi: 10.1038/npg.els.0001578. 884 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 39 3. Franz V. Das Auge von Orycteropus afer (Pallas). Denkschriften der medicinisch-885 naturwissenschaftlichen Gesellschaft zu Jena. 1909; 15:401–415, with plates XXV and 886 XXVI. 887 4. Ahnelt PK, Kolb H. The mammalian photoreceptor mosaic: adaptative design. Progr 888 Ret Eye Res. 2000; 19:711–777. doi: 10.1016/s1350-9462(00)00012-4. 889 5. Peichl L. Diversity of mammalian photoreceptor properties: Adaptations to habitat and 890 lifestyle? Anatomical Record, Part A, Discoveries in Molecular, Cellular, and 891 Evolutionary Biology. 2005; 287A:1001–1012. doi: 10.1002/ar.a.20262. 892 6. Meredith RW, Gatesy J, Emerling CA, York VM, Springer MS. Rod monochromacy 893 and the coevolution of cetacean retinal opsins. PLoS Genet. 2013; 9(4):e1003432. doi: 894 10.1371/journal.pgen.1003432. 895 7. Emerling CA, Springer MS. Eyes underground: Regression of visual protein networks 896 in subterranean mammals. Mol Phylogenet Evol. 2014; 78:260–270. doi: 897 10.1016/j.ympev.2014.05.016. 898 8. Emerling CA, Springer MS. Genomic evidence for rod monochromacy in sloths and 899 armadillos suggests early subterranean history for Xenarthra. Proc R Soc Lond. B. 900 2015; 282:20142192. doi: 10.1098/rspb.2014.2192. 901 9. Paszta W, Klećkowska-Nawrot JE, Goździewska-Harłajczuk K. Morphological 902 evaluation of the orbit, eye tunics, eyelids, and orbital glands in young and adult 903 aardvarks Orycteropus afer, Pallas, 1766 (Tubulidentata: Orycteropodidae): 904 Similarities and differences with representatives of the Afrotheria clade. Anat Rec. 905 2022; 305:3317–3340. doi: 10.1002/ar.24905. 906 10. Briggman KL, Helmstaedter M, Denk W. Wiring specificity in the direction-907 selectivity circuit of the retina. Nature. 2011; 471:183–188. doi:10.1038/nature09818. 908 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 40 11. Tegethoff L, Briggman KL. Quantitative evaluation of embedding resins for volume 909 electron microscopy. Front Neurosci. 2024; 18:1286991. doi: 910 10.3389/fnins.2024.1286991. 911 12. Schiviz AN, Ruf T, Kuebber-Heiss A, Schubert C, Ahnelt PK. Retinal cone 912 topography of artiodactyl mammals: Influence of body height and habitat. J Comp 913 Neurol. 2008; 507(3):1336–1350. doi: 10.1002/cne.21626. 914 13. Schleich CE, Vielma A, Glösmann M, Palacios AG, Peichl L. Retinal photoreceptors 915 of two subterranean tuco-tuco species (Rodentia, Ctenomys): morphology, topography, 916 and spectral sensitivity. J Comp Neurol. 2010; 518:4001–4015. doi: 917 10.1002/cne.22440. 918 14. Moritz GL, Lim NT-L, Neitz M, Peichl L, Dominy NJ. Expression and evolution of 919 short wavelength sensitive opsins in colugos: A nocturnal lineage that informs debate 920 on primate origins. Evol Biol. 2013; 40:542–553. doi: 10.1007/s11692-013-9230-y. 921 15. Nießner C, Denzau S, Malkemper EP, Gross JC, Burda H, Winklhofer M, et al. 922 Cryptochrome 1 in retinal cone photoreceptors suggests a novel functional role in 923 mammals. Sci Rep. 2016; 6:21848. doi: 10.1038/srep21848. 924 16. Peichl L, Radic T, Solovei I, Wolfram M, Glösmann M. On the retinae of Glis and 925 Graphiurus: photoreceptor and ganglion cell populations, an absence of shortwave-926 sensitive cones, and some other features (Rodentia: Gliridae). Lynx, n. s. (Praha). 927 2022; 53:185–205. doi: 10.37520/lynx.2022.013. 928 17. Schnitzer J. Astrocytes in mammalian retina. Progr Ret Res. 1988; 7:209–231. 929 18. Lewis GP, Fisher SK. Up-regulation of glial fibrillary acidic protein in response to 930 retinal injury: Its potential role in glial remodeling and a comparison to vimentin 931 expression. Int Rev Cytol. 2003; 230:263–290. doi: 10.1016/s0074-7696(03)30005-1. 932 19. Bringmann A, Iandiev I, Pannicke T, Wurm A, Hollborn M, Wiedemann P, et al. 933 Cellular signaling and factors involved in Müller cell gliosis: neuroprotective and 934 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 41 detrimental effects. Prog Ret Eye Res. 2009; 28:423–451. doi: 935 10.1016/j.preteyeres.2009.07.001. 936 20. Solovei I, Kreysing M, Lanctôt C, Kösem S, Peichl L, Cremer T, et al. Nuclear 937 architecture of rod photoreceptor cells adapts to vision in mammalian evolution. Cell. 938 2009; 137:356–368. doi: 10.1016/j.cell.2009.01.052. 939 21. Subramanian K, Weigert M, Borsch O, Petzold H, Garcia-Ulloa A, Myers EW, et al. 940 Rod nuclear architecture determines contrast transmission of the retina and behavioral 941 sensitivity in mice. eLife. 2019; 8:e49542. doi: 10.7554/eLife.49542. 942 22. Feodorova Y, Falk M, Mirny LA, Solovei I. Viewing nuclear architecture through the 943 eyes of nocturnal mammals. Trends Cell Biol. 2020; 30:276–289. doi: 944 10.1016/j.tcb.2019.12.008. 945 23. Walls GL. The Vertebrate Eye and its Adaptive Radiation. New York: Hafner 946 Publishing Company; 1942. 947 24. Heesy CP, Hall MI. The nocturnal bottleneck and the evolution of mammalian vision. 948 Brain Behav Evol. 2010; 75:195–203. doi: 10.1159/000314278. 949 25. Hall MI, Kamilar JM, Kirk EC. Eye shape and the nocturnal bottleneck of mammals. 950 Proc R Soc Lond B. 2012; 279:4962–4968. doi: 10.1098/rspb.2012.2258. 951 26. Gerkema MP, Davies WIL, Foster RG, Menaker M, Hut RA. The nocturnal bottleneck 952 and the evolution of activity patterns in mammals. Proc R Soc Lond B. 2013; 953 280:20130508. doi: 10.1098/rspb.2013.0508. 954 27. Maor R, Dayan T, Ferguson-Gow H, Jones KE. Temporal niche expansion in 955 mammals from a nocturnal ancestor after dinosaur extinction. Nature Ecology & 956 Evolution. 2017; 1:1889–1895. doi: 10.1038/s41559-017-0366-5. 957 28. Kolmer W. Die Netzhaut (Retina). In: v. Möllendorff W, editor. Handbuch der 958 mikroskopischen Anatomie des Menschen, Band 3, Teil 2 Auge. Berlin: Verlag Julius 959 Springer; 1936. pp. 295–468. 960 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 42 29. Rochon-Duvigneaud A. Les Yeux et la Vision des Vertébrés. Paris: Masson et Cie.; 961 1943. 962 30. Rahm U. Erdferkel oder Röhrchenzähner. In: Altevogt R, et al., editors. Grzimeks 963 Tierleben – Enzyklopädie des Tierreichs, 12. Band Säugetiere 3. Zürich: Kindler 964 Verlag; 1972. pp. 474–478. 965 31. Rahm U. Röhrchenzähner. In: Altevogt R, et al., editors. Grzimeks Enzyklopädie 966 Säugetiere, Band 4. München: Kindler Verlag; 1987. pp. 452–458. 967 32. Ratzloff E. Orycteropus afer (Online). 2011. Available from: 968 https://animaldiversity.org/accounts/Orycteropus_afer (last accessed October 17, 969 2024). 970 33. Klappenbach L. Aardvark fast facts (Online). 2019. Available from: 971 https://www.thoughtco.com/aardvark-profile-129412 (last accessed October 23, 2024). 972 34. Franz V. Vergleichende Anatomie des Wirbeltierauges. In: Bolk L, Göppert E, Kallius 973 E, Lubosch W, editors. Handbuch der vergleichenden Anatomie der Wirbeltiere, Vol. 974 2, Part 2 Sinnesorgane. Berlin and Vienna: Urban & Schwarzenberg; 1934. pp. 989–975 1292. 976 35. Kirk EC. Eye morphology in cathemeral lemurids and other mammals. Folia Primatol. 977 (Basel) 2006; 77:27–49. doi: 10.1159/000089694. 978 36. Schmitz L, Motani R. Morphological differences between the eyeballs of nocturnal 979 and diurnal amniotes revisited from optical perspectives of visual environments. Vis 980 Res. 2010; 50:936–946. doi: 10.1016/j.visres.2010.03.009. 981 37. Schwab IR, Yuen CK, Buyukmihci NC, Blankenship TN, Fitzgerald PG. Evolution of 982 the tapetum. Trans Am Ophthalmol Soc. 2002; 100:187–200. 983 38. Ollivier FJ, Samuelson DA, Brooks DE, Lewis PA, Kallberg ME, Komáromy AM. 984 Comparative morphology of the tapetum lucidum (among selected species). Vet 985 Ophthalmol. 2004; 7:11–22. doi: 10.1111/J.1463-5224.2004.00318.X. 986 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 43 39. Zueva L, Zayas-Santiago A, Rojas L, Sanabria P, Alves J, Tsytsarev V, et al. 987 Multilayer subwavelength gratings or sandwiches with periodic structure shape light 988 reflection in the tapetum lucidum of taxonomically diverse vertebrate animals. J 989 Biophotonics. 2022; 15:e202200002. doi: 10.1002/jbio.202200002. 990 40. Matas M, Wise I, Masters NJ, Stewart J, Holloway A. Unilateral eyelid lesion and 991 ophthalmologic findings in an aardvark (Orycteropus afer): case report and literature 992 review. Vet Ophthalmol. 2010; 13 (suppl. 1):116–122. doi: 10.1111/j.1463-993 5224.2010.00803.x. 994 41. Freeman K, Ben-Shlomo G, McMullen R, Moore BA. Ophthalmology of Afrotheria: 995 Aardvarks, Hyraxes, Elephants, Manatees, and Relatives. In: F. Montiani-Ferreira F, 996 Moore BA, Ben-Shlomo G, editors. Wild and Exotic Animal Ophthalmology, Vol. 2 997 Mammals. Springer Nature Switzerland; 2022. pp. 49–70. 998 42. Bellairs R, Harkness ML, Harkness RD. The structure of the tapetum of the eye of the 999 sheep. Cell Tissue Res. 1975; 157:73–91. doi: 10.1007/BF00223231. 1000 43. Tjälve H, Frank A. Tapetum lucidum in the pigmented and albino ferret. Exp Eye Res. 1001 1984; 38:341–351. doi: 10.1016/0014-4835(84)90189-1. 1002 44. Wen GY, Sturman JA, Shek JW. A comparative study of the tapetum, retina and skull 1003 of the ferret, dog and cat. Lab Anim Sci. 1985; 35(3):200–210. PMID: 4021435. 1004 45. Braekevelt CR. Fine structure of the bovine tapetum fibrosum. Anat Histol Embryol. 1005 1986; 15:215–222. doi: 10.1111/j.1439-0264.1986.tb00713.x. 1006 46. Bernstein MH, Pease DC. Electron microscopy of the tapetum lucidum of the cat. J 1007 Biophys Biochem Cytol. 1959; 5:35–40. doi: 10.1083/jcb.5.1.35. 1008 47. Braekevelt CR. Fine structure of the feline tapetum. Anat Histol Embryol. 1990; 1009 19:97–105. doi: 10.1111/j.1439-0264.1990.tb00892.x. 1010 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 44 48. Bortolami R, Callegari E, Lucchi ML. Ultrastructural observations on the tapetum 1011 lucidum fibrosum. Anat Histol Embryol. 1979; 8:72–78. doi: 10.1111/j.1439-1012 0264.1979.tb00680.x. 1013 49. Johnson GL. Contributions to the comparative anatomy of the mammalian eye, chiefly 1014 based on ophthalmoscopic examination. Phil Trans R Soc Lond Series B. 1901; 1015 194:1–82. 1016 50. Johnson GL. Ophthalmoscopic studies on the eyes of mammals. Phil Trans R Soc 1017 Lond Series B. 1968; 254 (no. 794):207–220. 1018 51. Chase J. The evolution of retinal vascularization in mammals: a comparison of 1019 vascular and avascular retinae. Ophthalmology. 1982; 89:1518–1525. doi: 1020 10.1016/s0161-6420(82)34608-4. 1021 52. Damsgaard C, Country MW. The opto-respiratory compromise: balancing oxygen 1022 supply and light transmittance in the retina. Physiology. 2022; 37:101–113. doi: 1023 10.1152/physiol.00027.2021. 1024 53. Dollery CT, Bulpitt CJ, Kohner EM. Oxygen supply to the retina from the retinal and 1025 choroidal circulations at normal and increased arterial oxygen tensions. Invest 1026 Ophthalmol. 1969; 8:588–594. PMID: 5395349. 1027 54. Buttery RG, Hinrichsen CF, Weller WL, Haight JR. How thick should a retina be? A 1028 comparative study of mammalian species with and without intraretinal vasculature. 1029 Vision Res. 1991; 31:169–187. doi: 10.1016/0042-6989(91)90110-q. 1030 55. Kim T-J, Jeon Y-K, Lee J-Y, Lee E-S, Jeon C-J. The photoreceptor populations in the 1031 retina of the greater horseshoe bat Rhinolophus ferrumequinum. Mol Cells. 2008; 1032 26:373–379. doi: 10.1016/S1016-8478(23)14011-8. 1033 56. Young HM, Vaney DI. Rod-Signal interneurons in the rabbit retina: 1. rod bipolar 1034 cells. J Comp Neurol. 1991; 310:139–153. doi: 10.1002/cne.903100202. 1035 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 45 57. Solovei I, Wang AS, Thanisch K, Schmidt CS, Krebs S, Zwerger M, et al. LBR and 1036 lamin A/C sequentially tether peripheral heterochromatin and inversely regulate 1037 differentiation. Cell. 2013; 152:584–598. doi: 10.1016/j.cell.2013.01.009. 1038 58. Jacobs GH. Evolution of colour vision in mammals. Phil Trans R Soc Lond Series B. 1039 2009; 364:2957–2967. doi: 10.1098/rstb.2009.0039. 1040 59. Hunt DM, Carvalho LS, Cowing JA, Parry JW, Wilkie SE, Davies WL, et al. Spectral 1041 tuning of shortwave-sensitive visual pigments in vertebrates. Photochem Photobiol. 1042 2007; 83:303–310. doi: 10.1562/2006-06-27-IR-952. 1043 60. Hunt DM, Peichl L. S cones: Evolution, retinal distribution, development, and spectral 1044 sensitivity. Vis Neurosci. 2014; 31:115–138. doi:10.1017/S0952523813000242. 1045 61. Emerling CA, Huynh HT, Nguyen MA, Meredith RW, Springer MS. Spectral shifts of 1046 mammalian ultraviolet-sensitive pigments (short wavelength-sensitive opsin 1) are 1047 associated with eye length and photic niche evolution. Proc R Soc Lond B. 2015; 1048 282:20151817. doi: 10.1098/rspb.2015.1817. 1049 62. Jacobs GH. The distribution and nature of colour vision among the mammals. Biol 1050 Rev. 1993; 68:413–471. doi: 10.1111/j.1469-185x.1993.tb00738.x. 1051 63. Röhlich P, van Veen T, Szél Á. Two different visual pigments in one retinal cone cell. 1052 Neuron. 1994; 13:1159–1166. doi: 10.1016/0896-6273(94)90053-1. 1053 64. Applebury ML, Antoch MP, Baxter LC, Chun LL, Falk JD, Farhangfar F, et al. The 1054 murine cone photoreceptor: a single cone type expresses both S and M opsins with 1055 retinal spatial patterning. Neuron. 2000; 27:513–523. doi: 10.1016/s0896-1056 6273(00)00062-3. 1057 65. Lukáts Á, Szabó A, Röhlich P, Vigh B, Szél Á. Photopigment coexpression in 1058 mammals: comparative and developmental aspects. Histol Histopathol. 2005; 20:551–1059 574. doi: 10.14670/HH-20.551. 1060 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 46 66. Szél A, van Veen T, Röhlich P. Retinal cone differentiation. Nature. 1994; 370:336. 1061 doi: 10.1038/370336a0. 1062 67. Müller B, Glösmann M, Peichl L, Knop GC, Hagemann C, Ammermüller J. Bat eyes 1063 have ultraviolet-sensitive cone photoreceptors. PLoS ONE. 2009; 4(7):e6390). doi: 1064 10.1371/journal.pone.0006390. 1065 68. Roenneberg T, Merrow M. Light reception: discovering the clock-eye in mammals. 1066 Curr Biol. 2002; 12:R163–R165. doi: 10.1016/s0960-9822(02)00731-5. 1067 69. Walmsley L, Hanna L, Mouland J, Martial F, West A, Smedley AR, et al. Colour as a 1068 signal for entraining the mammalian circadian clock. PLoS Biol. 2015; 1069 13(4):e1002127. doi: 10.1371/journal.pbio.1002127. 1070 70. Szé1 A, Röhlich P. Two cone types of the rat retina detected by anti-visual pigment 1071 antibodies. Exp Eye Res. 1992; 55:47–52. doi: 10.1016/0014-4835(92)90090-f. 1072 71. Müller B, Goodman SM, Peichl L. Cone photoreceptor diversity in the retinas of fruit 1073 bats (Megachiroptera). Brain Behav Evol. 2007; 70:90–104. doi: 10.1159/000102971. 1074 72. Peichl L, Kaiser A, Rakotondraparany F, Dubielzig RR, Goodman SM, Kappeler PM. 1075 Diversity of photoreceptor arrangements in nocturnal, cathemeral and diurnal 1076 Malagasy lemurs. J Comp Neurol. 2019; 527(1):13–37. doi: 10.1002/cne.24167. 1077 73. Mowat FM, Peichl L. Ophthalmology of canidae: foxes, wolves, and relatives. In: 1078 Montiani-Ferreira F, Moore BA, Ben-Shlomo G, editors. Wild and Exotic Animal 1079 Ophthalmology, Vol. 2 Mammals. Springer Nature Switzerland; 2022. pp. 181–214. 1080 doi: 10.1007/978-3-030-81273-7_11. 1081 74. Jacobs GH. Losses of functional opsin genes, short-wavelength cone photopigments, 1082 and color vision—A significant trend in the evolution of mammalian vision. Vis 1083 Neurosci. 2013; 30:39–53. doi:10.1017/S0952523812000429. 1084 75. Hughes A. The topography of vision in mammals of contrasting life style: comparative 1085 optics and retinal organisation. In: Crescitelli F, editor. Handbook of Sensory 1086 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 47 Physiology, Vol. VII/5: The Visual System in Vertebrates. New York: Springer; 1977. 1087 pp. 613–756. 1088 76. Moore BA, Tyrrell LP, Kamilar JM, Collin SP, Dominy NJ, Hall MI, et al. Structure 1089 and function of regional specializations in the vertebrate retina. In: Kaas JH, editor. 1090 Evolution of Nervous Systems (2nd ed). Amsterdam: Elsevier; 2017. pp. 351–372. 1091 doi: 10.1016/B978-0-12-804042-3.00008-7. 1092 77. Heukamp AS, Warwick RA, Rivlin-Etzion M. Topographic variations in retinal 1093 encoding of visual space. Annu Rev Vis Sci. 2020; 6:237–259. doi: 10.1146/annurev-1094 vision-121219-081831. 1095 78. tom Dieck S, Brandstätter JH. Ribbon synapses of the retina. Cell Tissue Res. 2006; 1096 326:339-346. doi: 10.1007/s00441-006-0234-0. 1097 79. Regus-Leidig H, Brandstätter JH. Structure and function of a complex sensory 1098 synapse. Acta Physiol. 2012; 204:479–486. doi: 10.1111/j.1748-1716.2011.02355.x. 1099 80. Ng L, Hurley JB, Dierks B, Srinivas M, Saltó C, Vennström B, et al. A thyroid 1100 hormone receptor that is required for the development of green cone photoreceptors. 1101 Nat Genet. 2001; 27:94–98. doi: 10.1038/83829. 1102 81. Roberts MR, Srinivas M, Forrest D, Morreale de Escobar G, Reh TA. Making the 1103 gradient: thyroid hormone regulates cone opsin expression in the developing mouse 1104 retina. Proc Natl Acad Sci U S A. 2006; 103:6218–6223. doi: 1105 10.1073/pnas.0509981103. 1106 82. Applebury ML, Farhangfar F, Glösmann M, Hashimoto K, Kage K, Robbins JT, et al. 1107 Transient expression of thyroid hormone nuclear receptor TRbeta2 sets S opsin 1108 patterning during cone photoreceptor genesis. Devel Dynamics. 2007; 236:1203–1212. 1109 doi: 10.1002/dvdy.21155. 1110 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 48 83. Glaschke A, Weiland J, Del Turco D, Steiner M, Peichl L, Glösmann M. Thyroid 1111 hormone controls cone opsin expression in the retina of adult rodents. J Neurosci. 1112 2011; 31:4844–4851. doi: 10.1523/JNEUROSCI.6181-10.2011. 1113 84. Neitz M, Neitz J. The uncommon retina of the common house mouse. Trends in 1114 Neurosciences. 2001; 24(5):248-249. doi: 10.1016/s0166-2236(00)01773-2. 1115 85. Liu Y, Ng L, Liu H, Heuer H, Forrest D. Cone photoreceptor differentiation regulated 1116 by thyroid hormone transporter MCT8 in the retinal pigment epithelium. Proc Natl 1117 Acad Sci U S A. 2024; 121:e2402560121. doi: 10.1073/pnas.2402560121. 1118 86. Hicks D, Molday RS. Differential immunogold-dextran labeling of bovine and frog 1119 rod and cone cells using monoclonal antibodies against bovine rhodopsin. Exp Eye 1120 Res. 1986; 42:55–71. doi: 10.1016/0014-4835(86)90017-5. 1121 87. Wang Y, Macke JP, Merbs SL, Zack DJ, Klaunberg B, Bennett J, et al. A locus 1122 control region adjacent to the human red and green visual pigment genes. Neuron. 1123 1992; 9:429–440. doi: 10.1016/0896-6273(92)90181-c. PMID: 1524826. 1124 88. Brown JL, Walker SL, Moeller T. Comparative endocrinology of cycling and non-1125 cycling Asian (Elephas maximus) and African (Loxodonta africana) elephants. Gen 1126 Comp Endocr. 2004; 136:360–370. doi:10.1016/j.ygcen.2004.01.013. 1127 89. Ortiz RM, MacKenzie DS, Worthy GAJ. Thyroid hormone concentrations in captive 1128 and free-ranging West Indian manatees (Trichechus manatus). J Exp Biol. 2000; 1129 203:3631–3637. doi: 10.1242/jeb.203.23.3631. 1130 90. Nixon DA, Akasha MA, Anderson RR. Free and total thyroid hormones in serum of 1131 Holstein cows. J Dairy Sci. 1988; 71:1152–1160. doi: 10.3168/jds.S0022-1132 0302(88)79669-1. 1133 91. Wahrendorf S. Schilddrüsenparameter und Cholesterol-Werte bei 1134 verhaltensunauffälligen Hunden. PhD Thesis, Faculty of Veterinary Medicine, Ludwig 1135 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 49 Maximilians University Munich; 2011. doi: 10.5282/edoc.12999. Available from: 1136 https://edoc.ub.uni-muenchen.de/12999/1/Wahrendorf_Silke.pdf 1137 92. Hegstad-Davies RL, Torres SMF, Sharkey LC, Gresch SC, Muñoz-Zanzi CA, Davies 1138 PR. Breed-specific reference intervals for assessing thyroid function in seven dog 1139 breeds. J Vet Diagn Invest. 2015; 27:716–727. doi: 10.1177/1040638715606953. 1140 1141 1142 1143 Supporting information 1144 S1 Fig. Cone photoreceptors. Sequentially double immunolabelled cones in two 1145 neighbouring flat-mounted pieces from an unknown location in midperipheral to peripheral 1146 retina (aardvark 2). The piece of field 1 was first incubated with the rabbit L opsin antiserum 1147 JH492, visualized with an Alexa488-coupled secondary antiserum (green). Then it was 1148 incubated with the rabbit S opsin antiserum JH455, visualized with a Cy3-coupled secondary 1149 antiserum (magenta). The merge shows that all cones are labelled by Cy3, because this 1150 secondary antiserum bound to JH455 as well as JH492. The pure S cones are exclusively 1151 labelled by Cy3 (arrowheads), all Alexa488-labelled cones contain the L opsin. The 1152 neighbouring piece of field 2 was first incubated with the rabbit S opsin antiserum JH455, 1153 visualized with the Cy3-coupled secondary antiserum (magenta). Then it was incubated with 1154 the rabbit L opsin antiserum JH492, visualized with the Alexa488-coupled secondary 1155 antiserum (green). The merge shows that all cones are labelled by Alexa488, because this 1156 secondary antiserum bound to JH492 as well as JH455. The pure L cones are exclusively 1157 labelled by Alexa488 (arrowheads), all Cy3-labelled cones contain the S opsin. The relative 1158 amount of L and S opsin (i.e., labelling intensity) differs between cones. For details of the 1159 procedure see Methods. The images are maximum intensity projections of confocal image 1160 stacks. Scale bar is 50 µm and applies to all images. 1161 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint 50 1162 S2 Fig. Cone photoreceptors and their synaptic pedicles in the outer plexiform layer. 1163 Flat-mounted piece of retina (aardvark 1), triple immunolabelled for S opsin, L opsin and 1164 PNA. (A) Focus on the outer segments of the opsin-immunolabelled cones; most cones 1165 express both opsins (A1, S opsin; A2, L opsin; A3, merge). The field contains two pure L 1166 cones. (B) Focus on the cone pedicles in the outer plexiform layer; the S opsin-labelled 1167 pedicles (B1) are also labelled by PNA (B2), as shown in the merge (B3). (C) Schematic 1168 illustration of the cones identified in A (S cones, magenta circles; L cones, green circles; dual 1169 pigment cones, bipartite circles) and the PNA-labelled pedicles identified in B (grey squares). 1170 The pairing between outer segments and pedicles (connecting lines) was checked by 1171 following the S opsin-labelled cone axons through the image stacks. The pedicles of the two 1172 pure L cones show no PNA label. For two pedicles at the edges of the image, the 1173 corresponding outer segments lie outside the frame. Due to faint labelling of some cones, not 1174 all cones shown in (C) can be seen in (A) and (B). Scale bar in (C) is 50 µm and applies to all 1175 images. 1176 1177 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 13, 2024. ; https://doi.org/10.1101/2024.11.09.622767doi: bioRxiv preprint