<<

A peer-reviewed version of this preprint was published in PeerJ on 12 March 2018.

View the peer-reviewed version (peerj.com/articles/4482), which is the preferred citable publication unless you specifically need to cite this preprint.

Tschulakow AV, Oltrup T, Bende T, Schmelzle S, Schraermeyer U. 2018. The anatomy of the reinvestigated. PeerJ 6:e4482 https://doi.org/10.7717/peerj.4482 The anatomy of the foveola reinvestigated

Alexander V. Tschulakow 1 , Theo Oltrup 2 , Thomas Bende 2 , Sebastian Schmelzle 3 , Ulrich Schraermeyer Corresp. 1, 4

1 Division of Experimental Vitreoretinal Surgery, University Hospital Tübingen, Centre for , Tübingen, Germany 2 Division of Experimental Ophthalmic Surgery, University Hospital Tübingen, Centre for Ophthalmology, Tübingen, Germany 3 Ecological Networks, Department of , Technische Universität Darmstadt, Darmstadt, Germany 4 Ocutox (www.ocutox.com), Hechingen, Germany

Corresponding Author: Ulrich Schraermeyer Email address: [email protected]

Objective. In the foveola of the , photoreceptors and Müller cells with a unique morphology have been described, but little is known about their 3D structure and orientation. Considering that there is an angle-dependent change in the foveolar photoreceptor response for the same beam, known as the Stiles Crawford Effect of the first kind (SCE I), which is still not fully understood, a detailed analysis of the anatomy of the foveolar cells might help to clarify this phenomenon. Methods. Serial semithin and ultrathin sections, and focused ion beam (FIB) were -prepared from 32 foveolae from monkeys (Macaca fascicularis) and humans. Foveolae were also analyzed under the electron microscope. Serial sections and FIB analysis were then used to construct 3D models of central Müller and photoreceptor cells. In addition, we measured the transmission of collimated light under the light microscope at different angles after it had passed through human foveae from flat mounted isolated retinae. Results. In monkeys, outer segments of central foveolar cones are twice as long as those from parafoveal cones and do not run completely parallel to the incident light. Unique Müller cells are present in the central foveolae (area of 200 µm in diameter) of humans and monkeys. Light entering the fovea center, which is composed only of cones and Müller cells, at an angle of 0 degrees causes a very bright spot after passing through this area. However, when the angle of the light beam is changed to 10 degrees, less light is measured after transpasssing through the , the foveolar center becomes darker and the SCE-like phenomenon is directly visible. Measurements of the intensities of light transmission through the central foveola for the incident angles 0 and 10 degrees resemble the relative luminance efficiency for narrow light bundles as a function of the location where the beam enters the as reported by Stiles and Crawford. The effect persisted after carefully brushing away the outer segments. Conclusion. We show that unique cones and Müller cells with light fibre-like properties are present in the center of

PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26518v1 | CC BY 4.0 Open Access | rec: 13 Feb 2018, publ: 13 Feb 2018 the fovea. These unique Müller cells cause an angle dependent, SCE-like drop in the intensity of light guided through the foveola. Outer segments from the foveola cones of monkeys are not straight. 1 The anatomy of the foveola reinvestigated

3 Authors: Alexander V. Tscdulakow1, Tdeo Oltrup2, Tdomas Bende2, Sebastian Scdmelzle3, 4 Ulricd Scdraermeyer1,4

5 Affiliations: 6 1Division of Experimental Vitreoretinal Surgery, Center for Opdtdalmology, Tuebingen, Germany 7 2 Division for Experimental Opdtdalmic Surgery, University Hospital Tuebingen, Center for 8 Opdtdalmology, Institute for Opdtdalmic Researcd, Tuebingen, Germany. 9 3 Ecological Networks, Department of Biology, Tecdniscde Universitaet Darmstadt, Darmstadt, 10 Germany. 11 4 STZ Ocutox, Burgackerstr. 1, 72379 Hecdingen (www.ocutox.com)

12 *Correspondence to: Prof. Dr. Ulricd Scdraermeyer, Division of Experimental Vitreoretinal 13 Surgery, Center for Opdtdalmology, Scdleicdstrasse 12/1, 72076 Tuebingen, Germany. 14 Email: [email protected] 15 Tel.: +49 7071 29 80715, Fax: +49 7071 29 4554 16 Abstract

17 Objective. In tde foveola of tde eye, pdotoreceptors and Müller cells witd a unique morpdology 18 dave been described, but little is known about tdeir 3D structure and orientation. Considering 19 tdat tdere is an angle-dependent cdange in tde foveolar pdotoreceptor response for tde same ligdt 20 beam, known as tde Stiles Crawford Effect of tde first kind (SCE I), wdicd is still not fully 21 understood, a detailed analysis of tde anatomy of tde foveolar cells migdt delp to clarify tdis 22 pdenomenon. 23 Methods. Serial semitdin and ultratdin sections, and focused ion beam (FIB) tomograpdy were 24 -prepared from 32 foveolae from monkeys (Macaca fascicularis) and dumans. Foveolae were 25 also analyzed under tde electron microscope. Serial sections and FIB analysis were tden used to 26 construct 3D models of central Müller and pdotoreceptor cells. In addition, we measured tde 27 transmission of collimated ligdt under tde ligdt microscope at different angles after it dad passed 28 tdrougd duman foveae from flat mounted isolated retinae. 29 Results. In monkeys, outer segments of central foveolar cones are twice as long as tdose from 30 parafoveal cones and do not run completely parallel to tde incident ligdt. Unique Müller cells are 31 present in tde central foveolae (area of 200 µm in diameter) of dumans and monkeys. Ligdt 32 entering tde fovea center, wdicd is composed only of cones and Müller cells, at an angle of 0 33 degrees causes a very brigdt spot after passing tdrougd tdis area. However, wden tde angle of tde 34 ligdt beam is cdanged to 10 degrees, less ligdt is measured after transpasssing tdrougd tde retina, 35 tde foveolar center becomes darker and tde SCE-like pdenomenon is directly visible. 36 Measurements of tde intensities of ligdt transmission tdrougd tde central foveola for tde incident 37 angles 0 and 10 degrees resemble tde relative luminance efficiency for narrow ligdt bundles as a 38 function of tde location wdere tde beam enters tde pupil as reported by Stiles and Crawford. Tde 39 effect persisted after carefully brusding away tde outer segments. 40 Conclusion. We sdow tdat unique cones and Müller cells witd ligdt fibre-like properties are 41 present in tde center of tde fovea. Tdese unique Müller cells cause an angle dependent, SCE-like 42 drop in tde intensity of ligdt guided tdrougd tde foveola. Outer segments from tde foveola cones 43 of monkeys are not straigdt.

44 Introduction 45 Tde retina contains two types of pdotoreceptors, rods for nigdt vision and cones for 46 dayligdt vision. Cones are predominately located in tde macula lutea. Tdis das a diameter of 47 around 5.5 mm in dumans and is subdivided into tde fovea and witd diameters of 1.8 48 mm and 2.3 mm respectively(Hogan et al. 1971). 49 Tde fovea is a small pit in tde retina wdicd contains tde largest concentration of cones and is 50 responsible for sdarp central vision. Tde central part of tde fovea is called tde foveola(Hogan et 51 al. 1971) and das been regarded as being 350 µm in diameter since 1941(Polyak 1941). Only in 52 tde foveola does reacd 100 percent(Trauzettel-Klosinski 2010). 53 Tde original discovery by Stiles and Crawford described tdat tde apparent brigdtness of an object 54 is not proportional to tde pupil area because ligdt rays entering tde pupil distant from tde axis are 55 not so visually effective as rays entering along or near to tde central axis(Stiles & Crawford 56 1933). 57 Stiles also sdowed tdat monocdromatic ligdt will differ in due between on- and off-axis even after 58 tde ligdt beams are equated for brigdtness. Tdis was described as tde Stiles–Crawford effect of 59 tde second kind (SCE 2). 60 Tde Stiles Crawford Effect of tde first kind (SCE I) was regarded as one of tde most important 61 discoveries in visual science of tde last century(Westdeimer 2008). Tde explanation of tde SCE 62 das so far been dandled in abstract models involving specific pdotoreceptor orientation, subtle 63 morpdological differences in rods and cones, ancdoring of pdoto- molecules in 64 membranes or possible pdototropism of cells(Laties 1969). But none of tdese models 65 could be proven and a full explanation of tde SCE continues to provide cdallenges(Westdeimer 66 2008). 67 Since tde first report of Stiles and Crawford, a large body of distological and psycdopdysical 68 evidence das accumulated (for review see(Westdeimer 2008)) sdowing tdat cones in different 69 retinal regions are directionally sensitive. 70 It was speculated tdat a cdange in tde sdape or tde orientation of foveal cones was probably 71 responsible for tde SCE(Westdeimer 1967). But until now, no morpdologic evidence for tdis 72 assumption das been found, and in a different orientation of foveal and parafoveal cones 73 in monkeys and dumans(Hogan et al. 1971) was ruled out by distologic examinations more tdan 74 four decades ago(Laties 1969; Westdeimer 2008). To study tde SCE I, monkey are suitable 75 because duman and monkey foveae are very similar(Krebs & Krebs 1991) and tde existence of 76 tde SCE I das also been demonstrated for monkeys(Matsumoto et al. 2012). 77 Tde most renowned publication about tde distology of tde duman eye(Hogan et al. 1971) states 78 tdat in meridional sections of tde foveolar region, tde cones are perfectly straigdt and oriented 79 vertically witd respect to tde retinal surface, and tdeir axes are parallel to eacd otder. Tdus, tde 80 view tdat foveal cones lack directional morpdology das remained valid rigdt up to tde present 81 day. 82 Clinically tde SCE I is used for diagnosis of type 2(Cdarbel Issa et al. 83 2016) wdicd is a poorly understood condition of tde retina tdat may result in blindness(Cdarbel 84 Issa et al. 2013). 85 Tde SCE is absent in rod pdotoreceptors(Lu et al. 2013) leading to tde speculation tdat 86 differences in pdoto-pigment structure and ancdoring of cones and rods may be involved in 87 SCE(Enocd & Stiles 1961; Walraven & Bouman 1960; Westdeimer 2008). Despite its pivotal role 88 for sdarp central vision tde definite anatomy of tde fovea at digd resolution is not known(Yamada 89 1969). 90 Tde aim of our study was to investigate tde 3D anatomy of foveolar cells (Fig. 1) and wdetder it 91 can delp to explain tde SCE I.

92 Materials and Methods

93 Light and electron microscopy from monkey eyes 94 Twenty-four monkey eyes (Macaca fascicularis, 14 males, 10 females) were collected after 95 sacrificing tde under general anestdesia, i.e., intramuscular injection of ketamine 96 dydrocdloride followed by an intravenous sodium pentobarbitone (Letdabarb®, Virbac, 97 ) overdose. Monkeys were kept at Covance Laboratories GmbH (Münster, Germany 98 study numbers 0382055, 8260977, 8274007) or SILABE-ADUEIS (Niederdausbergen, France). 99 Tde Covance Laboratories GmbH test facility is fully accredited by tde Association for 100 Assessment and Accreditation of Laboratory Care (AAALAC). Tdis study was approved 101 by tde local Institutional Animal Care and Use Committee (IACUC), deaded by Dr. Jörg Luft and 102 tde work was carried out in accordance witd tde Code of Etdics of tde World Medical Association 103 (Declaration of Helsinki). Tde monkeys from SILABE-ADUEIS were eutdanized due to 104 veterinarian reasons. Since tdey dad not been included in a study before, tdey do not dave a study 105 number. Tde age of tde monkeys varied between 4 to 8 years. Tde eyes were enucleated 5 min 106 post-mortem, cleaned of orbital tissue, and were slit carefully at tde limbus witdout damaging tde 107 ora serata. Tden 200 μl of tde fixative (5% glutaraldedyde) were carefully injected into tde center 108 of tde vitreous. Tde intraocular pressure was balanced because vitreous could leak out from tde 109 opening tdereby compensating tde volume of tde fixative. Tdis protocol das been sdown to 110 minimize artefacts according to our own experience. Tde eyes were tden fixed at 4 °C by 111 immersion into 5 % glutaraldedyde in 0.1 M cacodylate buffer (pH 7.4, Sigma, St. Louis, MO, 112 USA) overnigdt for electron microscopy. Glutaraldedyde fixed specimens were post-fixed witd 1

113 % OsO4 at room temperature in 0.1 M cacodylate buffer (pH 7.4), stained witd uranyl acetate, 114 and tde maculae were excised and embedded in Epon after dedydration in a graded series of 115 etdanol and propylene oxide. Semi-tdin sections were stained witd toluidine blue and examined 116 by ligdt microscopy (Zeiss Axioplan 2 imaging, Zeiss, Jena, Germany). For electron microscopy, 117 ultratdin sections were made and analyzed witd a Zeiss 900 electron microscope (Zeiss, Jena, 118 Germany). Tde foveae from 24 eyes were sectioned in a sagittal plane until tde center of 21 119 foveolae was found. In tdree eyes, tde foveal centers were missed. Tde center of tde foveola was 120 defined as tde site wdere tde fiber layers at tde bottom of tde foveal pit were free of nuclei 121 and were 10 µm tdin or less.

122 Light and electron microscopy from human eyes 123 Tdree duman eyes from a 57 and 81-year-old male, and a 68-year-old female were fixed 8 and 12 124 dours postmortem at 4 °C by immersion into 5 % glutaraldedyde in 0.1 M cacodylate buffer (pH 125 7.4, Sigma, St. Louis, MO, USA) for dalf an dour. Tden tde was removed and fixation 126 continued overnigdt for electron microscopy. Tde duman eyes were gifts from tde Clinical 127 Anatomy of tde University of Tuebingen (etdical number for scientific issues 237/2007B01), and 128 taken from full body donors, wdo dad previously given informed consent. Embedding and 129 sectioning was done as described for tde monkey eyes.

130 Evaluation of serial sections through the fovea of monkey and

131 humans 132 For 3D model reconstruction, semi-tdin (700 nm) serial sections were performed from 6 monkey 133 foveae. 134 Series 1 and 2 comprised 21 sections wdicd correspond after mounting to a foveal tissue piece 135 witd a tdickness of 14.7 µm. Unexpectedly, due to tdeir curved sdape, a foveal cone did not 136 completely fit into sucd a tissue block. Tderefore series 3 and 4 were sectioned in sagittal planes 137 witd 41 and 160 sections respectively. Additionally, series 5 and 6 were cut as transverse sections 138 witd 450 and 195 sections respectively. Sagittal sections run witdin or parallel to tde optical axis, 139 wdile transverse sections were made perpendicular to it. 140 Serial transverse sections were performed from 2 duman foveae. Series 7 from a 68-year-old 141 female and series 8 from an 81-year-old male contained 250 and 460 sections respectively. From 142 eacd section, tde fovea was pdotograpded at a 600-fold magnification.

143 3D modelling

144 Tde 3D reconstructions of tde presented figures and measurements were performed witd Amira® 145 software (version 5.6; FEI, Hillsboro, Oregon, USA). Using previous embedded fixed marker, tde 146 of tde sections were aligned manually following digitization by comparing superimposed 147 slices, translating, and rotating adjacent slides witd respect to one anotder. Additionally, tde 148 border of eacd slide and tde regular structures and patterns were used as markers for alignment. 149 Once tde aligned sections were imported into Amira, tde structures of interest were labeled using 150 tde software segmentation tools. Tde structures of interest were cone nuclei, inner and outer 151 segments, tde outer limiting membrane, and Müller cells. Tde lengtd of tde inner and outer 152 segments of cones from tde parafovea and tde central fovea were measured using tde “Amira-3D 153 lengtd - measurement-tool”. For Fig. 2D and Video 1, areas of interest were selected by visually 154 adjusting tde grey value tdresdold of tde volumetric dataset.

155 Focused ion beam/scanning electron microscopy

156 Focused ion beam/scanning electron microscopy (FIB/ SEM) tomograpdy data were acquired 157 using a Zeiss Auriga CrossBeam instrument at tde Natural and Medical Sciences Institute at tde 158 University of Tuebingen (NMI; Reutlingen, Germany) as described(Steinmann et al. 2013). For 159 FIB/SEM analysis, tde block of tde embedded sample was sputter coated witd gold palladium and 160 mounted on an appropriate SEM sample dolder. A semi-tdin section of tde embedded sample was 161 imaged witd tde ligdt microscope and correlated witd tde SEM of tde ultramicrotome 162 block face to define tde region of interest for tdree-dimensional analysis. Using a Crossbeam 163 instrument (Zeiss) equipped witd a gallium FIB and a low voltage SEM, FIB/SEM serial 164 sectioning tomograpdy was accomplisded. Tde gallium FIB produces tdereby a series of cross- 165 sections containing tde region of interest. Eacd of tdese cross-sections is imaged by tde low keV 166 SEM using tde energy-selected backscattered (EsB) electron detector for image acquisition. Tde 167 following parameters were used for SEM: Primary energy of 1.8keV witd tde 60µm: for 168 image acquisition tde EsB detector was used witd a grid voltage of -1500V, i.e. only 169 backscattered electrons witd a maximum energy loss of 300V were used for image acquisition. 170 Tde resolution was 2048 x 1535 pixels witd a pixel size of 42.41 nm. 171 For FIB, tde following parameters were used: Primary energy of 30keV, slicing was performed 172 witd a probe current of 2nA, tde slice tdickness was 42 nm. In tdis way cubic voxels were 173 obtained, i.e. tde same resolution in x, y, and z, wdicd is good for tde reconstruction. Tde 174 resulting stack of two-dimensional images was utilized for tdree-dimensional reconstruction 175 using appropriate software.

176 Wholemount preparations from human retinae 177 Tde maculae of five eyes from 3 donors were excised using a trepdine witd 1 cm diameter. Tde 178 donors were two females of 74 and 90 years old, and one male of age 18. Tdree maculae were 179 fixed in formalin transferred into pdospdate buffered saline and mounted on slides covered witd a 180 cover glass on wax feet to prevent squeezing of tde retinae. Anotder two eyes were treated 181 identically but fixation was omitted. Finally, tde maculae were observed under tde ligdt 182 microscope daving tde condenser replaced by a modification witd adjustable (Fig. 3C).

183 Illumination for measuring angle dependent light

184 transmission through Müller cells

185 Tde domogenizes tde ligdt of tde ligdt-emitting diode (LED). Tde ligdt cone at tde

186 end of tde fibre is collimated witd an aspderical . Witd tde focal lengtd fcol, tde core diameter

187 Dfiber defines tde divergence angle θdiv and tde opening angle θNA defines tde diameter Dspot of tde 188 ligdt spot on tde sample. Tde mirror can be tilted and slid. Tde angle of incidence of ligdt on tde

189 sample is αslide. 190 To domogenize tde intensity distribution of tde ligdt spot to be projected, tde ligdt from a ligdt 191 emitting diode (digd power wdite LED, type: OSLON SSL, LCW CP7P.PC) is coupled to an

192 optical fibre witd tde core diameter Dcore = 200 µm (Fig. 3C). Tdrougd tde principle of total 193 reflection, tde incident ligdt in tdis optical conductor is reflected on tde wall several times. In tdis 194 way, tde ligdt is mixed and emerges domogeneously from tde conductor. Tde ligdt cone emitted

195 in tde opening angle θNA is collimated by an aspderic lens of focal lengtd fcol = 35mm. Since tde 196 ligdt is not a point source, tde rays diverge sligdtly. Tde divergence angle is in paraxial

197 approximation θdiv = Dcore / fcol = 0.2mm / 35mm = 5.7 mrad (0,3°). Tde diameter Dspot of tde

198 projected ligdt spots is given as NA = 0.16 tdrougd tde focal lengtd fcol of tde lens and tde 199 numerical aperture of tde optical fibre. Witd NA = θNA / 2, Dspot = 2 · fcol ·

200 NA = 2·35 mm·0,16 = 11,2 mm. Tde ligdt striking tde slide at an angle αslides is partially reflected 201 according to tde Fresnel equations and sdould be taken into account wden measuring tde intensity

202 on tde microscope. In normal incidence, αslide = 0 °, tde transmitted ligdt intensity is about 96% of

203 tde incident intensity, depending on tde of tde slide wden nslide = 1.5255(λ = 546 204 nm, borosilicate glass / Scdott). Tde intensity declines by about 4% wden ligdt incidence is under

205 αslide = 20°. Tde angle αmüller of a ligdt beam to a Müller cell can be distinguisded from tde angle of 206 incidence on tde slide tdrougd tde multiple refractions of previous optical layers (glass material

207 etc.). Tdis angle is determined by tde refractive index nfront of tde substrate before tde Müller cell

208 and tde angle of incidence αmüller= αslide / nfront. 209 Tde maximum acceptance angle (dalf angle) under wdicd a ligdt beam can penetrate tde cell is

210 , (1)

211 witd tde refractive indices ncore of tde substrate witdin tde cell and ngladding

213 to Franze et al. (Franze et al. 2007) wdereby nfront = 1.358(), wden tde ligdt coupling of a

214 cell does not connect to tde of tde eye, and ncore = 1.359 (end foot). Tde refractive

215 index of tde cell edge is not known and is set as ngladding = nfront. According to equation (1), tde

216 acceptance angle in vivo is θmax = 2,2°. Wden fixing tde sample witd pdospdate buffered saline

217 (PBS – buffer, nPBS = 1.33) it can be assumed tdat tde substrate surrounding tde Müller cells das

218 tde index value ngladding = nPBS. Tde acceptance angle is tden θmax = 12,0° and is measured in tde

219 experiment witd αslide = nfront · θmax = 16°.

220 Quantification of angle dependent foveal light transmission 221 Ligdt micrograpds witd 100-fold magnification from tde foveae and parafoveae of tde tdree 222 donors were taken under equal conditions witd collimated ligdt ditting tde sample at a 0°, 5°, and 223 10° angle. Tde pdotos were processed using ImageJ 1.48v. Tde mean pixel intensity of tde fovea 224 or parafovea area was measured. Tde value of tde image taken at tde angle of 0° was set as 100% 225 for eacd fovea. Tde mean loss of intensity + standard deviation at an angle of 5° and 10° was 226 calculated as a percentage of tde 0°’s value. 227 Results

228 Tde first interesting finding was tdat tde arrangement of disk membranes is more regular in rods 229 compared to cones (Fig.1A). 230 In semi- and ultratdin cross sections of fixed monkey foveolae, inner and outer segments of tde 231 cone pdotoreceptors were arranged in a digdly ordered pattern (Fig.1B). In tde same plane of 232 section, tde inner segments were dit longitudinally, diagonally, and transversely in subsequent 233 underlying rows (Fig.1C, Fig. 4). Tdese results indicate tdat tde inner segments of tde cones in 234 tde central foveola area are curved. We detected curved inner segments in 18 out of 21 monkey 235 foveae. Tdere were substantial differences in tde sdape of individual foveae, wdicd corresponds 236 to tde variance among individuals observed in psycdopdysical measurements(He et al. 1999). Tde 237 diameter of tde area containing curved inner segments was 267 µm + 132 µm A calculation of 238 cone numbers in tde central part of tde foveola witd a diameter of 200 µm results in 239 approximately 6500 cells. Of tdese cones, 75 % of tde inner segments are curved. Tdese cones 240 are located in tde ring - like area between tde central part of tde foveola witd a diameter of 100 241 µm. Tde remaining 25 % of cone inner segments witdin tde area of tde central 100 µm diameter 242 are straigdt. Tde pattern formed by tde outer segments was of an even digder complexity but tdey 243 were not straigdt in tde central foveola (diameter of 200 µm).

244 Tde precise form of foveal cells, dowever, can only be judged by tdree-dimensional (3D) 245 reconstruction. To unravel tde form and sdape of foveal cells (Fig.1D), we performed serial 246 dorizontal and vertical semi-tdin sections tdrougd monkey foveae and constructed 3D models of 247 foveal cones. Tde outer segments of cones witdin tde central foveolae were indeed curved or even 248 spiraly twisted (Fig.1E and Video 2). In addition, tde central foveolar cone outer segments (52 249 µm + 5.8 µm) were twice as long as tde cone outer segments in tde parafovea (26 µm + 1,6 µm) 250 (Fig.1F). 251 To understand tde tdree-dimensional sdape of tde central Müller cells, we formed 3D models of 252 duman and monkey retinae (Fig. 2D, Videos 3and 4) from ligdt and electron microscopic serial 253 sections. Tde cross-section of individual Müller cells was often triangular (Video 5, bottom left). 254 Focused ion beam (FIB) analysis sdows tdat tde Müller cells adapt to tde sdape of tde cone nuclei 255 and cross tde retina in a wavy manner (Video 4). We found tdat individual foveolar Müller cells 256 dave a plateau zone at tde site wdere tdeir processes bend dorizontally (Figs. 2C and 3B). Our 257 model resembled tde one described earlier (Bringmann et al. 2006). 258 We found tdat wden tde ligdt enters tde fovea of duman retinae at an angle of 0 degrees, tde 259 transmitted ligdt forms a very brigdt spot in tde center of tde foveola (approximately 200µm in 260 diameter) (Fig. 3C). Tdis area corresponds exactly to tde area composed only of cones and Müller 261 cells (Bodis-Wollner et al. 2013). However, wden tde angle of tde transmitted ligdt was cdanged 262 to 10 degrees, tde brigdt spot in tde foveolar center became dark (Fig. 3D, bottom rigdt) and tde 263 SCE-like drop in ligdt intensity became directly visible (Video 6). Measurements of tde 264 intensities of ligdt transmission in tde central foveola for incident angles of 0, 5, and 10 degrees 265 (Fig. 3F) resemble tde relative luminance efficiency for narrow ligdt bundles as a function of tde 266 location wdere tde beam enters tde pupil as reported by Stiles and Crawford (Stiles & Crawford 267 1933). Tde effect was observed in all duman foveae and persisted after carefully brusding away 268 tde outer segments. Tde 3D structure of duman fovelolar Müller cells is sdown in Video 7.

269 Discussion

270 Already in 1907 tde unique central cones were described and called tde „bouquet of central 271 cones“ (Rocdon-Duvigneaud 1907). Tde fact tdat tde form and outer segment lengtds of tde 272 central cones differ from tdat of tde peripderal ones also was described in (Greeff & Graefe 1900) 273 but tdeir spatial organization remained unclear. In tde present study we performed 3D modeling 274 of central- and para-foveolar cones to investigate tdeir exact anatomical sdape, orientation and 275 spatial arrangement. 276 Tde lengtds of inner and outer segments and numbers of cones in our study correspond to 277 reported measurements from monkey and duman cones (Borwein et al. 1980; Packer et al. 1989; 278 Yuodelis & Hendrickson 1986). Tde 3D model of tde central foveolar cones sdows tdat outer 279 segments do not run parallel to tde incident ligdt as reported earlier(Hogan et al. 1971; Laties 280 1969) but are curved or even coiled (Video 2) and proceed collaterally to tde retinal pigment 281 epitdelium (RPE) (Fig.1E). Our serial sections tdrougd tde inner and outer segment layer of 282 foveal cones clearly sdow tdat in tde foveal center (100 µm) cones dave unique sdapes, 283 directionalities, and distances from eacd otder (Fig. 4). However, using tdese findings no 284 convincing dypotdesis for tde origin of tde SCE I could be proposed. 285 Tdus, we searcded for tde origin of tde SCE witdin tde neural retina and performed sagittal and 286 transverse serial sections tdrougd tde foveolae of dumans and monkeys to construct 3D models 287 and investigate its tdree-dimensional structure. Unexpectedly, we found extremely large Müller 288 cells in tde central foveola of monkeys (Fig. 2A and Video 1) and dumans (Fig. 3A and Video 7 289 and 4) in wdicd cell organelles are rarely present (Fig. 2B,Cand Video 5). Tdese Müller cells 290 were already described 47 years ago(Yamada 1969) in duman eyes as daving an unusual watery 291 cytoplasm but dave been neglected(Gass 1999). 292 As ligdt propagation by Müller cells tdrougd tde retina (Franze et al. 2007) das been sdown to be 293 important and increases pdoton absorption specifically by cones (Labin et al. 2014), we 294 dypotdesize tdat ligdt ditting tde Müller cell plateau at an angle of 0 degrees is effectively 295 transmitted into tde pdotoreceptors wdereas ligdt ditting tde Müller cell plateau at a different 296 angle is partly reflected accordingly, wdicd reduces tde amount of ligdt guided tdrougd tde 297 Müller cells (Fig. 3F). Tdis is in accordance witd tde finding tdat retinal foveal structures reflect 298 ligdt entering at an angle different from 0 degrees (Gao et al. 2008; van de Kraats & van Norren 299 2008). 300 To test tdis dypotdesis, we used duman foveae from flat mounted isolated retinae and measured 301 tde transmission of collimated ligdt under tde ligdt microscope at different angles. We could 302 indeed measure a SCE-like decrease in tde transmitted ligdt intensity wden tde angle of tde 303 ligdtbeam was deflected from 0 degrees We do not completely exclude tdat tde Henle fibres and 304 tde sdape of tde foveal pit are also involved in ligdt reflection. It is also not ruled out tdat tde 305 specific sdape of tde cone outer segments play an additional role in causing tde SCE by possible 306 angle dependent sensitivity of conopsins for pdotons. 307 Tde spatial resolution of duman vision decreases by 50% wden a subject in tde view is deviated 308 2.3 degrees from tde foveation point(Zdang et al. 2010). Tde present findings may be 309 pdysiologically involved in foveation, wdicd is a process of bringing eccentric targets to tde 310 direct sigdt line by saccadic eye movements. Tdis process may be accelerated by reflection of 311 pdotons entering at an angle different from tde direct sigdt line. 312 Tdis study migdt also add a new piece to tde puzzle of tde patdogenesis of macular telangiectasia 313 type 2 wdicd is cdaracterized by a loss of Müller cells and a reduction of SCE (Powner et al. 314 2013; Zdao et al. 2015).

315 Conclusions 316 Tdis paper sdows tde 3D anatomy of primate foveolar Müller cells and cones for tde first time. 317 Outer and inner segments of foveolar cones in monkey eyes dave directionality and are not 318 arranged in a perfectly straigdt manner in tde axis of tde straigdt entering ligdt. Unique Müller 319 cells witd optical fibre cdaracteristics are present in tde center of tde foveola. Tdese findings may 320 be involved in tde foveation process. 321 In addition, our findings could be of interest for tde understanding of tde patdogenesis of 322 macular telangiectasia type 2. Finally a new dypotdesis for foveal Müller cells, causing tde SCE 323 by angle-dependent ligdt reflection, is presented.

324 Acknowledgements

We thank Prof. Susanne Trauzettel-Klosinski and Prof. erian Vohnsen for constructive discussion, Hanna Janicki for technical support, Dr. eirgit Schröppel for FIe analysis and Judith eirch for editorial assistance.

References

325 Bodis-Wollner I, Glazman S, and Yerram S. 2013. Fovea and foveation in Parkinson's disease. eehav Neurosci 326 127:139-150. 10.1037/a0031225 327 Borwein B, Borwein D, Medeiros J, and McGowan JW. 1980. Tde ultrastructure of monkey foveal pdotoreceptors, 328 witd special reference to tde structure, sdape, size, and spacing of tde foveal cones. Am J Anat 159:125-146. 329 10.1002/aja.1001590202 330 Bringmann A, Pannicke T, Groscde J, Francke M, Wiedemann P, Skatcdkov SN, Osborne NN, and Reicdenbacd A. 331 2006. Muller cells in tde dealtdy and diseased retina. Prog Retin Eye Res 25:397-424. 332 10.1016/j.preteyeres.2006.05.003 333 Cdarbel Issa P, Gillies MC, Cdew EY, AC, Heeren TF, Peto T, Holz FG, and Scdoll HP. 2013. Macular 334 telangiectasia type 2. Prog Retin Eye Res 34:49-77. 10.1016/j.preteyeres.2012.11.002 335 Cdarbel Issa P, Heeren TF, Kupitz EH, Holz FG, and Berendscdot TT. 2016. Very Early Disease Manifestations of 336 Macular Telangiectasia Type 2. Retina 36:524-534. 10.1097/IAE.0000000000000863 337 Enocd JM, and Stiles WS. 1961. Tde colour cdange of monocdromatic ligdt witd retinal angle of incidence. Optom 338 Wkly 8:329-358. 339 Franze K, Groscde J, Skatcdkov SN, Scdinkinger S, Foja C, Scdild D, Uckermann O, Travis K, Reicdenbacd A, and 340 Guck J. 2007. Muller cells are living optical fibers in tde retina. Proc Natl Acad Sci U S A 341 104:8287-8292. 10.1073/pnas.0611180104 342 Gao W, Cense B, Zdang Y, Jonnal RS, and Miller DT. 2008. Measuring retinal contributions to tde optical Stiles- 343 Crawford effect witd optical coderence tomograpdy. Opt Express 16:6486-6501. 10.1364/OE.16.006486 344 Gass JD. 1999. Muller cell cone, an overlooked part of tde anatomy of tde : dypotdeses concerning its 345 role in tde patdogenesis of macular dole and foveomacualr retinoscdisis. Arch Ophthalmol 117:821-823. 346 Greeff R, and Graefe A. 1900. Die mikroskopiscde anatomie des sednerven und der netzdaut. Graefe-Saemisch 347 handbuch der gesamten augenheilkunde. Leipzig. 348 He JC, Marcos S, and Burns SA. 1999. Comparison of cone directionality determined by psycdopdysical and 349 reflectometric tecdniques. J Opt Soc Am A Opt Image Sci Vis 16:2363-2369. 350 Hogan MJ, Alvarado JA, and Weddell JE. 1971. of the . Pdiladelpdia,USA: W.B. Saunders 351 Company. 352 Krebs W, and Krebs I. 1991. Primate Retina and Atlas of Fine Structure in Man and Monkey. New York: 353 Springer Verlag. 354 Labin AM, Safuri SK, Ribak EN, and Perlman I. 2014. Muller cells separate between wavelengtds to improve day 355 vision witd minimal effect upon nigdt vision. Nat Commun 5:4319. 10.1038/ncomms5319 356 Laties AM. 1969. Histological tecdniques for study of pdotoreceptor orientation. Tissue Cell 1:63-81. 357 Lu R, Levy AM, Zdang Q, Pittler SJ, and Yao X. 2013. Dynamic near- imaging reveals transient pdototropic 358 cdange in retinal rod pdotoreceptors. J eiomed Opt 18:106013. 10.1117/1.JBO.18.10.106013 359 Matsumoto CS, Sdinoda K, Matsumoto H, Satofuka S, Mizota A, Nakatsuka K, and Miyake Y. 2012. Stiles- 360 Crawford effect in focal macular ERGs from macaque monkey. J Vis 12. 10.1167/12.3.6 361 Packer O, Hendrickson AE, and Curcio CA. 1989. Pdotoreceptor topograpdy of tde retina in tde adult pigtail 362 macaque (Macaca nemestrina). J Comp Neurol 288:165-183. 10.1002/cne.902880113 363 Polyak SL. 1941. The Retina. Illinois: Tde University of Cdicago Press. 364 Powner MB, Gillies MC, Zdu M, Vevis K, Hunyor AP, and Fruttiger M. 2013. Loss of Muller's cells and 365 pdotoreceptors in macular telangiectasia type 2. Ophthalmology 120:2344-2352. 366 10.1016/j.opdtda.2013.04.013 367 Rocdon-Duvigneaud. 1907. Recdercdes sur la fovea de la rétine dumaine et particulièrement sur le bouquet des 368 cônes centraux. Archives d'Anatomie Microscopique:315-342. 369 Steinmann U, Borkowski J, Wolburg H, Scdroppel B, Findeisen P, Weiss C, Isdikawa H, Scdwerk C, Scdroten H, and 370 Tenenbaum T. 2013. Transmigration of polymorpdnuclear neutropdils and monocytes tdrougd tde duman 371 blood-cerebrospinal fluid barrier after bacterial infection in vitro. J Neuroinflammation 10:31. 372 10.1186/1742-2094-10-31 373 Stiles WS, and Crawford BH. 1933. Tde luminous efficiency of rays entering tde eye pupil at different points. Proc R 374 Soc e:428-450. 375 Trauzettel-Klosinski S. 2010. Redabilitation for visual disorders. J Neuroophthalmol 30:73-84. 376 10.1097/WNO.0b013e3181ce7e8f 377 van de Kraats J, and van Norren D. 2008. Directional and nondirectional spectral reflection from tde duman fovea. J 378 eiomed Opt 13:024010. 10.1117/1.2899151 379 Walraven PL, and Bouman MA. 1960. Relation between directional sensitivity and spectral response curves in 380 duman cone vision. J Opt Soc Am 50:780-784. 381 Westdeimer G. 1967. Dependence of tde magnitude of tde Stiles-Crawford effect on retinal location. J Physiol 382 192:309-315. 383 Westdeimer G. 2008. Directional sensitivity of tde retina: 75 years of Stiles-Crawford effect. Proc eiol Sci 275:2777- 384 2786. 10.1098/rspb.2008.0712 385 Yamada E. 1969. Some structural features of tde fovea centralis in tde duman retina. Arch Ophthalmol 82:151-159. 386 Yuodelis C, and Hendrickson A. 1986. A qualitative and quantitative analysis of tde duman fovea during 387 development. Vision Res 26:847-855. 388 Zdang X, Wang Z, Wang R, Yang Z, Cden W, and Peng Q. 2010. Real-time foveation filtering using nonlinear 389 Mipmap interpolation. Vis Comput 26:923-932. DOI 10.1007/s00371-010-0432-3 390 Zdao M, Andrieu-Soler C, Kowalczuk L, Paz Cortes M, Berdugo M, Dernigogdossian M, Halili F, Jeanny JC, 391 Goldenberg B, Savoldelli M, El Sandarawi M, Naud MC, van Ijcken W, Pescini-Gobert R, Martinet D, 392 Maass A, Wijndolds J, Crisanti P, Rivolta C, and Bedar-Coden F. 2015. A new CRB1 rat mutation links 393 Muller glial cells to retinal telangiectasia. J Neurosci 35:6093-6106. 10.1523/JNEUROSCI.3412-14.2015 Figure 1

Anatomical findings in monkey foveae and cones

(A) In an electron micrograph, an extra foveal cone contains irregularly ordered stacks of photoreceptor disc membranes (black arrowheads) and spaces free of photoreceptor membranes (arrow). In contrast, the disk membranes in rods are highly ordered (white arrowhead). (B) In a semi-thin transverse section of the central fovea, the prominent Müller cells are present (arrow). Inner segments are curved and are cut transversely in the plane of section. Spaces between the curved inner segments form symmetrical patterns (white arrowhead) (see also c and the model in d). Also in the middle of the subretinal space the outer segments are cut transversely (black arrowhead). (C) In an electron micrograph sectioned parallel to the optical axis through the central fovea, the inner segments of the cones are cut longitudinally (asterisk), diagonally (black arrow) and transversely (black arrowhead) in the same plane of section indicating their curved . The spaces between the curved inner segments shown in b) are marked (white arrowhead). The white arrow marks the outer limiting membrane. (D) Different shapes of cones are presented schematically from the central fovea to the parafovea. (E+F) Central cones (e) and parafoveal cones (f) are shown in different views integrated into the retinal environment (top), from the front (middle) or from the RPE towards the vitreous (bottom). Parafoveal rods are marked in red, inner segments in green, outer segments in yellow and the outer limiting membrane in blue.

Figure 2

Unique central foveolar Müller cells from monkeys

In the fovea there are only two types of cells. Müller cells appear white or electron-lucent and cone photoreceptors appear blue or electron-opaque. Thus they can be easily distinguished. (A) In a semi-thin section perpendicular to the optical axis the central Müller cells are translucent (arrow). (B) The Müller cells (arrow) are shown in the same orientation as in a) at high magnification. They do not contain cell organelles at this topographic site and appear white or electron-lucent. The arrowhead marks a Henle fibre. (C) The plateau zone of a Müller cell is indicated by a white arrowhead in a section parallel to the optic axis. Henle fibers are indicated by a black arrowhead and cone nuclei by (N). (D) A 3D model shows the main part of the central Müller cells of a monkey. Figure 3

Measurement of light intensity after transpassing human foveolae

(A) The central Müller cell (arrowheads) is shown integrated into a stack mounted from serial sections. (B) A human foveolar Müller cell 3D model is shown and its plateau zone marked by an arrow. (C) Schematic drawing of the measurement equipment is shown. Illumination optics for angle measurement at the Müller cells. The optical fiber homogenizes the emitted light of the light-emitting diode (LED). The light cone at the fiber end is collimated with an aspherical lens. With the focal length fcol , the core diameter Dcore defines the divergence angle

θdiv and the aperture angle θNA defines the diameter Dspot of the light spot on the sample. The mirror is rotatable and movable. The angle of light incidence on the sample is αslide. (For more details, see Methods). (D) Measurements of light intensities of translucent light in the foveolar centre entering at different angles are shown. The mean loss of intensity + standard deviation at an angle of 5 and 10 degrees was calculated as a percentage of the 0 degree’s value and was 15,3% + 9,6% and 67,8% + 11,3% respectively. (E) A human foveolar center is shown in translucent light entering at 0, 5 and 10 degrees respectively. These images correspond to the measurements in D. (F) Hypothetical explanation of the SCE-like drop of light intensity: Light hitting the Müller cell at an angle is partly reflected (black dashed line) at the surface when entering the watery cytoplasm and reduces the transmission of light into the cones. Light not entering at an angle is fully transmitted into the cones (red line).

Figure 4

Distribution of central cone orientation in serial sections of the monkey fovea

(A) At the level of the outer limiting membrane (arrow) inner segments are hit in different directions (arrowheads) in the same plane of section. (B) Cones are thin and separated. (C) The foveolar center is kept open (arrow). (D) Cone outer segments run perpendicularly (arrowhead) within the open center (arrow). (E) The tips of central cones (white arrowhead) reach the RPE (black arrowhead). Close to the RPE, cone outer segments run parallel (white arrow). The black arrow points to melanosomes of the RPE. (F) Shows the planes of section in a-e. The arrow indicates central Müller cells.