Development of Synaptic Arrays in the Inner Plexiform Layer of Neonatal Mouse Retina LESLIE J. FISHER The University of Michigan, Ann Arbor, Michigan 481 09 ABSTRACT Retinas from mice of the C57BL/6 strain were sampled at fre- quent intervals from birth to postnatal day 33 to determine the numerical den- sity of conventional and ribbon synapses within the inner plexiform layer (IPL) as a function of time. Synaptic arrays of the IPL were formed in three phases. During Phase I, from day 3 to day 10, conventional synapses were produced at a mean rate of 0.44 synapsesl1,OOO pm3/hour, but no ribbons were seen. During Phase 11, from day 11 to day 15, ribbons formed at a rate of 0.38 ribbons/1,000 pm3/hour and conventional synapses were produced at a rate of 1.15 synapsed 1,000 p.m31hour. Phase 111 began at day 15, the approximate time of eye open- ing in these animals, and was characterized by a sharp reduction in the rate of production of both ribbons and conventional synapses. During this phase rib- bons achieved a final mean density of 113 ribbons/1,000 prn' and conventionals achieved a final mean density of 250 synapses/1,000 fim3. Serial synapses ap- peared in Phase I1 but remained at low densities. The retinal inner plexiform layer (IPL) is MATERIALS AND METHODS particularly suitable for quantitative analysis A colony of the inbred strain of C57BL/6 because of its precisely defined borders, its mice originally purchased from Microbio- involvement with only three presynaptic neu- logical Associates was maintained under a ronal elements, and the ability to distinguish 14-hour light, 10-hour dark cycle with food ribbon from conventional synaptic complexes. (Purina Rat Chow) and water supplied ad Ribbon synapses in the IPL are unique to bi- libitum. polar cells while symmetrical or asymmet- After random mating, pregnant females rical conventional synapses are found only were segregated and kept in separate cages in the processes of amacrine cells (Dowling with their litters. Daily inspections deter- and Boycott, '66; Raviola and Raviola, '67; mined the day of birth (day 0). Neonates of Witkovsky and Dowling, '69) or interplex- both sexes were studied. iform cells (Dowling et al., '76; Kolb and West, Unanesthetized animals were killed by cer- '77). Furthermore, serial synapses can be iden- vical dislocation at the same time of day (1300 tified as a subset of conventional synapses in 2 1 hour). Eyes were sutured with 7-0 suture which a process is simultaneously postsynap- silk at the dorsal-most aspect-which allowed tic to a conventional synapse from an ama- orientation during embedding and section- crine cell or an interplexiform cell and presyn- ing-and then excised. After hemisection and aptic to a bipolar, amacrine, interplexiform, or lentectomy the eyes were fixed in a combined ganglion cell (Dowling and Boycott, '66; fixative of 3% glutaraldehyde, 2% formalde- Dubin, '70; Kolb and West, '77).This ability to hyde, 1% acrolein and 2.5% dimethyl sulfoxide precisely identify the input and interneuronal in 0.1 M sodium cacocylate buffer at pH 7.2 for connections allows a detailed analysis to be approximately 18 hours, washed in 0.1 M made of the temporal and spatial patterns of cacodylate buffer and post-fixed in 2%osmium connectivity within the neuropil. The purpose tetroxide in the same buffer. Eyes were of this paper is to describe quantitatively the stained en bloc with 0.5% aqueous uranyl ace- temporal formation of synaptic arrays in the tate, rinsed, dehydrated through alcohols, and mouse from an undifferentiated state at birth embedded in Epon. until the prepubertal period at 33 days of age. A second set of animals (6 adults, 8 pups- J. COMP. NEUR. (1979) 187: 359372 359 360 LESLIE J. FISHER 13 days old) were fixed by one cycle through were associated with membrane specializa- the Golgi-Kopsch procedure reported by Col- tions in the same plane of section, occasionally onnier ('64).The eyes were immersed for seven ribbons not lying near a membrane were days each in the chromate and silver nitrate found. The statistics reported below include solutions. These Golgi-stained retinas were ribbons both with and without an associated used to identify neuronal characteristics membrane specialization. In every instance which could be used to determine cell types in counts of synapses and ribbons were done the experimental retinas. using a blind procedure in which the age and Eyes were serially sectioned at 75 pm in the animal were not known until the data were horizontal plane beginning at the ventral- reduced. most point and mounted onto plastic slides. The section which passed through the optic Quantitative analysis disc was located and a portion of retina within The synapses are first defined in terms of 750 pm of the optic disc was mounted on a their numerical densities (Weibel, '69) which plastic blank for semi-thin (1 pm) and thin are expressed in numbers per 1,000 pm3. This (600-700 A) sectioning. All electron micro- statistic is the mean number of nuclei or syn- graphs were taken between 150 pm and 500 apses per unit volume of the retinal layer in pm from the optic nerve. question. Nuclei were counted and layer thicknesses Statistics of developmental interest are the were measured on semi-thin sections stained numbers of synapses or cells present at a given with methylene blue. Nuclear densities were locus at a given time, and the rate of synapse computed from data obtained by counting and formation. Retinal planimetric density and using an Abercrombie correction (Abercrom- synapses per neuron expressed as a function of bie, '46). age supply these data. Electron micrographic mosaics (final mag- Retinal planimetric density is computed by nification 26,000 X) of the entire thickness multiplying numerical density by thickness of (vitreal to scleral dimension) of IPL were con- the layer in question. The product is given in structed using micrographs of thin sections numbers per 1,000 pmLand represents the stained with both uranyl acetate and lead cit- number of synapses or nuclei contained in a rate. The average area of IPL covered by one column which has a cross-sectional area of mosaic was 900 pm2 as measured by polar 1,000 pm2and a height equal to the thickness planimetry. of the layer. The area on which these nuclei or To insure that the magnification of each synapses can be considered to be located is in mosaic was precisely known an electron mi- the plane of the layer at a level midway be- crograph of a calibration grid (Fullam No. tween the layer's boundaries with the adja- 1002) was taken each time the magnification cent layers. Planimetric densities, therefore, of the electron microscope was changed. At are the number of nuclei or synapses referred least one calibration micrograph was taken to a unit area in the plane of the retina. per session, and all were processed with the Planimetric density gives the number of syn- mosaics. Every mosaic, therefore, has a apses or nuclei which directly underlie the re- calibration micrograph to which it is directly ceptors excited by a spot of light of unit area. related. The data are also expressed as the mean Ribbon, conventional, and serial-conven- number of synapses per inner nuclear layer tional synapses were identified and the (INL) nucleus by dividing synaptic plani- lengths of their profiles were measured on the metric by nuclear planimetric density. Rates mosaics. The resulting data were reduced of ribbon and conventional synapse formation using a modified Abercrombie correction per neuron can be calculated from these sta- (Dubin, '70). Serial-conventionals were in- tistics when plotted as a function of time. cluded in the data to be presented for con- RESULTS ventionals. The criterion used to identify a conven- Golgi description tional synapse was the presence of at least To express the synaptic data as a function of three synaptic vesicles grouped next to a the number of amacrine or bipolar nuclei, a membrane specialization (fig. 1). The pres- reliable method of identifying the individual ence of the ribbon was the criterion for ribbon nuclei in the INL was needed. Accordingly, synapses (fig. 2). While most of the ribbons Golgi-stained retinas were analyzed to iden- SYNAPTIC ARRAY DEVELOPMENT IN MOUSE 361 Fig. 1 Conventional synapse from the mouse IPL (arrow). Calibration Bar = 0.5 pm Fig. 2 Ribbon from the mouse IPL (arrow). Calibration Bar = 0.5 pm 362 LESLIE J. FISHER Fig. 3 Muller cells of the mouse retina. In the mouse the nuclei of Muller cells form a single stratum (arrow). INL, Inner Nuclear Layer; IPL, Inner Plexiform Layer. Golgi-Kopsch stain. Calibration Bar = 10 pm. Fig. 4 Bipolar cell (single arrow) and Muller cell (double arrow) of the mouse retina as stained by the Golgi-Kopsch method. In all of the Golgi-stained material examined, bipolar nuclei were found sclerad to the level of Muller nuclei. INL, Inner Nuclear Layer; IPL, Inner Plexiform Layer. Calibration Bar = 10 pm. SYNAPTIC ARRAY DEVELOPMENT IN MOUSE 363 Fig. 5 Amacrine cell (single arrow) and bipolar cell (double arrow) of the mouse retina. Amacrine nuclei were always located vitread to the Muller nuclei. INL, Inner Nuclear Layer; IPL, Inner Plexiform Layer. Golgi-Kopsch stain. Calibration Bar = 10 Fm. Fig. 6 Mouse retina. The stratum of darkly stained nuclei (arrow) corresponds to the Muller nuclei seen in the Golgi-stained material (fig. 3). OPL, Outer Plexiform Layer; GCL, Ganglion Cell Layer; INL, Inner Nuclear Layer; IPL, Inner Plexiform Layer. Methylene Blue stain. Calibration Bar = 10 pm, 364 LESLIE J.