Proc. NatW Acad. Sci. USA Vol. 80, pp. 3069-3073, May 1983 Medical Sciences Electroanatomy of a unique amacrine cell in the rabbit retina (horseradish peroxidase staining/intracellular electrophysiology) ROBERT F. MILLER AND STEWART A. BLOOMFIELD* Departments of Ophthalmology, Physiology and Biophysics, Washington University School of Medicine, St. Louis, Missouri 63110 Communicated by Oliver H. Lowry, February 10, 1983 ABSTRACT Intracellular electrophysiological recordings were (ii) to permit the cell to maintain a comparatively large number obtained from a specialized class of "starburst" amacrine cells by of relatively short processes without compromising the synaptic using an isolated superperfused retina-eyecup preparation of the efficiency of single distal branches. A preliminary account of rabbit. These cells were injected intracellularly with horseradish these findings has been presented (10). peroxidase and identified with light microscopy. A computer-con- trolled image-processing system was used to map and display the MATERIALS AND METHODS three-dimensional dendritic organization and provide information on length and sublaminar distribution of dendritic processes. Physiology and Anatomy. Intracellular recordings were ob- Starburst amacrines show an unusual dendritic architecture that tained from single neurons of the rabbit retina by using the iso- includes thin intermediate dendritic segments. Analysis with steady- lated superfused eyecup preparation as described (11, 12). Im- state cable equations suggests that these thin segments may pro- paled neurons were physiologically characterized and subse- vide electrical isolation of distal processes, raising the possibility quently stained with HRP. Stained cells were examined micro- that a single dendrite, which lies beyond the thin segment, may scopically in flatmount view. Well-stained neurons were fur- constitute a functional subunit of the cell. ther analyzed by using a computer-assisted image;processing system (13), which compiled structural data in three dimen- For nearly a century, silver-staining techniques applied to the sions through control of the focus adjustment and the stage mi- vertebrate retina have revealed a rich variety of neuronal cell crometers. Once the cell was logged onto the computer, its three- types that vary in size, density, and form of dendritic branching dimensional structure could be visualized in any plane through patterns (1). Variations in morphology such as the overall den- computer processing and display on a plotter. dritic spread may relate to differences in receptive field size (2); The diameters of dendritic segments were measured on a the sublaminar distribution of dendritic branching among third- Jena Amplival microscope with a Zeiss long-working-distance order neurons also may determine physiological polarity of the oil-immersion objective. A micrometer-driven filar eyepiece cell (ON, OFF, ON-OFF) (3). What about variations in the was calibrated and used for diameter measurements. The av- number of dendrites and the pattern of dendritic architecture? erage diameter of a single process was determined for three What functional significance can be attributed to the micro- adjacent regions of each dendritic segment. Although tapering scopic design of dendrites and dendritic branching? This prob- of dendrites is a property of some amacrines (14), the dendrites lem has received comparatively little attention, although a re- of starburst amacrines reasonably approximate cylinders. cent analysis of Golgi-stained ganglion cells has appeared (4). Microelectrodes were fabricated from thin-walled triangular The question of dendritic architecture is especially relevant for tubing (Glass Co. of America, Bargaintown, New Jersey) and considerations of amacrine cells because the dendrites of these filled with 10% HRP (Boehringer Mannheim) dissolved in 0.5 neurons have both post- and presynaptic anatomy (5). A ques- M potassium acetate adjusted to pH 7.6 with 0.1 M Tris. In- tion often posed concerning amacrine cells is whether single tracellular injection of HRP was done by applying an alternat- dendrites or subdendritic regions may operate independently ing current of 3 Hz at a level of 1 nA for 10 min. The use of and serve to parcel the amacrine into functional subunits. Rall alternating current minimized polarization of the microelec- (6) has pointed out that a narrowing of processes may augment trode. After several cells were stained, the retina was super- the synaptic operations upstream from the stricture, and one perfused for an additional hour, followed by fixation in 1.5% study (7), emphasizing this concept, has suggested that single glutaraldehyde/1.5% paraformaldehyde (pH 7.3 in 0.1 M varicosities of some amacrine dendrites may form functional phosphate buffer) for 10 min. Prior to fixation, the retina-pig- subunits. ment epithelium was fixed to a gelatinized cover slip to min- We have obtained intracellular recordings from a specialized imize shrinking during the histological procedure. After fixa- class of amacrine cells that have cell bodies localized on op- tion, the pigment epithelium was removed and placed in Ringer's posing sides of the inner plexiform layer. Both ON and OFF solution to wash for 16 hr at 40C. This was followed by a brief types have been observed. When these cells are stained with wash with distilled water. The HRP reaction utilized benzidene intracellularly injected horseradish peroxidase (HRP), they clearly dihydrochloride according to the method outlined by Mesulam conform to the "starburst" amacrines described previously (8, (15). 9) on the basis of Golgi-stained material. These cells possess an We estimated the amount of tissue shrinkage in several unusual dendritic organization that consists of thin interme- preparations by measuring the shrinkage of slices of retina diate segments and varicosities confined to the more distal mounted on gelatinized cover slips or the shrinkage between branches. The application of steady-state cable equations sug- two dye injection "spots" with fast Green B iontophoretically gests that these thin segments serve (i) to promote electrical applied extracellularly. If the tissue was not secured to a cover isolation of dendritic segments distal to the thin segments and slip, tissue shrinkage in the range of 20-25% was seen, but the The publication costs ofthis article were defrayed in part by page charge Abbreviation: HRP, horseradish peroxidase. payment. This article must therefore be hereby marked "advertise- * Present address: The Biological Laboratories, Harvard Univ., Cam- ment" in accordance with 18 U.S.C. §1734 solely to indicate this fact. bridge, MA. 3069 Downloaded by guest on September 27, 2021 3070 Medical Sciences: Miller and Bloomfield Proc. Nad Acad. Sci. USA 80 (1983) gelatinized cover slip reduced shrinkage to 6-14%. branches. At each branch point, Gd is evaluated as three par- Mathematical Analysis. We estimated the steady-state elec- allel conductances. The two "outward looking" terms were de- trotonic cable losses along dendritic trees using the method- rived from the initial calculation (before Gn was computed), ology of Rall (16), whose initial description also includes the whereas the soma-looking segment is solved after Gn is com- mathematical derivation of these equations. The electrotonic puted. It is worth mentioning here that the method we have lengths of dendritic segments were computed by the well-known used is an explicit solution of the dendritic tree and is not de- equation rived from any assumption about branching patterns or whether dendrites conform to equivalent cylinders as outlined by Rall A = (Rm/Ri d/4)12, [1] (16, 17). In fact, dendrites of starburst amacrines deviate con- where A = characteristic length or length constant, Rm = spe- siderably from equivalent cylinder conditions. cific membrane resistance (Q'cm2), Ri = specific internal re- To calculate steady-state voltage decay along the dendrite, sistance (fl'cm), and d = diameter. For all our calculations, we from a point of current injection, one is constrained by differ- assumed an Rm value of 4,000 fi.cm2 and an Ri value of 70 tlcm, ent boundary conditions that depend on the direction of cur- values commonly used for such calculations in mammalian neu- rent flow within the dendrite. Towards the periphery, rons. Based on the anatomical measurements of length and di- cosh (1 - x)A ameter, we calculated the following: [7] Gn = Gs + I Gd, [2] cosh l/A where Gn = whole neuron conductance, G. = soma conduc- where V = voltage at a point x and Vo = voltage at point of tance, and I Gd = summation of each individual dendritic con- application. Voltage decay towards the soma is determined by ductance at the soma. If each dendrite is similar, one need only evaluate Gd for a single dendrite and I Gd = n(Gd), where n V = number of equivalent dendrites. If dendrites show consid- =VOcosh(l/A) + B1 sinh(l/A)'[8[8] erable dissimilarity in structure, it is necessary to solve each dendrite separately. In order to solve Gd for any single den- where B1 is as in Eq. 5. drite, one begins at the distal dendritic branches and computes, collectively, towards the soma. RESULTS AND DISCUSSION The input conductance of a dendrite at the origin (Gda) is During the course of this study, we recorded from and stained given by: several types of neurons (18, 19). One type of amacrine ap- = Bo (1T/2)(Rm Ri)-12 (do)3/2 [3] peared to be unique and striking in its dendritic architecture. Gd. Fig. 1A shows a slightly retouched photograph of a HRP-stained where Bo is a computed value that expresses the conductance OFF-type starburst amacrine. In this cell and in all starburst of the dendrite as a fraction of an infinitely long extension of amacrines stained, the stain only lightly penetrated the den- the primary dendrite. Bo is determined by sequential analysis dritic tree, and some dye leakage obscured the anatomic ar- of the entire dendritic tree beginning at the most distal branches: rangement of the branches near the soma. For this reason, spe- Bj+l + tanh(l/A) cial efforts were required to obtain a photographic representation of a flat-mount view of the cell.