Psychophysical results suggest that the primate visual system is equally sensitive to both the onset and offset of short-wavelength light and that these responses are carried by separate pathways. However, physiological studies of cells in the retina and lateral geniculate nucleus find far fewer OFF-center than ON-center cells whose receptive-field centers are driven by short-wavelength-sensitive (S) cones. To determine whether S cones contact ON and OFF midget bipolar cells as well as (ON) “blue-cone bipolar” cells (Mariani, 1984), we examined 118 contiguous cone terminals and their bipolar cells in electron micrographs of serial sections from macaque foveal retina. Five widely spaced cone terminals do not contact ON midget bipolar cells. These five cone terminals contact the dendrites of “blue-cone bipolar” cells instead, showing that they are the terminals of S cones. These S-cone terminals are smaller and contain more synaptic ribbons than other terminals. Like neighboring cones, each S cone contacts its own OFF midget bipolar cell via triad-associated (flat) synaptic contacts. Moreover, each S-cone OFF midget bipolar cell has a synaptic terminal in the outer half of the inner plexiform layer, where it contacts an OFF midget ganglion cell.
- S cone
- blue cone
- retinal ganglion cell
- color perception
- retinal anatomy
- retinal histology
Psychophysical results suggest that the primate visual system is equally sensitive to both the onset and offset of short-wavelength light and that these responses are carried by separate pathways (Smith et al., 1989; DeMarco et al., 1994; Chichilnisky and Wandell, 1996; McLellan and Eskew, 2000). However, physiological studies of cells in the retina and lateral geniculate nucleus (LGN) find far fewer OFF-center cells than ON-center cells whose receptive-fields are driven by short-wavelength-sensitive (S) cones (Krüger, 1977; Malpeli and Schiller, 1978; de Monasterio, 1979; Zrenner and Gouras, 1981; Zrenner, 1983) (but see Valberg et al., 1986).
The scarcity of cells with S-cone OFF-center receptive fields was surprising because Golgi studies suggested that each foveal cone is presynaptic to one invaginating (ON) and one flat (OFF) midget bipolar cell and that each midget bipolar cell is presynaptic to a single midget ganglion cell (Polyak, 1941; Boycott and Dowling, 1969; Kolb et al., 1969; Kolb, 1970). Moreover, physiologists had found approximately equal numbers of ON and OFF responses among ganglion cells and LGN cells with receptive-field centers dominated by middle- or long-wavelength-sensitive (M or L) cones (Wiesel and Hubel, 1966; de Monasterio and Gouras, 1975). The rarity of cells with S-cone OFF-center receptive fields raised the possibility that S cones, unlike L and M cones, might not drive OFF midget bipolar and ganglion cells and that an S-cone-specific OFF circuit might not exist.
The picture of S-cone circuitry was further complicated by the discovery of a special “blue-cone bipolar” cell contacted exclusively by several S cones (Mariani, 1984; Kouyama and Marshak, 1992, 1997). This S-cone-specific ON bipolar cell provides the ON signals to a ganglion cell that has been described anatomically as a small bistratified cell and physiologically as receiving ON signals from S cones and coextensive OFF signals from L and M cones (Rodieck, 1991; Dacey, 1993; Dacey and Lee, 1994; Calkins et al., 1998; Chichilnisky and Baylor, 1999). The discovery of this S-cone-specific ON bipolar cell raised another question: are S cones also presynaptic to ON midget bipolar cells?
We identified each cone and each midget and diffuse bipolar cell in a small region of macaque monkey foveal retina by reconstructing cells from serial sections (see Fig. 1A). S cones contact the distinctive (ON) blue-cone bipolar cell described above (see Fig. 1B). Each L or M cone is presynaptic to its own ON midget bipolar cell, but none of the S cones are. In contrast, every cone, including every S cone, is presynaptic to an OFF midget bipolar cell. Like other midget bipolar cells, these S-cone OFF midget bipolar cells are presynaptic to midget ganglion cells. We conclude that S cones, like L and M cones, provide input to both ON and OFF cone-specific retinal circuits. The S-cone ON circuit is unique, with its own ON bipolar and ganglion cells, whereas the S-cone OFF circuit uses midget bipolar and ganglion cells.
Materials and Methods
As described by Tsukamoto et al. (1992), a 6 kg adult male monkey (Macaca fascicularis) was given a preanesthetic dose of ketamine (10 mg/kg) and a deep anesthetic dose (40 mg/kg) of pentobarbital, followed by perfusion through the aorta. Perfusion consisted of a brief (20 sec) flush with saline, followed by 10 min of a solution containing 100 mm NaCl, 30 mm sodium phosphate buffer, pH 7.40, 2% glutaraldehyde, and 2% paraformaldehyde. The retina was removed and postfixed in 2% OsO4 in 8% dextrose for 30 min at 4°C and 60 min at 22°C. A piece of the retina containing the fovea was dissected, stained en bloc in 1% aqueous uranyl acetate, dehydrated, and embedded in Epon. A series of 319 consecutive ultrathin (90 nm) vertical sections were cut parallel to the horizontal meridian, through the fovea. The sections were mounted and then stained with uranyl acetate and lead citrate.
The region from 480 to 640 μm nasal to the center of the retina was photographed en montage at 400 and 2000× magnification using an electron microscope and printed with 2.5× enlargement for final magnifications of 1000 and 5000×. The outer plexiform layer and parts of the inner plexiform layer (IPL) were rephotographed at 5000× and enlarged 3.5× for 17,500× prints. Selected photoreceptor terminals were rephotographed at 10,000× and enlarged 2.5×. Cells in this region connect to foveal cone inner segments centered around 1° of visual eccentricity because of the lateral displacement of cone terminals from cone inner and outer segments (Schein, 1988; Tsukamoto et al., 1992). The region studied is well within foveal retina because the fovea extends to 500 μm, or 2.5° (Polyak, 1941).
Cone terminals, bipolar cells, and ganglion cells were tracked and reconstructed from the series of electron micrographs (see Fig. 2 A). Each bipolar cell was classified as an ON or OFF midget bipolar, ON or OFF diffuse bipolar, rod bipolar, or blue-cone bipolar cell. Because the orientation of the sections is not perfectly vertical, midget bipolar and ganglion cells of some cones fall outside of the series and could not be completely reconstructed. For example, the midget bipolar cells contacted by cone terminals in the low-numbered sections at the bottom of Figure 1 A synapse with ganglion cells that are located in the high-numbered sections at the top of the same figure. Contours were traced from 5000 or 17,500× prints onto clear plastic sheets and digitized using the Montage software package (Smith, 1987). Some wire-frame models were tiled with the ContourFitter program (Meyers et al., 1992), visualized with Geomview (http://www.geomview.org/), and rendered with the Blue Moon Rendering Tools (NVIDIA, Santa Clara, CA). Montage was also used to calculate the volumes of 43 reconstructed cone terminals by summing volumes (surface area × section thickness) section by section.
S cones do not contact ON midget bipolar cells
We identified all of the midget bipolar cells postsynaptic to the 118 shaded cone synaptic terminals in Figure 1A. Foveal cone terminals have ∼20 ribbons (Esfahani et al., 1993; Calkins et al., 1996), each presynaptic to one or more “triads” of postsynaptic dendrites: two lateral elements, provided by the dendrites of horizontal cells, and one or more central elements, provided by dendrites of invaginating (ON) bipolar cells (Fig. 2B). Invaginating bipolar cells, whose terminals synapse with ganglion cell dendrites in the inner half of the inner plexiform layer, have ON-center receptive fields (Famiglietti and Kolb, 1976; Nelson et al., 1978).
We identified all of the ON midget bipolar cells according to the following criteria: (1) their dendrites are postsynaptic to a single cone and (2) provide central elements at ribbon synapses (Kolb, 1970); (3) their synaptic terminals are located in the inner half of the IPL; and (4) their somas are more densely stained in electron micrographs compared with somas of OFF bipolar cells (Ahmad et al., 2003). One ON midget bipolar cell is “two headed” and is shared by two L or M cones (Polyak, 1941; Hopkins and Boycott, 1992). Of 118 cone terminals, 113 are presynaptic to an ON midget bipolar cell.
Five cones (identified in Fig. 1A) do not contact ON midget bipolar cells. The central elements postsynaptic to these five exceptional cones are provided instead by ON bipolar cells whose dendrites branch widely and contact several cones (Fig. 1B). Unlike the ON and OFF diffuse bipolar cells in our series, which contact all of the cones within a small region, these bipolar cells reach past all intervening cones and contact only the exceptional cones. The synaptic terminals of two of these bipolar cells are contained within the series and are located in the innermost IPL, closer to the ganglion cell layer than the terminals of ON midget bipolar cells (Fig. 3). These bipolar cells can thus be classified as blue-cone bipolar cells (Mariani, 1984; Kouyama and Marshak, 1992, 1997), and the exceptional cones can be classified as S cones. We rename this bipolar cell type the “S-cone ON bipolar” to distinguish it from other bipolar cells that are postsynaptic to S cones, to indicate that it is depolarized by the onset of short-wavelength light and to emphasize the difference between the spectral sensitivity of S cones and the perceptual quality of blueness. Each S cone contacts several S-cone ON bipolar cells (Fig. 1B) (Herr et al., 2003).
Four of the five S-cone terminals are completely contained within the series and were reconstructed. The number of ribbons in each of the four complete terminals is either 22 or 23, more than that in any other cone terminal in our series (Fig. 4A) (Esfahani et al., 1993). The volumes of the four cone terminals are also among the lowest of all terminals at any given eccentricity (Fig. 4B). As a consequence, the density of ribbons is highest in the S-cone terminals, as reported by Ahnelt et al. (1990) and Haverkamp et al. (2001).
A sixth cone terminal (Fig. 1A, 118, dashed outline) contacts two S-cone ON bipolar cells and so was also classified as an S cone. This cone terminal is only partially contained within the series, so we could not determine conclusively whether or not it contacts an ON midget bipolar cell. The proportion of S cones (6 of 118, ∼5%) is consistent with other reports of S-cone density at similar eccentricities (Williams et al., 1981; de Monasterio et al., 1985; Curcio et al., 1991; Roorda et al., 2001).
S cones contact OFF midget bipolar cells
OFF midget bipolar cells postsynaptic to L or M cones: (1) make “flat” contacts with the basal surface of a single cone terminal (Kolb, 1970); (2) have synaptic terminals located in the outer half of the IPL (Famiglietti and Kolb, 1976; Nelson et al., 1978); and (3) appear pale in electron micrographs (Ahmad et al., 2003). This description applies equally well to the subset of OFF midget bipolar cells postsynaptic to S cones.
The flat synaptic contacts from S cones to OFF midget bipolar cell dendrites are symmetrically thickened (Fig. 2B) (Ahmad et al., 2003) and located near the synaptic ribbon in the triad-associated position, adjacent to contacts with the central elements provided by dendrites of S-cone ON bipolar cells (Fig. 5). The dendritic arbors of two S-cone OFF midget bipolar cells (cells 35 and 52), reconstructed from higher-magnification photomicrographs, receive, respectively, 34 and 35 of these triad-associated contacts from their (S) cone terminals, similar to the number of contacts between some L- and M-cone terminals and their OFF midget bipolar cells (Herr et al., 1995).
We reconstructed two S-cone OFF midget bipolar cells that are completely contained within the sectioned retina. Their synaptic terminals are located in the outer half of the IPL, at the same level as the terminals of OFF midget bipolar cells postsynaptic to non-S cones (Fig. 6). The overall morphology of S-cone OFF midget bipolar cells, as well as finer details of synaptic structure, are indistinguishable from those of OFF midget bipolar cells contacted by L or M cones.
All six of the S cones, and all of the other 113 cones in our region, are presynaptic to an OFF midget bipolar cell.
S-cone OFF midget ganglion cells
Like those of other midget bipolar cells, the synaptic terminals of S-cone midget bipolar cells contain synaptic ribbons (Fig. 7), each presynaptic to a “dyad” of postsynaptic elements (Dowling and Boycott, 1966). In almost all cases, one postsynaptic element is provided by a midget ganglion cell and one by an amacrine cell (Calkins et al., 1996). From photomicrographs of the IPL, we reconstructed the synaptic terminals of OFF midget bipolar cells and the dendritic arbors of their midget ganglion cells. The synaptic terminal of each S-cone OFF midget bipolar cell (Fig. 8A) is enveloped by the dendrites of a single midget ganglion cell (Fig. 8B).
The synaptic terminals of the two S-cone OFF midget bipolar cells shown in Figure 6 (cells 144 and 35) contain, respectively, 30 and 32 synaptic ribbons, similar to the number of ribbons (32 ± 3) reported for some other OFF midget bipolar cells from the same serial sections (Calkins et al., 1994). The dendrites of the two corresponding OFF midget ganglion cells receive 24 and 29 ribbon synaptic inputs from their bipolar cells, also similar to other midget ganglion cells. The two reconstructed S-cone OFF midget bipolar cells provide the only ribbon inputs to their partner OFF midget ganglion cells, although a small number of dyadic ribbons are presynaptic to dendrites from other, nonmidget, ganglion cells (Calkins et al., 1994).
S-cones and midget bipolar cells
S cones are not presynaptic to ON midget bipolar cells. Instead, as described previously, each S cone is presynaptic to several S-cone ON bipolar cells (Mariani, 1984; Kouyama and Marshak, 1992, 1997), which provide all of the central elements at cone ribbon synapses (Herr et al., 2003). In contrast, each L and M cone is presynaptic to an ON midget bipolar cell.
S cones are presynaptic to OFF midget bipolar cells, which in turn contact their own OFF midget ganglion cells. The OFF midget cells of S cones resemble L- and M-cone OFF midget cells in every feature that we examined, including the following: overall appearance (Fig. 6); symmetric cone synaptic contacts (Fig. 2B); the number and triad-associated location of those synapses (Figs. 2B, 5B); the location of the bipolar cell–ganglion cell synapses in the outer half of the IPL (Figs. 6, 7); the number of ribbons in the synaptic terminal of the bipolar cell; and the exclusive synaptic relationship of the midget ganglion cell with a single midget bipolar cell (Fig. 8).
Kolb et al. (1997) reconstructed S-cone terminals in peripheral human retina but were unable to identify contacts that were clearly presynaptic to an OFF midget bipolar cell. It is possible that peripheral S cones do not contact OFF midget bipolar cells; alternatively, in the work of Kolb et al. (1997), these synapses may have been missed because the bipolar cells were not reconstructed.
S-cone OFF midget circuits are more difficult to record
Some investigators reported that S-cone OFF-center cells were uncommon in the retina (Malpeli and Schiller, 1978; de Monasterio, 1979; Zrenner and Gouras, 1981; Zrenner, 1983) and LGN (Krüger, 1977; Malpeli and Schiller, 1978). However, De Valois et al. (1966) and Valberg et al. (1986) reported substantial numbers of S-cone OFF cells in the LGN. Because our data suggest that every foveal S cone contacts one OFF midget bipolar cell–midget ganglion cell pair, why did most physiological studies find strikingly fewer S-cone OFF cells than S-cone ON-cells?
First, there are two small bistratified (ON) ganglion cells, but only one OFF midget ganglion cell, for each S cone (Ahmad et al., 2003; Herr et al., 2003). Second, axial chromatic aberration spreads the image of a tiny short-wavelength spot over a large “blur circle” with a diameter similar to the spacing between S cones (Williams et al., 1981; Curcio et al., 1991; Wandell, 1995). The receptive-field center of an S-cone OFF midget cell, driven by a single S cone, consequently intercepts a small fraction of the light in the blurred short-wavelength spot, whereas L and M cones in the receptive-field ON surround intercept most of the light and might conceal the OFF-center response (de Monasterio et al., 1975). Long-wavelength stimulation would strongly activate the opponent surround, giving the appearance of a large yellow-ON receptive field. S-Cone OFF midget cells might therefore have been misclassified as noncolor opponent (physiological type III) cells (Wiesel and Hubel, 1966; de Monasterio and Gouras, 1975; Dreher et al., 1976; Marrocco, 1976). Supporting this hypothesis, chromatic adaptation by long-wavelength light revealed OFF inputs from S cones to type III cells (Zrenner and Gouras, 1983, their Fig. 3), and selective stimulation of S cones by silent substitution revealed cells with S-cone OFF centers (Derrington et al., 1984; Valberg et al., 1986).
Third, S-cone OFF midget cells might have been misclassified as cells with coextensive receptive-field center and surround (physiological type II) because blurring attributable to chromatic aberration increases the size of an S-cone receptive-field center. In fact, many S-cone OFF ganglion and LGN cells were classified as type II cells (Wiesel and Hubel, 1966, their Fig. 7) (de Monasterio and Gouras, 1975; Reid and Shapley, 2002).
Finally, physiologists generally record outside the fovea, where the proportion of S cones is lower and midget ganglion cells may receive mixed cone input (Watanabe and Rodieck, 1989; Dacey, 1999, 2000). Consequently, the S-cone input to peripheral midget cells may be overwhelmed by signals from the more numerous L and M cones.
Psychophysical evidence for separate ON and OFF S-cone channels
In contrast to the mixed physiological results, there is ample psychophysical evidence for the existence of an OFF midget pathway for S-cone signals. Measurements of grating acuity imply that the retina contains a pathway that preserves signals from single foveal S cones (Metha and Lennie, 2001). Shinomori et al. (1999) reported that the threshold for detecting an S-cone stimulus with a rapid-ON sawtooth temporal waveform was elevated more by adaptation to a rapid-ON waveform than to a rapid-OFF waveform and vice versa, suggesting that S-cone signals are carried by two pathways that can be differentially adapted. Using a noise-masking technique to reduce sensitivity separately in the S+ and S– chromatic directions, Sankeralli and Mullen (2001) reported similar results.
Interactions between cone types also suggest separate ON and OFF pathways. Using transient tritanopia, an elevation of detection thresholds by S cones after the offset of a long-wavelength background, McLellan and Eskew (2000) found that L-cone stimulation in the background affected detection of S-cone increments more than decrements. Color judgments showed a similar asymmetry: the appearance of test patches darker than a surrounding field could be described by a linear combination of S-, M- and L-cone quantal absorptions, whereas the appearance of test patches lighter than the background required a nonlinear contribution from the S cones (Chichilnisky and Wandell, 1996; Knoblauch and Shevell, 2001).
Finally, inactivating the ON pathway by exposing the retina to l-2-amino-4-phosphonobutyric acid (l-APB or l-AP-4), which blocks the glutamate receptors of ON but not OFF bipolar cells (Slaughter and Miller, 1981; Schiller et al., 1986), did not change the shape of the spectral sensitivity function, as would be expected if S-cone signals were carried only by ON cells (Smith et al., 1989). Instead, l-APB reduced sensitivity by the same amount at all wavelengths, suggesting the existence of an OFF pathway sensitive to short-wavelength stimuli. Together with other studies demonstrating equal sensitivity to short-wavelength increments and decrements (DeMarco et al., 1994; Schwartz, 1996), the l-APB experiment suggests that the strength of S-cone signals relative to L-and M-cone signals is similar in the ON and OFF pathways.
A role for S-cone OFF midget cells in color perception
The circuitry of the small bistratified ganglion cell is consistent with the coextensive +S/-L-M physiology required for cells contributing to blueness–yellowness (Dacey and Lee, 1994; Calkins et al., 1998; Cottaris and De Valois, 1998; Dacey, 2000; Calkins, 2001). In contrast, no ganglion cell type yet identified has the properties expected for a redness–greenness channel, i.e., coextensive receptive fields driven by L and S cones in opposition to M cones (Hurvich and Jameson, 1957; Wooten and Werner, 1979; Shevell and Humanski, 1988; Rodieck, 1991). It is therefore widely believed that the sensation of redness–greenness is constructed in the visual cortex from signals carried by midget ganglion cells (for an alternative model, see Calkins and Sterling, 1999).
Do S-cone OFF midget bipolar and ganglion cells contribute to these cortically constructed redness–greenness color channels? In particular, do S-cone OFF midget cells contribute to the short-wavelength component of redness? The results of Shinomori et al. (1999) suggest that they do. Their experiments identified separate OFF and ON channels driven by S cones when an adapting sawtooth stimulus left the perception of redness unadapted and available for the detection task, suggesting that both S-cone OFF and ON cells signal redness.
In contrast, S-cone OFF cells do not appear to contribute to blueness: Shinomori et al. (1999) also found that thresholds were determined by the ON channel alone when an adapting sawtooth left only the perception of blueness available for the detection task.
The hypothesis that S-cone OFF midget cells contribute to redness but not blueness leads to a specific prediction: exposure of the retina to l-APB should block blueness but leave intact an S-cone contribution to redness, causing short-wavelength stimuli to appear red rather than violet.
Sankeralli and Mullen (2001) concluded that yellowness and blueness are carried by separate pathways. Is yellowness signaled by a reduction in the firing rate of S-cone ON ganglion cells or by an increase in the firing rate of S-cone OFF ganglion cells, or by both? Because l-APB greatly reduces the sensitivity of small bistratified (S-cone ON) ganglion cells to yellow-OFF stimuli (Dacey, 2000), the perception of yellowness should be intact in the presence of l-APB if it is carried by S-cone OFF ganglion cells but will be eliminated or reduced if carried wholly or partly by S-cone ON ganglion cells.
This work was supported by National Institute of Mental Health Grant MH15795-18 and National Institutes of Health Grants EY11153 and EY08124. We thank Pooneh Esfahani for counting ribbons and for reconstructing cone terminals, Jack Ribble, Garrett Eggers, and Vanessa Preciado for help with bipolar and ganglion cell reconstructions, Dr. Yoshihiko Tsukamoto, Patricia Masarachia, and Sally Shrom for preparing the EM material, and Kazuki Uema and Lisa Travis for printing.
Correspondence should be addressed to Stan Schein, Department of Psychology, Franz Hall, Mail code 951563, University of California, Los Angeles, Los Angeles, CA 90095-1563. E-mail:.
Copyright © 2003 Society for Neuroscience 0270-6474/03/239881-07$15.00/0