Modeling Circadian Phototransduction: Retinal Neurophysiology and Neuroanatomy

Abstract

The retina is a complex, but well-organized neural structure that converts optical radiation into neural signals that convey photic information to a wide variety of brain structures. The present paper is concerned with the neural circuits underlying phototransduction for the central pacemaker of the human circadian system. The proposed neural framework adheres to orthodox retinal neuroanatomy and neurophysiology. Several postulated mechanisms are also offered to account for the high threshold and for the subadditive response to polychromatic light exhibited by the human circadian phototransduction circuit. A companion paper, modeling circadian phototransduction: Quantitative predictions of psychophysical data, provides a computational model for predicting psychophysical data associated with nocturnal melatonin suppression while staying within the constraints of the neurophysiology and neuroanatomy offered here.

Keywords: circadian phototransduction; photic sub-additivity; retinal neuroanatomy; retinal neurophysiology; shunting inhibition.

FIGURE 1

Retinal circuit diagram illustrating the revised model of circadian phototransduction. On the left side of the figure are the conventional labels for the different layers of the retina: OS, outer segment of the rod (R) and cone [L (long-wavelength sensitive), M (middle-wavelength sensitive), and S (short-wavelength sensitive)] photoreceptors; ONL, outer nuclear layer containing the cell bodies of the rod and cone photoreceptors and two types of horizontal cells (H1 and H2); OPL, outer plexiform layer containing the distal plexus of the photoreceptor (efferent, orange lines) axons and the (afferent, black lines) dendrites of the horizontal and bipolar neurons; INL, inner nuclear layer containing the cell bodies of the rod bipolar (RB) and two cone bipolar neurons, one achromatic cone bipolar (CB) and one S-cone bipolar (SB) neuron and amacrine neurons (AII, A17, A18); IPL, inner plexiform layer containing the plexus of axons and dendrites of the bipolar, amacrine, and ganglion neurons, which is divided into OFF and ON sublayers; GCL, ganglion cell layer containing the cell bodies of the conventional retinal ganglion cells (RGCs) and the intrinsically photosensitive retinal ganglion cells (ipRGCs) as well as a displaced S-cone amacrine (SCA). At the bottom of the figure are the two targets for ganglion cell axons, the RGC axons reaching the lateral geniculate nucleus (LGN) and the ipRGC axons reaching the suprachiasmatic nucleus (SCN). Also shown in the diagram are blue circles (Processes A–C) that represent important processes in the revised model. Process A represents cone inhibition of rods and thereby a reduction in shunting inhibition of the ipRGC by AII amacrine cells in process B. Process B also includes decoupling of shunting inhibition of AII amacrine cells via the A18 amacrine cells when the spectrally opponent SB signals “yellow.” Process C represents the en passant complex involving the SB, ipRGC, and A18 neurons.

FIGURE 2

Model predictions of the spectral sensitivity of the circadian phototransduction circuit when exposed to either monochromatic (narrow-band) or polychromatic light sources at 300 scotopic lx at the eye. The spectral sensitivities to monochromatic spectral lights from two studies (closed diamonds Thapan et al., 2001; open circles Brainard et al., 2001) are shown together with the revised model predictions for monochromatic sources (black dashed line) and for polychromatic light sources where the b – y channel signals “blue” (blue solid line) or “yellow” (red dot/dash line). The cross-hatched area represents enhanced spectral sensitivity of the circadian phototransduction circuit to very short wavelengths (< 470 nm) when the SB signals “blue” and can provide added input to the ipRGC. The solid gray area represents the area of reduced spectral sensitivity, relative to that from the ipRGC alone (470–500 nm). This solid gray area of transition to longer wavelengths is due to a systematic loss of S-cone inhibition of rods as the b – y channel comes closer to its spectral cross-point at approximately 500 nm. As the inhibitory SB input to the AII amacrine is reduced (Process A in Figure 1), rod shunting inhibition of the ipRGC increases (Process B in Figure 1). For wavelengths longer than about 500 nm, the SB signals “yellow” OFF, and through the A18 amacrine neuron (Process C in Figure 1) decouples the shunting inhibition of the ipRGC altogether. It should be noted that the light level of 300 scotopic lx was chosen for illustration because rod inhibition is modeled to control threshold for ipRGC activation, thereby providing a common basis for comparing different spectral sensitivity functions and because, empirically, this light level is approximately equal to that producing the half-saturation response levels of nocturnal melatonin suppression. As described in the original model, a progressively smaller “notch” is predicted as light levels increase because cone inhibition of rods becomes relatively larger.