Retinal Bioengineering

Abstract

The retina is a small piece of the central nervous system responsible for the first steps in vision, so understanding how it works has great importance for daily life. In addition, features of the retina make it attractive as a model neural system. The only input to the retina is light, which can be easily manipulated, and recordings have been made for many decades from individual output cells of the retina, the retinal ganglion cells (RGCs), allowing application of linear (and to some extent nonlinear) systems analysis methods that define the transfer functions of the retina. The small, non-spiking photoreceptors and retinal interneurons make recordings from these earlier stages difficult in mammals, but this has been partially surmounted by the ability to record massed activity of some types of retinal neurons, including photoreceptors and bipolar cells, in the electroretinogram (ERG) in humans as well as animals. ERG analyses have led to models of signal processing prior to the RGCs. Engineering methods in combination with physiology have thus elucidated the basic features of the retinal network that allow the convergence of signals from many millions of photoreceptors to yield the center-surround organization and response properties of the primary types of RGCs in cats and primates. However, some of the approximately 20 types of RGCs that send parallel signals to the brain are still poorly understood. Recent work has used isolated retinas and multielectrode arrays to record from many retinal ganglion cells simultaneously. Specific contributions of interneurons to the retinal circuits have also been addressed with new methods, some of which are reviewed here. Another aspect of retinal bioengineering concerns the retinal microenvironment. Diffusion models and spatially precise intraretinal measurements of oxygen and pH provide information about retinal metabolism that is useful in understanding dysfunction of the retina in some diseases.

Homework

Fig. 21.32

• 1. Calculate the relative g_Na/g_K (or P_Na/P_K if you prefer) for a photoreceptor whose resting potential in the dark is −30 mV. Make reasonable assumptions for E_Na and E_K (or Na⁺ and K⁺ concentrations) and assume that Cl⁻ is at equilibrium. How does this differ from most neurons at rest?

• 2. The dark current of photoreceptors is about −30 pA. Assume all the current is carried by Na⁺. All the Na⁺ has to be pumped out of the inner segment (IS) to maintain the normally low intracellular [Na⁺]_i. The pump exchanges 3 Na⁺ for 2 K⁺ as usual, and each pump cycle (i.e., 3 Na⁺) requires one molecule of ATP.

  • (a) What is the usage of ATP/min in the dark for an individual rod? (This is not the only function requiring ATP but it is by far the largest in the dark-adapted retina. Actually about 85% of the current is due to Na⁺, and 15% is due to Ca⁺², but Ca⁺² is pumped out by a secondary active transporter that moves Ca⁺² out and Na⁺ in in the outer segment, which makes the load of Na⁺ higher than assumed in the problem statement.)

  • (b) There are 180,000 rods per mm² at the peak of rod density. The IS are about 25 μm long. As noted in the text, other layers of the outer retina use no oxygen, so this ATP usage is over a volume of 1 mm² × 25 μm. Roughly what is the oxygen usage of the IS, in μM-ml⁻¹-min⁻¹ of IS volume (essentially per gram since tissue density is about 1.05 g/ml), if all of the metabolism is oxidative metabolism (1 glucose + 6 O₂ → 6 CO₂ + 6 H₂O). Also assume that 36 moles of ATP are produced per mole of glucose. (After you do the calculation, you will be able to compare this with the typical oxygen consumption of the brain, which is around 2 μmoles O₂-ml⁻¹-min⁻¹ or as it is often expressed, 4 ml O₂-100g⁻¹-min⁻¹.)

• 3. The ganglion cell center and surround are usually viewed as being antagonistic to each other, but this is actually true only for certain stimulus conditions, as in Fig. 21.18. Under what conditions do the center and surround of ganglion cells add rather than subtract? Justify your answer.

• 4. Gauthier et al. [38] hypothesized that the receptive fields of primate retinal ganglion cells were arranged to tile the retina (or visual world). They suggested that the interdigitation of adjacent receptive fields was not random but was nearly optimal, with minimal gaps between ganglion cells and minimum overlap of receptive fields. Using RF data like those shown in Fig. 21.24, suggest a method to test this hypothesis.

• 5. Figure 21.14 shows that the small ERG signals that comprise the multifocal ERG vary in amplitude across the visual field. In fact, the stimulus elements are not equal in size, and the smaller elements in the middle of the stimulus array (left) produce the largest responses (right). Generate at least one testable hypothesis about why this might be true, recognizing that the ERG comes largely from photoreceptors (here cones) and bipolar cells.

• 6. The chapter shows difference of Gaussian receptive field profiles for selected cat retinal ganglion cells, but as noted in Fig. 21.19, primate retinal ganglion cells can be characterized in the same way. Receptive fields vary a great deal across the retina.

  • (a) For the P (midget) cells with the smallest and largest receptive field centers, plot the sensitivity of center and surround of the receptive field in spatial coordinates, as in Fig. 21.18a. For the larger P cell, also show the surround sensitivity multiplied by 10. The cells in Fig. 21.19 were recorded between about 1 and 35 degrees of eccentricity. For convenience, the centers and surrounds of P cells from Croner and Kaplan [218] are shown separately below.

  • (b) In response to a large (or diffuse) stimulus, both center and surround will be maximally activated. The area under the center curve represents this “integrated center strength” and is K_cr_c ². The integrated surround strength is K_sr_s ². For these two cells, compare the integrated center strengths. Also, what is the strength of the surround relative to the strength of the center? From your graphs of the center and surround, the answers to these questions may surprise you, but they seem to reveal a logic about the way ganglion cell receptive fields vary with eccentricity.

  • (c) What is the highest spatial frequency that each of these cells can detect? (In the units of the figures and the equation in the text, this is where contrast sensitivity falls to 0.01, meaning that 100% contrast is needed.)

• 7. Which eye diseases could a retinal prosthesis be used to treat and why? What is the definition of legal blindness in the USA in terms of visual acuity? No currently available retinal prosthesis has succeeded in providing this minimal level of acuity. Why do you think that this is the case and what has limited our ability to reach this standard?

• 8. Barlow and Levick in Fig. 7 of their 1965 paper “The mechanisms of directionally selective units in the rabbit’s retina” (Journal of Physiology 178, 477–504) proposed a model for the receptive field of a rabbit retinal ganglion cell that has directional selectivity. It is known now that retinal ganglion cells with similar receptive field properties exist in most, if not all, vertebrate retinas, including those of the primate. Suggest a model for the creation of directional selectivity based on retinal circuitry involving bipolar and amacrine cells.