ScienceWeek

NEUROBIOLOGY: ON RETINA-CORTEX WIRING

The following points are made by A.M. Derrington and B.S. Webb (Current Biology 2004 14:R14):

1) Forty years ago Hubel and Wiesel [1] elucidated our understanding of the visual cortex with the discovery that the receptive fields of the visual cortex are made up of elongated mutually antagonistic zones. A single retinal output neuron transmits to primary visual cortex through multiple pathways with different strengths, and these pathways converge on a single cortical neuron.[1,2].

2) Cortical receptive fields, with their elongated subregions, are very different from the circular receptive fields found earlier in the visual pathway. The antagonistic subfields of retinal and lateral geniculate receptive fields are concentric, with a center -- within which light can be excitatory or inhibitory -- and an opposed surround. This raises the question: how is a cortical receptive field constructed from its input? There are two possibilities. First, the cortical receptive fields are simply cortical manifestations of the centers and surrounds of the receptive fields of retinal neurons. Second, the subregions of a cortical cell's receptive field are formed from scratch in the cortex, by combining inputs from separate sets of receptive fields whose centers have opposite polarities.

3) Although in principle this question can be addressed indirectly by comparing measurements of receptive field transfer functions at different levels of the pathway [3], it is very difficult to attack it head on. Untangling the kilometers of wiring in every cubic millimeter of cortex makes it very difficult to trace the connections anatomically. However, the question has now yielded to a combination of neurophysiology and statistics. Kara and Reid [4] have shown, in a series of physiological experiments, that cortical subfields are formed from the centers of retinal receptive fields.

4) The experiment sounds straightforward enough. It involves recording simultaneously from a simple cell in layer IV of the cortex and from the retinal ganglion cells that provide the input to the cortical receptive field. Retino-cortical connections can be identified by cross-correlating the streams of action potential spikes from paired retinal and cortical recordings. Extra spikes in a cortical cell that occur a fixed time after each spike in a retinal cell indicate that there is a direct disynaptic connection between those retinal and cortical cells. This is actually very difficult to do. Fortunately, the difficulties in the experiment highlight an interesting aspect of the design of the pathway from retina to the primary visual area V1.

5) If the divergent pathways through the lateral geniculate nucleus converge in the cortex, there could be strong connections between individual retinal output neurons and cortical cells. In principle, reconvergence in the cortex would allow the divergence between retina and the lateral geniculate nucleus to act as an amplifier at the level of the cortex, by providing several synchronous geniculate spikes arriving in the cortex from a single spike leaving the retina [5]. On the other hand, if the different geniculate targets of a given retinal neuron project to different cortical targets, the links between retina and cortex could be undetectable.(5)

References (abridged):

1. Hubel, D.H. and Wiesel, T.N. (1962). Receptive fields, binocular interaction and functional architecture in cat's visual cortex. J. Physiol. (Lond) 160, 106-154

2. Hubel, D.H. and Wiesel, T.N. (1968). Receptive fields and functional architecture of monkey striate cortex. J. Physiol. (Lond) 195, 215-243

3. Cooper G.F., and Robson J.G. (1968). Successive transformations of spatial information in the visual system. Conference on Pattern Recognition. I.E.E./N.P.L, Conference No. 47

4. Kara, P. and Reid, R.C. (2003). Efficacy of retinal spikes in driving cortical responses. J. Neurosci. 23, 8547-8557

5. Usrey, W.M., Reppas, J.B., and Reid, R.C. (1998). Paired-spike interactions and synaptic efficacy of retinal inputs to the thalamus. Nature 395, 384-387

Current Biology http://www.current-biology.com

--------------------------------

Related Material:

NEUROBIOLOGY: ON RAPID EVENTS IN THE RETINA

The following points are made by Kendall J. Blumer (Nature 2004 427:20):

1) Imagine walking out of a dark theater into a bright and sunny Sunday afternoon. You are momentarily blinded, but your eyes rapidly adjust to the change and you continue on your way. For some people with a rare visual defect, however, this momentary blindness can last for up to ten seconds. A similar, but potentially more dangerous, prolonged blindness occurs when these individuals drive from daylight into a darkened tunnel. Moreover, people with this problem also suffer from difficulties in seeing certain moving objects (such as balls thrown during a sporting event). Nishiguchi et al(1) have described a genetic cause of this condition. In so doing, they have revealed that visual perception requires rapid deactivation of the light-stimulated responses shown by neurons in the eye.

2) Light streaming into the eye is detected by specialized neurons (photoreceptors) in the retina. In response to light, a coordinated series of molecular events -- the so-called phototransduction cascade -- is triggered in these cells(2). Photons excite pigment-containing proteins called rhodopsins, which then switch on the protein transducin by loading it with the small molecule guanosine triphosphate (GTP). When bound to GTP, transducin turns on a phosphodiesterase, an enzyme that breaks down cyclic guanosine monophosphate (cGMP -- another small molecule). High concentrations of cGMP open specialized ion channels in the outer cell membrane. Thus, by reducing the concentration of cGMP, light changes the flow of ions across the membrane of photoreceptive neurons, producing an electrical signal that is necessary for communicating with the brain.

3) Once this light-activated switch is on, how do cells turn it off? One mechanism is to limit the amount of time that GTP-bound transducin can keep the phosphodiesterase enzyme active. Transducin can accomplish this task itself by converting --hydrolyzing -- its bound GTP molecule into guanosine diphosphate, GDP. (This conversion from GTP to GDP is a commonly used molecular "switch" in a variety of cellular signalling pathways.) Because transducin bound to GDP has a low affinity for phosphodiesterase, it releases the enzyme in an inactive form, allowing cGMP levels to rise again and return the flow of ions across the cell membrane to the "dark" state. In this molecular cascade, then, the conversion of GTP to GDP by transducin is the rate-limiting step that defines the amount of time for which a photoreceptor responds to a light pulse.

4) But this presents a problem. Photoreceptor cells can turn off in less than a second in response to a brief flash of light(2). In contrast, the hydrolysis of GTP by transducin requires tens of seconds to complete, making it difficult to understand how such a mechanism could account for the rapid turn-off of photoreceptor cells. To get around this problem, photoreceptor cells possess a protein called regulator of G-protein signalling-9 (RGS9) that accelerates transducin's ability to hydrolyze GTP3. Indeed, mice that lack the RGS9 gene exhibit slow photoreceptor deactivation(4,5).

References (abridged):

1. Nishiguchi, K. M. et al. Nature 427, 75-78 (2004)

2. Arshavsky, V. Y., Lamb, T. D. & Pugh, E. N. Jr Annu. Rev. Physiol. 64, 153-187 (2002)

3. He, W., Cowan, C. W. & Wensel, T. G. Neuron 20, 95-102 (1998)

4. Chen, C. K. et al. Nature 403, 557-560 (2000)

5. Kooijman, A. C., Houtman, A., Damhof, A. & van Engelen, J. P. Doc. Ophthalmol. 78, 245-254 (1991)

Nature http://www.nature.com/nature

--------------------------------

Related Material:

ON NEURAL OPTIMIZATION OF RETINAL INPUTS

The following points are made by Martin Wilson (Current Biology 2002 12:R625):

1) Every neuron in the brain faces the task of discriminating useful incoming signals from a background of useless noise. Much of this noise is inevitable in any miniaturized device and comes from the stochastic behavior of the signaling molecules from which neurons are constructed. In a recent study, Field and Rieke [1] have taken a reverse engineering approach to an early step in vision to examine how this is done; they find, counterintuitively, that to optimize performance much of the signal is thrown out with the noise.

2) Because light arrives as discrete quanta, the problem of seeing in very low light conditions is essentially that of signaling the capture of enough photons for reliable statistical estimates about the relative brightness of different parts of the visible world. Mammals, particularly nocturnal mammals, have retinas with large numbers of rod photoreceptors dedicated to low light vision. Each rod is a very high gain detector with 10 million copies of a light-capturing molecule, rhodopsin, coupled to a multistage biochemical amplifier. When a photon is absorbed by one rhodopsin molecule, activation of the biochemical amplifier results in the closure of cation channels in the plasma membrane, giving rise to a current blip lasting about one second and about one picoamp in amplitude.

2) The consequent small voltage signal is propagated to the axon terminal, where it is passed on to the next layer of cells, the bipolar cells, through three different pathways. "OFF" bipolar cells, which signal a light increase with hyperpolarization, receive rod signals in two ways: first, via a recently discovered direct synaptic connection between rods and OFF bipolar cells; and second, indirectly via rod cone electrical coupling and cone synapses. The most prominent pathway, however, is a direct synaptic connection between rods and a type of ON bipolar cell, the rod bipolar cell, that draws input exclusively from rods and is thought to be part of a special pathway used for vision in the dimmest environments.

3) Dozens or even hundreds of rods are synaptically wired to every rod bipolar cell. This makes engineering sense, as only by pooling the scarce photon signals is it possible for the visual system to make statistically reliable discriminations. Convergence brings a problem, however, as the high gain of rods results in their being noisy, even in darkness. A particularly troublesome kind of noise is a continuous fluctuation in the current stemming from the instability of the phosphodiesterase molecules that form an intermediate stage in the rod's biochemical amplifier [2] . This kind of noise would swamp the signal in a bipolar cell at the very dimmest intensities, where no more than one rod is likely to contribute signal but every rod contributes noise. One trick often used by the nervous system to remove noise is to filter out the temporal frequencies unique to the noise, thereby leaving mostly signal. This strategy seems to be used at the rod-to-rod bipolar cell synapse [3] , but unfortunately the temporal frequencies of signal and noise overlap too much for this to be enough. Nevertheless, as Field and Rieke [1] have shown, noise can be optimally filtered out solely on the basis of the amplitude distributions of signal and noise [4,5].

References (abridged):

1. Field G.D. and Rieke F. (2002) Nonlinear signal transfer from mouse rods to bipolar cells and implications for visual sensitivity. Neuron, 34:773-785.

2. Rieke F. and Baylor D.A. (1996) Molecular origin of continuous dark noise in rod photoreceptors. Biophys. J., 71:2553-2572.

3. Bialek W. and Owen W.G. (1990) Temporal filtering in retinal bipolar cells. Elements of an optimal computation? Biophys. J., 58:1227-1233.

4. Rieke F. and Baylor D.A. (1998) Single-photon detection by rod cells of the retina. Rev. Mod. Phys., 70:1027-1036.

5. Euler T. and Masland R.H. (2000) Light-evoked responses of bipolar cells in a mammalian retina. J. Neurophysiol., 83:1817-1829.

Current Biology http://www.current-biology.com

ScienceWeek http://scienceweek.com