Helga Kolb

1. General characteristics.

Amacrine cells of the vertebrate retina are interneurons that interact at the second synaptic level of the vertically direct pathways consisting of the photoreceptor-bipolar-ganglion cell chain. They are synaptically active in the inner plexiform layer (IPL) and serve to integrate, modulate and interpose a temporal domain to the visual message presented to the ganglion cell. Amacrine cells are so named because they are nerve cells thought to lack an axon (Cajal, 1892). Today we know that certain large field amacrine cells of the vertebrate retina can have long axon-like processes which probably function as true axons in the sense that they are output fibers of the cell (see later section on dopaminergic amacrine cells). However these amacrine axons remain within the retina and do not leave the retina in the optic nerve as do the ganglion cell axons. Figure 1 shows one of the earliest depictions of the retinal cell types including amacrine cells drawn by Ramon y Cajal (circa 1890). These retinal cell types were visualized using the anatomical silver impregnation method devised by the Italian anatomist Camillo Golgi in the nineteenth century (Fig. 2).

Fig. 1. Drawing of the retina made by Cajal

Since the time of Cajal we have known that amacrine cells come in all shapes, sizes and stratification patterns. Since those days many more morphological subtypes have and continue to be described from further Golgi studies, intracellular recordings and immunocytochemical staining. Thus, we presently have a classification of amacrine cells consisting of about 40 different morphological subtypes.

Fig. 2. Picture of Camillo Golgi

It is useful and most easily understandable to group the amacrine cell types into the general descriptors of narrow-field (30-150 um), small-field (150-300 um), medium-field (300-500 um) and wide-field (>500 um) based on a measurement of their dendritic field diameters (Kolb et al., 1981). Then the next most important criterion of classification involves knowing the cells stratification. It is generally agreed now that the IPL can be subdivided into five equi-thickness strata or sublayers (Cajal, 1892) into which amacrine, bipolar and ganglion cell processes can be assigned. All of these cell types are now classified primarily on the stratum or strata of the IPL in which their dendrites or axons are located. This is because, as mentioned in previous chapters, the IPL of vertebrate retinas can be divided up into areas of neuropil where specific cells are put into synaptic contacts and form circuits only with cells earmarked for a particular functional role.

Many varieties of amacrine cell are monostratified, restricted to a single stratum, while others are bi- or tri-stratified. When amacrine or ganglion cell processes pass through all the strata of the IPL from distal to proximal or vice versa, they are called diffuse cells. Superimposed upon Cajals five strata subdivision of the IPL, is a sublaminar division of the IPL. The first two strata, 1-2, are known as sublamina a of the IPL while strata 3-5 are known as sublamina b by this scheme (Famiglietti and Kolb, 1976). It will be remembered from previous chapters that sublamina a contains bipolar axons and ganglion cell connections that lead to OFF-center ganglion cell physiology, while sublamina b contains bipolar to ganglion cell connections resulting in ON-center ganglion cell physiology (Nelson et al., 1978).

Figure 3 shows drawings of some small field amacrine cells of the monkey retina as seen in vertical sections (Polyak, 1941). Small-field cells like these can be well visualized in section because their dendritic trees are contained within the depth of the section. However, large field cells are not so well described in section where their dendrites get cut off.

Fig. 3. Amacrine cells of the monkey retina. Adapted from Polyak, 1941.

It was only when wholemount preparations, from Golgi staining (Stell and Witkovsky, 1972; Boycott and Kolb, 1973) or immunocytochemical staining (Karten and Brecha, 1980) were attempted that we could classify such cells. Then the full extent of their dendritic trees which can be up to one millimeter in spread could be visualized (Fig. 4) and a whole new understanding of amacrine cells became available.

Fig.4. Amacrine cells as seen in wholemountcat retina

A new technique of intracellular staining by a photochemical method has been developed in Richard Maslands group, as an alternative to the unreliable Golgi technique (MacNeil and Masland, 1998). Amacrine cells of the rabbit retina are labelled with the nuclear stain DAPI and then selected single nuclei are irradiated by a narrow beam of light to drive DAPI to the oxidation of non-fluorescent dihydrorhodamine 123 to the fluorescent rhodamine 123. The complete cell body and the dendritic tree is thus revealed under viewing in the fluorescence or confocal microscopes. By this method, 30 or so different varieties of amacrine cell can be photographed and drawn in full detail in the rabbit retina. 22 varieties of amacrine cell have been seen in Golgi preparations in cat and primate retinas so either some have been missed that were seen in rabbit, or else they are not as well deveoped in these less complex mammalian retinas. In any event the narrow field and medium field types revealed by MacNeil and Maslands work (1998) are shown in Figures 4a and b below. A further 5 different wide field monostratified types were also encountered in rabbit retina by this method (not illustrated). They correspond closely to the wide field types seen in monkey, cat and human (Fig. 4, above) (Mariani, 1990; Kolb et al., 1981, Kolb et al. 1992).

2. Amacrine cell circuitry as revealed by electron microscopy.

Kidd (1962) and later Dowling and Boycott (1966) were the first to identify the three types of profile that contribute to the IPL by electron microscopy. The electron micrograph (Fig. 5) below shows the cytological criteria on which we now recognize bipolar, amacrine and ganglion cell profiles in the neuropil. Thus bipolar cell axonal endings are recognized by being filled with synaptic vesicles and having a ribbon-shaped density (Fig. 5, red spots) pointing to two postsynaptic profiles (amacrine and ganglion). Amacrine profiles are also filled with synaptic vesicles but make synapses characterized by membrane densities at which the vesicles are particularly clustered (Fig. 5, yellow spots). Ganglion cell profiles are recognized as being only postsynaptic to either bipolar axons or amacrine processes, containing no vesicles but instead a content of neurotubules, ribosomes and glycogen granules.

Fig. 5. Electron micrograph of several profiles in the IPL

Amacrine cell synapses are frequently seen to be reciprocal to bipolar ribbon input, i.e. the amacrine returns a synapse in the vicinity of the ribbon input synapse (arrowheads). Most amacrine cells are inhibitory neurons in the vertebrate retina, containing the common inhibitory neurotransmitters GABA or glycine. GABAergic amacrine cells, in particular, typically make reciprocal synapses with bipolar cells. A17 is the most well studied of the GABAergic reciprocal amacrine cells in the retina and we shall return to this cell later.

We have learned much concerning the synaptic relationships of certain narrow-field amacrine cells as well as bipolar and small ganglion cell types such as midget ganglion cells of the primate retina, from reconstructions of serial-section electron micrographs. The circuitry of the AII amacrine cell in the cat retina was first appreciated by this means (Kolb and Famiglietti, 1974; Famiglietti and Kolb, 1975; Kolb, 1979). However, with the advent of intracellular dye injection of electron-dense materials (horseradish peroxidase, HRP, or the photoreduction of Lucifer yellow) after physiological recordings or the development of electron dense immunostains for electron microscopy, neurocircuitry was made easier for us. We could look at amacrine cells and their synaptic inputs by study of fewer sections and it was not as critical to photograph every single section in a series. The amacrine cell of interest would always be clearly marked black, and easily found in the synaptic neuropil. It is from this technique that we have learned most about amacrine cells and their circuitry in the mammalian retina. The remainder of this chapter will describe the morphology, circuitry and intracellular responses of the amacrine cells that are most completely understood at present.

3. A2: narrow-field, cone pathway amacrine cell.

A2 is a narrow-field amacrine with a 20-60 um wide dendritic tree composed of multibranched, beaded and appendage-bearing dendrites mostly confined to stratum 2 of the IPL (Fig.6).

Fig. 6. Golgi drawings of A2 amacrine cellsin cat and human retinas

Intracellular recordings from A2 cells (formerly called A4) indicate that these cells give true slow potential hyperpolarizing response to light (OFF-center) at all positions of the slit in their receptive fields and they have no sign of an inhibitory surround (Fig. 7) (Kolb and Nelson, 1984).

A2 cells receive bipolar input from OFF-center types of cone bipolar cell of sublamina a and make reciprocal synapses to these bipolar axons (Fig. 7). A2 amacrine cells then synapse upon OFF-center ganglion cell dendrites of sublamina a. The A2 cell makes an inhibitory synapses upon these ganglion cells, because it is thought to be a glycinergic cell type (Wassle et al., 2009 ).

A possible role for A2 amacrines is in disinhibition of the ganglion cells center responses. Alternatively, A2 cells, despite being small-field types, might have a role in the generation of antagonistic surrounds of ganglion cells (Kolb and Nelson, 1993). A2 cells receive a great many amacrine inputs to their dendritic trees which could be from wider field cells than they are are themselves so giving them a much larger receptive field size than their actual dendritic tree size would indicate.

Fig. 7. Summary diagram of A2 amacrine cells wiring pattern and physiological responses to light

A3, knotty Type 2 amacrine cells.

A small-field amacrine cell that branches in S2 and S3 (i.e. across the sublamina a/b border) is seen in cat and human retina with Golgi staining (Kolb et al., 1981; Kolb et al., 1992) (Fig. 7a). The same amacrine has been called a knotty Type2 amacrine cell in Golgi studies (Mariani, 1990) and immunostaining in the macaque retina (Klump et al., 2009). The A3 cell is clearly immunostained with parvalbumin (Fig. 7b) and has the typical A3 morphology and small field multibranched dendrites with large varicosities (Fig. 7a, b) (Klump et al., 2009). The varicose dendrites branch broadly through strata S2 and S3, so they are in a position to interact with both flat midget cone bipolars or other narrow field cone bipolars of the OFF sublamina a, and with cone bipolar cells of stratum 3 in the ON sublamina b.

Fig. 7a. A3 or knotty type 2 amacrines are seen in whole mount Golgi staining (left) and their physiological response to light is an ON center response (right) (adapted from Klump et al., 2009).

Fig. 7b. Parvalbumin-mmunostained A3 or knotty type2 cells and their neurotransmitter and connectivity. a) An isolated PARV+ amacrine is a small field amacrine cell that branches primarily in S2 and S3 of the IPL (arrow heads show the IPL borders to the inner nuclear layer (INL) and to the ganglion cell layer, GCL). b-d) Shows that the A3 cell colocalizes PARV+ (b) (green) and the glycine transporter Glyt-1 (c) (red) (d) colocalization in yellow). e) The A3 amacrine (green) makes synaptic contact (f) with recoverin-IR flat midget bipolar cells (red) in S2 of the IPL.

A3 or the PARV+ amacrine can be seen to be glycinergic when double immunostained for either glycine (Wassle et al., 2009) or the glycine transporter Glyt-1 (Fig. 7b, the PARV+, A3 colocalizes parvalbumin and Glyt-1, a-d). This small field A3 amacrine has been intracellulaly recorded from by Klump and co-authors (2009) and proves to give an ON response to light stimulation (Fig. 7a, right traces). The PARV+ amacrine is situated amongst the axon terminals of flat (OFF) midget bipolar cells in S2 of the IPL and appears to be either pre or postsynaptic where apposing immunostained profiles occur (Fig. 7b, e and f). The A3 PARV+ cell is also know to be presynaptic to OFF parasol ganglion cells in sublamina a and interact with AII amacrine lobular appendages and starburst amacrine cells of sublamina a (Bordt et al., 2006). A3 cells are also extensively coupled into a network of the same cell type by gap junctions (Klump et al., 2009). The latter authors suggest that A3, knotty Type2 amacrines are driven by ON pathway cone bipolars and inhibit OFF pathways and through synapses upon AII amacrine cells inhibit the transmission of rod signals to these same OFF pathways.

4. AII: a bistratified rod amacrine cell.

Fig. 8. AII amacrine cells intracellularly stained after physiological recording by different methods

Above are shown four examples of the best studied amacrine of all in the vertebrate retina: the AII rod amacrine of the mammalian retina. These cells have been recorded from by microelectrodes and dyes have been iontophoresed into the cell after the intracellular recordings (Nelson, 1982). The AII cell, was first described from Golgi staining and electron microscopic examination (Famiglietti and Kolb, 1975; Kolb and Famiglietti, 1974).

AII is a narrow field amacrine (dendritic tree diameter typically 30-70 um) with a bistratified morphology: the mitral shaped cell body gives off a single, stout apical dendrite and a cluster of lobular appendages (round blobs just below the cell body, Fig. 9) arise from the main dendrite in sublamina a of the IPL (Fig 9). The finer arboreal dendrites (Vaney et al., 1991) penetrate down into sublamina b to end close to the ganglion cell layer (Fig. 9). AII amacrine cells are glycine-immunoreactive (Pourcho and Goebel, 1985; Crooks and Kolb, 1992) and contain the calcium binding proteins parvalbumin, calbindin and calretinin (Wssle et al., 1995). Figure 10 shows a Golgi stained AII amacrine cell in cat and human retinas as seen in a surface view of a wholemount.

Fig. 9. Parvalbumin staining of AII amacrine cells in hamster retina.

Fig. 10 Golgi stained AII amacrine cells seen in wholemount.

In cat and rabbit retinas where AIIs have been recorded from, the AII cell is a rod-dominated depolarizing (ON-center) cell (Bloomfield, 1992; Dacheux and Raviola, 1986; Nelson, 1982). Thus, in the center of its receptive field the cell gives a transient depolarizing response with a pronounced sustained plateau (ON-center) and a long drawn out hyperpolarization after light off (Fig. 11). By 140 um to either side of the center, the response to a light flash is now an inverted response indicating a hyperpolarizing surround (OFF-surround) (Fig. 11) (Nelson, 1982).

Fig. 11. Schematic diagram of the morphology, physiology and wiring pattern of the AII amacrine cell

Electron microscopy has shown that the AII, is primarily postsynaptic to rod bipolar axon terminals in lower sublamina b of the IPL (30% of its input, Strettoi et al., 1992) (Fig. 12, left). Some OFF cone bipolar input is directed at the AIIs lobular appendages in sublamina a (Fig. 13). AIIs major output is upon ganglion cells that have dendrites only in sublamina a. AII cell lobular appendages synapse upon OFF-center ganglion cells (Fig. 12, right) and to OFF-center cone bipolar cell axons (possibly cb1 and cb2 types) in sublamina a (Fig. 13) (Kolb, 1979).

Fig. 12. Electron micrographs of AII amacrine synapses

The AII also passes rod-driven information through the ON-center cone bipolar axons in sublamina b to ON-center ganglion cells by means of gap junctions (Fig. 12, center panel) (Kolb and Famiglietti, 1974; Famiglietti and Kolb, 1975; Kolb, 1979). Several, if not all, cone bipolar axons of sublamina b have gap junctions with AII cell dendrites. A new finding is that even the blue-cone bipolar receives rod signals through this gap junction pathway (Field et al., 2009). AII cells also join with other AII cells by gap junctions in sublamina b of the IPL (Fig. 13, lowest gj) (Famiglietti and Kolb, 1975; Nelson, 1982; Vaney, 1994a). Summary Figure 13 shows the major input and output circuitry of the AII amacrine cell.

Thus, AII cells are almost totally rod-driven by the rod bipolar input in sublamina b of the IPL. However some cone pathway bipolar cell input occurs to their ON-center responses. This could come from from excitatory input from ON-center cone bipolars at the gap junctions (gap junction transmitting both ways), from direct OFF center cone bipolar axon synapses in sublmina a, unlikely, or from other intermediary ON amacrine cells of the cone system (such as A8 and A13, knotty parvalbumin-IR amacrines; Klump et al., 2009).

Fig. 13. Schematic drawing of the wiring pattern of the AII amacrine cell

Click here to see an animation of the wiring pattern of the AII amacrine cells (Quicktime movie)

Several amacrine synapses are seen throughout the AII cells dendritic tree by electron microscopy (Kolb, 1979; Strettoi et al., 1992, Fig. 13, red as). Most of these are unidentified as yet. However, a dopaminergic amacrine cell provides a considerable number of synapses to the AII cell, either directly upon its cell body or upon its lobular appendages (Voigt and Wssle, 1987; Kolb et al, 1991) (see later section on dopamine amacrine cells, A18). Dopamine cells are thought to have a function in the inner retina to uncouple AII amacrine cells from both their contacts with the depolarizing cone bipolar and the AII amacrine coupled network (Daw et al., 1990; Vaney, 1994a). As much as 51% of the input to AII amacrine cells is from various other amacrine cells though, and most of these inputs occur in the central part of the cells dendritic tree in strata 3-4 (Strettoi et al., 1992).

Thus the AII amacrine cells are the major carriers of rod signals to the ganglion cells in the retina. As such they play a role in speeding up the slow potential rod messages for presentation to ganglion cells (Nelson, 1982; Smith, 1994). Their distribution in the retina suggests that they tile the complete retina (Vaney, 1990). AII amacrine cells peak in density at 1.5 mm from the foveal center in monkey and at the area centralis in cat (Vaney, 1984). In addition, because of their high density across all parts of the retina and their synaptic involvement with millions of rod bipolar cells, they may contribute in a major way to the pattern ERG (Zrenner, 1990).

5. A8: a bistratified cone amacrine cell.

A8 is a bistratified, narrow-field amacrine cell which is easy to confuse with AII in wholemount, stained retina (Fig. 14). It actually looks like an upside-down AII cell. A8 has short, wispy processes coming from the apical dendrite to ramify in sublamina a of the IPL whereas heavy beaded dendrites penetrate down to sublamina b, to run in strata 4 and 5. This cell type may correspond to the DAPI-3 of the rabbit retina, described by Vaney (1990) and Bloomfield (1992). It is a glycinergic amacrine cell (Pourcho and Goebel, 1985; Crooks and Kolb,1992; Wassle et al., 2009, Neumann and Haverkamp, 2012).Very recently the A8 amacrine cell population has been demonstrated to immunostain for the protein synaptotagmin-2 (Syt2) in macaque monkey retina (Neumann and Haverkamp, 2012). Syt2 cells peak at a density of 1400/mm2 at 1-2 mm and subside to 22/mm2 at 9-10 mm of eccentricity (Neumann and Haverkamp, 2012).

Fig. 14. Golgi and HRP appearance of A8 amacrine cells

The A8 cell has been intracellularly recorded and studied by electron microscopy after iontophoresis of horseradish peroxidase. In Fig. 15 we can see the synapses of this cell type.

Fig.15. Synapses in the IPL of a HR-peroxidase injected A8 cell

The A8 amacrine cell is involved in the cone pathways of the cat retina, rather than the rod pathways, that the AII is committed to. Thus, in sublamina a, excitatory cone driven signals come from cone bipolar cells like cb2 which we know are OFF center in physiology (Fig. 15a), and in sublamina b from cb6, another OFF-center bipolar cell (Fig. 15d) (Nelson and Kolb, 1983). Altogether cone bipolar synapses account for 42% of the input to A8 cells. Lesser rod bipolar input (20%) also occurs to the lower dendrites in sublamina b of the IPL. Like AII amacrine cells, A8 cells also engages in gap junctions with a cone bipolar type of sublamina b, but the bipolar is a different type, possibly cb6, and in addition to the gap junction makes the common ribbon synapse to A8 dendrites (Figs. 15c and d). A8s major output is to OFF-center beta ganglion cell dendrites in sublamina a of the IPL (Fig. 15b). We have not seen A8 synapses to OFF-center alpha cells in the cat retina (Kolb and Nelson, 1996). The input and output synapses for A8 amacrine cells are seen in the summary diagram of Figure 16 and the animation below.

View original post here:
Roles of Amacrine Cells by Helga Kolb Webvision