Introduction
Vesicular glutamate transporters (vGluT) are a family of membrane proteins that transport glutamate from the cytoplasm into synaptic vesicles. The first members, vGluT1 and vGluT2, were identified based on their physiological activity. vGluT1 and vGluT2 are expressed by neurons in the brain that utilize glutamate as their primary neurotransmitter. Later, vGluT3 was identified based on the homology of its amino acid sequence with the other members of the family. vGluT3 is less abundant than the other vesicular glutamate transporters, and it is expressed in neurons that utilize other conventional neurotransmitters, including monoamines, acetylcholine, and amino acids (Fremeau et al., Reference Fremeau, Voglmaier, Seal and Edwards2004). In the retina, the distribution of vesicular glutamate transporters is very similar. vGluT1 and vGlu2 are widely expressed by typical excitatory neurons, and vGluT3 is expressed by a small number of neurons that also release another conventional neurotransmitter. Rods, cones, and bipolar cells express vGluT1, and retinal ganglion cells express vGluT2 (Fujiyama et al., Reference Fujiyama, Hioki, Tomioka, Taki, Tamamaki, Nomura, Okamoto and Kaneko2003; Johnson et al., Reference Johnson, Tian, Caywood, Reimer, Edwards and Copenhagen2003). In mammalian retinas, a single type of glycinergic amacrine cell expresses vGluT3 (Fremeau et al., Reference Fremeau, Burman, Qureshi, Tran, Proctor, Johnson, Zhang, Sulzer, Copenhagen, Storm-Mathisen, Reimer, Chaudhry and Edwards2002), and now there is good evidence that the two neurotransmitters are released at distinct synapses. In rat retinas, the vGluT3 cells comprise only 1% of all amacrine cells (Haverkamp & Wassle, Reference Haverkamp and Wassle2004), but several recent electrophysiological studies have shown that they play important roles in visual information processing. Their glycinergic synapses inhibit a subset of cone bipolar cells and ganglion cells, and their glutamatergic synapses excite a subset of GABAergic amacrine cells and some other types of retinal ganglion cells.
Morphology of vGluT3 amacrine cells
The vGluT3 cells are a subtype of narrow-field amacrine cells, which have relatively small dendritic fields and branch in two or more of the five strata that comprise the inner plexiform layer (IPL). They typically use the neurotransmitter glycine and make inhibitory synapses onto bipolar cells, ganglion cells, and other amacrine cells. Several types of narrow-field amacrine cells have dendrites in both the inner strata of the IPL, where bipolar cells and ganglion cells that respond to increments in light intensity ramify, and in the outer strata of the IPL, where bipolar cells that respond to decrements in light intensity typically ramify (Wassle et al., Reference Wassle, Heinze, Ivanova, Majumdar, Weiss, Harvey and Haverkamp2009). The vGluT3 cells are bistratified, having dendrites that ramify in two relatively broad plexuses, one on either side of the center of the IPL. In primates, the plexus in the outer half of the IPL extends from approximately 20–40% of the IPL depth, and the plexus in the inner half extends from approximately 50–80%, with 0% being the inner nuclear layer and 100% being the ganglion cell layer. The neurons that branch in these strata near the center of the IPL typically have relatively transient responses to light and are highly sensitive to moving and other rapidly changing stimuli (Marshak et al., Reference Marshak, Chuang, Dolino, Jacoby, Liu, Long, Sherman, Suh, Vila and Mills2015). In primates, these cells were previously identified based on their distinctive morphology, and they were named knotty bistratified type 1 cells (Mariani, Reference Mariani1990). These amacrine cells were called narrow-field bistratified in mice (Badea & Nathans, Reference Badea and Nathans2004) and AB broad diffuse cells in rabbits (MacNeil et al., Reference MacNeil, Heussy, Dacheux, Raviola and Masland1999).
Synaptic inputs to vGluT3 amacrine cells
The vGluT3 amacrine cells receive excitatory synapses from bipolar cell axon terminals. In an electron microscopic immunolabeling study of rat retina, inputs from cone bipolar cells were seen in both the inner and outer halves of the IPL (Haverkamp & Wassle, Reference Haverkamp and Wassle2004). In electrophysiological studies of vGluT3 cells in mouse retina, glutamatergic, excitatory input is observed in response to both increments and decrements in light intensity, but the responses to decrements are larger (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014; Kim et al., Reference Kim, Soto and Kerschensteiner2015). In a primate retina, on the other hand, inputs from bipolar cells are only detected in the inner half of the IPL, and excitatory inputs would be expected only at light onset (Marshak et al., Reference Marshak, Chuang, Dolino, Jacoby, Liu, Long, Sherman, Suh, Vila and Mills2015).
The majority of inputs to vGluT3 cells are inhibitory synapses from other amacrine cells. These include both GABAergic and glycinergic amacrine cells, but not vGluT3 cells, themselves (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014). In primates, over 99% of the synapses onto vGluT3 cells are from other amacrine cells. ON-starburst cells may be among the presynaptic amacrine cells in S4. The large synapses from amacrine cells onto the primary dendrites of vGluT3 cells suggest that the entire cell might be inhibited under some conditions (Marshak et al., Reference Marshak, Chuang, Dolino, Jacoby, Liu, Long, Sherman, Suh, Vila and Mills2015). This input might originate from the dopaminergic cells, which have ON responses to light and inhibit multiple amacrine cell types via synapses onto their somas and primary dendrites (Debertin et al., Reference Debertin, Kantor, Kovacs-Oller, Balogh, Szabo-Meleg, Orban, Nyitrai and Volgyi2015). In mouse retina, a major inhibitory input to the vGluT3 amacrine cells comes from spiking amacrine cells (Kim et al., Reference Kim, Soto and Kerschensteiner2015). Another presynaptic amacrine cell in mouse retina is the type two tyrosine hydroxylase-positive cell, a nonspiking GABAergic amacrine cell with ON–OFF responses to light (Knop et al., Reference Knop, Feigenspan, Weiler and Dedek2011).
Light responses of vGluT3 amacrine cells
Light responses of identified vGluT3 cells have only been recorded from mouse retinas to date. With full-field stimuli, there is no response or else a small hyperpolarization. Voltage clamp studies under those conditions reveal a transient excitatory input that is amplified by voltage-gated sodium channels in the vGluT3 cells (Grimes et al., Reference Grimes, Seal, Oesch, Edwards and Diamond2011). With a small spot of light in the receptive field center, the vGluT3 cells depolarize transiently at both at the onset and the offset; the OFF response is larger. However, they hyperpolarize at the onset and offset of a larger spot. This results from both a reduction in bipolar cell input to the vGluT3 cells and an enhancement of direct inhibition from amacrine cells (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014). This very strong surround inhibition is consistent with the anatomical studies indicating a preponderance of amacrine cell input to vGluT3 cells. vGluT3 cells are also highly sensitive to movement. They depolarize in response to moving bars and gratings in the receptive field center, but they are inhibited when these stimuli extend into the receptive field surround. This response selectivity depends on input from spiking amacrine cells (Kim et al., Reference Kim, Soto and Kerschensteiner2015).
Neurotransmitters of vGluT3 amacrine cells
Fremeau et al. (Reference Fremeau, Burman, Qureshi, Tran, Proctor, Johnson, Zhang, Sulzer, Copenhagen, Storm-Mathisen, Reimer, Chaudhry and Edwards2002) were the first to propose that vGluT3 cells in the retina released both glutamate and the inhibitory amino acid glycine, and this has now been confirmed by both anatomical and physiological studies. Electron microscopic immunolabeling experiments in baboon retina provided evidence suggesting that vGluT3 cells release glycine at some of their synapses and glutamate at others (Marshak et al., Reference Marshak, Chuang, Dolino, Jacoby, Liu, Long, Sherman, Suh, Vila and Mills2015). At synapses made by vGluT3 cells, some of the postsynaptic densities are relatively thin, as in symmetric synapses. At the other synapses made by vGluT3 cells, the postsynaptic densities were relatively thick, as in asymmetric synapses. These likely correspond to the rare, conventional synapses with E-face particles in the postsynaptic membrane that were observed in the macaque IPL in a study using freeze-fracture (Raviola & Raviola, Reference Raviola and Raviola1982). In the brain, symmetric synapses are known to be inhibitory and asymmetric synapses are known to be excitatory (Harris & Weinberg, Reference Harris and Weinberg2012).
Besides the presence of vGluT3, three lines of evidence indicate that these amacrine cells use glutamate as a neurotransmitter. The vGluT3 amacrine cells contain immunoreactive glutamate (Haverkamp & Wassle, Reference Haverkamp and Wassle2004), their dendrites are closely associated with puncta containing metabotropic glutamate receptor 4 (Johnson et al., Reference Johnson, Sherry, Liu, Fremeau, Seal, Edwards and Copenhagen2004), and they make excitatory glutamatergic synapses onto several types of retinal ganglion cells (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014; Krishnaswamy et al., Reference Krishnaswamy, Yamagata, Duan, Hong and Sanes2015; Tien et al., Reference Tien, Kim and Kerschensteiner2016). There is good evidence that vGluT3 cells also use glycine as a neurotransmitter. They contain immunoreactive glycine (Haverkamp & Wassle, Reference Haverkamp and Wassle2004; Johnson et al., Reference Johnson, Sherry, Liu, Fremeau, Seal, Edwards and Copenhagen2004), puncta containing the α2 subunit of the glycine receptor are found inside the presumed postsynaptic cells at appositions with dendrites of vGluT3 cells (Haverkamp et al., Reference Haverkamp, Muller, Zeilhofer, Harvey and Wassle2004; Jusuf et al., Reference Jusuf, Haverkamp and Grunert2005; Knop et al., Reference Knop, Feigenspan, Weiler and Dedek2011), and they make glycinergic synapses onto mouse retinal ganglion cells whose responses are suppressed by stimulus contrast, also known as uniformity detectors (Lee et al., Reference Lee, Zhang, Chen and Zhou2016; Tien et al., Reference Tien, Kim and Kerschensteiner2016).
Postsynaptic targets of vGluT3 amacrine cells
vGluT3 cells make synapses onto axon terminals of bipolar cells. In primates, vGluT3 cells make synapses onto OFF bipolar cells, including the DB3a type known to be presynaptic to parasol ganglion cells (Marshak et al., Reference Marshak, Chuang, Dolino, Jacoby, Liu, Long, Sherman, Suh, Vila and Mills2015). In mice, the vGluT3 cells make contacts with axons of type 3 and type 5 bipolar cells (Haverkamp & Wassle, Reference Haverkamp and Wassle2004). These two types of bipolar cells typically have transient responses to light (Euler et al., Reference Euler, Haverkamp, Schubert and Baden2014). There is physiological evidence suggesting that the vGluT3 cells release glycine at their synapses onto ON bipolar cell axon terminals in mouse retina (Tien et al., Reference Tien, Kim and Kerschensteiner2016). In primates, approximately half of the output of vGluT3 cells is directed to other amacrine cells, and one of these, the Aii amacrine cell, was identified by light microscopy. The Aii amacrine cells are expected to be postsynaptic at the contacts with vGluT3 cells because Aii cells receive synapses from amacrine cells but do not make chemical synapses onto other amacrine cells (Wassle et al., Reference Wassle, Grunert, Chun and Boycott1995). ON starburst amacrine cells may also receive synapses from vGluT3 cells. Starburst cells are both presynaptic and postsynaptic to amacrine cells in macaque retina (Yamada et al., Reference Yamada, Dmitrieva, Keyser, Lindstrom, Hersh and Marshak2003), and, therefore, it is not possible to predict whether the vGluT3 cells or the starburst cells are presynaptic at their contacts (Marshak et al., Reference Marshak, Chuang, Dolino, Jacoby, Liu, Long, Sherman, Suh, Vila and Mills2015). In mice, there is physiological evidence suggesting that vGluT3 cells release glutamate at their synapses with unidentified GABAergic amacrine cells (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014).
vGluT3 cells also make synapses onto dendrites of retinal ganglion cells (Fig. 1). In primates, approximately half of the synaptic output of vGluT3 cells is onto retinal ganglion cells. These include both ON and OFF subtypes of midget and parasol ganglion cells, and there is indirect evidence that at least some of these synapses are inhibitory (Marshak et al., Reference Marshak, Chuang, Dolino, Jacoby, Liu, Long, Sherman, Suh, Vila and Mills2015). In mice, vGluT3 cells release glutamate at synapses onto several types of retinal ganglion cells that are excited by stimulus contrast and motion. These include: ON direction selective, ON–OFF direction selective, W3B and OFF alpha cells. In general, the effects of the excitatory synapses from vGluT3 cells are to enhance the responses of each type to its characteristic trigger feature (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014; Kim et al., Reference Kim, Soto and Kerschensteiner2015). Excitation from the vGluT3 is particularly important for the OFF component of the responses of W3B cells (Krishnaswamy et al., Reference Krishnaswamy, Yamagata, Duan, Hong and Sanes2015). vGluT3 cells release glycine at synapses onto at least one other type of retinal ganglion cell, the spontaneously active cells whose responses are suppressed by stimulus contrast. These glycinergic synapses are one of four sources of inhibition to these ganglion cells, and they have a distinct contribution, namely inhibition at the offset of small stimuli (Lee et al., Reference Lee, Zhang, Chen and Zhou2016; Tien et al., Reference Tien, Kim and Kerschensteiner2016). The ganglion cells that are suppressed by contrast are bistratified, and they receive the glycinergic synapses from vGluT3 cells on the dendrites connecting their two dendritic arbors, where the inhibition would be particularly effective (Tien et al., Reference Tien, Kim and Kerschensteiner2016).
Contributions of vGluT3 cells to neural circuits in the retina
In mouse retina, the amacrine cells containing vGluT3 have resting membrane potentials that are relatively depolarized, and therefore, they may release their neurotransmitters spontaneously (Grimes et al., Reference Grimes, Seal, Oesch, Edwards and Diamond2011). This spontaneous activity might increase the sensitivity of vGluT3 cells and their postsynaptic targets to changes in illumination.
vGluT3 cells clearly impart selective sensitivity to small stimuli to their postsynaptic targets in the mouse retina. The specificity of the light responses of the vGluT3 cells closely resembles that of the glycinergic inhibitory input recorded from ganglion cells that are suppressed by contrast (Lee et al., Reference Lee, Zhang, Chen and Zhou2016; Tien et al., Reference Tien, Kim and Kerschensteiner2016). The same is true of the mouse retinal ganglion cells that receive glutamatergic excitatory input from vGluT3 cells (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014, Reference Lee, Zhang, Chen and Zhou2016). In mice without vGluT3, excitation generated by selective stimulation in the receptive field center of W3B ganglion cells was greatly reduced, but excitation generated by other stimuli was unaffected (Kim et al., Reference Kim, Soto and Kerschensteiner2015). Because the signal passes through an additional neuron, the major excitatory input to W3B cells is slower than direct excitation via bipolar cells. The combination of rapid inhibition and slow excitation is thought to enhance the selectivity of W3B cells for object motion (Krishnaswamy et al., Reference Krishnaswamy, Yamagata, Duan, Hong and Sanes2015).
Because the vGluT3 cells ramify in both the inner and outer sublaminae of the IPL, they are able to influence neurons with ON and neurons with OFF responses. The OFF responses of vGluT3 cells are the most prominent, and they might provide crossover inhibition to ON cells via their glycinergic synapses. Through their glutamatergic synapses, they could also provide crossover excitation, and this has been proposed as an explanation for the paradoxical ON responses of OFF ganglion cells under some conditions (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014). Acting within a single sublamina, vGluT3 cells might provide either lateral inhibition or lateral excitation. Lateral excitation has been proposed as a mechanism to enhance synchronous firing of neighboring ganglion cells of the same type (Lee et al., Reference Lee, Chen, Chen, Ye, Seal and Zhou2014).
Another, more global function has also been proposed for the vGluT3 cells based on their actions in mouse retina. Glutamate released from vGluT3 cells excites ganglion cells that are highly sensitive to contrast and object motion. The same stimuli suppress the firing of ganglion cells that are inhibited by glycine released from the vGluT3 cells. The role proposed for vGluT3 cells is to coordinate signaling between the multiple parallel pathways in the inner retina. A single type of amacrine cell releasing both neurotransmitters simultaneously would, in theory, be able to accomplish this with much greater temporal precision than two distinct types (Lee et al., Reference Lee, Zhang, Chen and Zhou2016; Tien et al., Reference Tien, Kim and Kerschensteiner2016). The same may also be true for other dual-neurotransmitter amacrine cells.
Conclusions
Many electrophysiological studies of synaptic inputs to neurons of the inner retina have utilized blockers of GABA and glycine receptors or else clamped the membrane voltage at the equilibrium potential for anions in order to study the inputs from bipolar cells. These experimental designs were based on the assumption that bipolar cells provide the only excitatory input to the IPL, but this is no longer true. It is clear now that vGluT3 amacrine cells release glutamate onto some of their postsynaptic targets. Under the conditions assumed to block amacrine cell input, the glutamatergic synapses of vGluT3 cells would still be active, and their effects would be interpreted, incorrectly, as originating from bipolar cells. Other possible sources of excitatory input are the monopolar neurons described recently in mouse retina (Della Santina et al., Reference Della Santina, Kuo, Yoshimatsu, Okawa, Suzuki, Hoon, Tsuboyama, Rieke and Wong2016) and the axon collaterals of melanopsin cells (Joo et al., Reference Joo, Peterson, Dacey, Hattar and Chen2013). Under physiological conditions, all three types of cells receive significant inhibitory input via GABA and glycine. When those inputs are blocked, the contributions of their excitatory synapses would appear much larger. These discoveries will also change the way electron micrographs of the neuropil are interpreted. Profiles in the IPL that make asymmetric synapses have always been assumed to originate from bipolar cell axon terminals, but they might also originate from amacrine cells containing vGluT3, monopolar cells or axon collaterals of intrinsically-photosensitive retinal ganglion cells.