Full Text

Turn on search term navigation

ARTICLE

Received 28 Jun 2016 | Accepted 12 Apr 2017 | Published 7 Jun 2017

The retina processes visual images to compute features such as the direction of image motion. Starburst amacrine cells (SACs), axonless feed-forward interneurons, are essential components of the retinal direction-selective circuitry. Recent work has highlighted that SAC-mediated dendro-dendritic inhibition controls the action potential output of direction-selective ganglion cells (DSGCs) by vetoing dendritic spike initiation. However, SACs co-release GABA and the excitatory neurotransmitter acetylcholine at dendritic sites. Here we use direct dendritic recordings to show that preferred direction light stimuli evoke SAC-mediated acetylcholine release, which powerfully controls the stimulus sensitivity, receptive eld size and action potential output of ON-DSGCs by acting as an excitatory drive for the initiation of dendritic spikes. Consistent with this, paired recordings reveal that the activation of single ON-SACs drove dendritic spike generation, because of predominate cholinergic excitation received on the preferred side of ON-DSGCs. Thus, dendro-dendritic release of neurotransmitters from SACs bi-directionally gate dendritic spike initiation to control the directionally selective action potential output of retinal ganglion cells.

DOI: 10.1038/ncomms15683 OPEN

Dendro-dendritic cholinergic excitation controls dendritic spike initiation in retinal ganglion cells

A. Brombas1,*, S. Kalita-de Croft1,*, E.J. Cooper-Williams1 & S.R. Williams1

1 Queensland Brain Institute, The University of Queensland, Brisbane, Queensland QLD 4072, Australia. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to S.R.W. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683

The algorithms underlying neuronal circuit-based computations are embedded in network connectivity, and implemented by the functional operations of component

synapses and neurons. The retina provides an ideal neuronal circuit to investigate such algorithms1, as the visual world is computed by a relatively simple three-layered network to drive action potential output in classes of retinal ganglion cells which signal salient visual features, such as the direction of image motion to the higher brain29. In the retina the direction of image motion is signalled by classes of direction-selective retinal ganglion cells (DSGCs), which generate robust patterns of action potential output in response to light stimuli moving across or within their receptive elds in a preferred direction, but respond weakly to identical stimuli moving in the opposite null direction24,6,10.

The direction-selective action potential output of DSGCs is believed to be computed by the integration of a directionally un-tuned excitatory input, predominately mediated by glutamate release from bipolar cells1113, with a directionally tuned inhibitory synaptic input generated by the dendritic release of GABA from axonless feed-forward interneurons, termed starburst amacrine cells (SACs)9,1417. SACs represent essential components of the direction-selective circuitry of the retina18. Both ON- and OFF-SACs are distributed, in a mosaic throughout the retina19, and possess a unique radial dendritic morphology, in which each dendrite operates in electrical isolation14,20,21. Functionally, two-photon calcium imaging has revealed that SAC dendritic calcium responses are preferentially generated when light stimuli move from the soma towards terminal dendritic sites14, a direction-selective calcium signal that is thought to generate the dendritic release of neurotransmitters from terminal dendritic synaptic output zones2224. The directional tuning of DSGCs is disrupted by the pharmacological antagonism of GABAA receptors23,25, evincing a prominent role of SAC-mediated synaptic inhibition. Consistent with this, electrophysiological and high-resolution morphological studies have demonstrated that the SAC-mediated inhibitory synaptic control of null direction light responses is mediated by a greater GABAA receptor-mediated synaptic conductance, synapse number, and distribution of dendro-dendritic synapses on the null side of DSGCs16,17,2628. SACs are, however, not simply feed-forward inhibitory interneurons, as ultrastructural and functional evidence indicates that both GABA and acetylcholine (ACh) are co-released by SACs to drive postsynaptic inhibition and excitation16,23,2527,2936. The physiological role of the co-released neurotransmitter ACh is however less well understood, and controversy remains on the contribution of this feed-forward excitatory signal to the generation of light-evoked DSGC action potential output, and its role in the computation of direction selectivity12,16,23,26,29,30,3641. Notably, SAC-mediated cholinergic signalling has been demonstrated to powerfully control the action potential output of DSGCs in response to time-varying visual stimuli, and act as an essential excitatory signal under mesopic, low contrast, conditions40,41. Light-evoked feed-forward SAC-mediated cholinergic excitation may therefore function to provide local dendritic depolarization that acts to gate bipolar-cell-mediated glutamatergic excitation of DSGCs41, which is largely mediated by NMDA receptors42. Recent work has, however, highlighted that dendritic spike generation plays a major role in computation of direction selectivity4345. Direct dendritic recordings have revealed that light-evoked action potential ring of rabbit DSGCs is driven by the initiation of a cascade of dendritic spike generators, a dendritic computation that is silenced by SAC-mediated inhibition45. As GABA and ACh are released from dendritic

sites of SACs16,23,2527,3336, these observations suggest that SAC-mediated cholinergic excitation may also function to control dendritic electrogenesis.

Here we use multi-site electrophysiological recording techniques to demonstrate that the dendro-dendritic release of ACh from SACs powerfully positively gates the initiation of dendritic spikes to control the stimulus-sensitivity, receptive eld structure and the magnitude of light-evoked action potential output of DSGCs.

ResultsPaired recordings from ON-SACs and ON-DSGCs. The role of SACs in the directional selective circuitry of the retina has largely been explored with reference to ON-OFF-DSGCs, a cell-type which receives synaptic input in both the ON- and OFF-sublamina of the inner plexiform layer (IPL)46. In contrast, ON-DSGCs are exclusively driven by the ON pathway, and possess a dendritic tree restricted to the ON- sublamina of the IPL46, providing a simplied circuit for the dissection of the role of SACs in the computation of direction selectivity. Direct electrical recording from the dendrites of ON-DSGCs has revealed that active dendritic integration plays an instrumental role in the generation and control of light-evoked action potential ring and the computation of direction selectivity45. To investigate the role that SAC-mediated cholinergic signalling plays in this computation we rst made paired recordings from morphologically identied ON-SACs and ON-DSGCs maintained in fragments of the adult rabbit retina ex vivo (Fig. 1).

Paired recordings revealed that the generation of threshold electrogenesis in ON-SACs evoked either net excitatory or net inhibitory postsynaptic potentials (PSPs) in synaptically connected ON-DSGCs (Fig. 1; SAC current input: amplitude 0.650.06 nA; duration 10 ms; n 16 pairs).

In connections that exhibited excitatory PSPs, the selective pharmacological blockade of nicotinic ACh receptors (nAChRs) abolished the excitatory response to reveal postsynaptic inhibition (Fig. 1a-c; amplitude distribution in mecamylamine (MMA, 10 mM) signicantly different from control, Po0.0001;

KolmogorovSmironov test; median signicantly different from control, Po0.0001, MannWhitney test; n 9). Conversely, the

specic GABAA receptor antagonist GABAzine (10 mM) unmasked nAChR-mediated excitation in connections that exhibited net inhibition under control conditions (Fig. 1d,e; Po0.0001; KolmogorovSmironov test; median signicantly different from control, Po0.0001, MannWhitney test; n 6).

Pharmacologically isolated cholinergic excitatory and GABAergic inhibitory PSPs exhibited a characteristic difference in their time to onset following SAC activation, with the onset of GABAergic inhibitory PSPs leading that of cholinergic excitatory PSPs (Fig. 1fh, Supplementary Fig. 1ab; IPSP onset time 6.60.9

ms; EPSP onset time 10.31.3 ms, P 0.0432; T 2.295; onset

time measured at 5% of PSP amplitude; IPSP rise time 8.91.9

ms; EPSP rise time 5.80.5 ms; rise time measured between 10

and 90% of PSP amplitude; n 14). This difference could not be

accounted for by a presynaptic mechanism, such as temporal jitter in the engagement of dendritic transmitter release, as the distribution of onset latencies around the mean was similar for pharmacologically isolated excitatory and inhibitory PSPs (Supplementary Fig. 1c). The onset time of SAC-evoked PSPs could however be inuenced by the time course of transmitter diffusion from SAC dendritic sites. Previous studies have suggested cholinergic signalling may occur in a paracrine fashion in the retina, and so may not be reliant on structurally determined dendro-dendritic synapses17,41, an idea consistent with the observed slower onset time of the pharmacologically

2 NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683 ARTICLE

a c

DSGC

SAC

APs

n = 7

Cumulative prob.

1.0

DSGC

SAC

0.5

n = 9

10 0 10 75 90

d e

1.0

GABAzine

Cumulative prob.

DSGC

SAC

0.5

Control GABAzine

100 m

10 0 10 75 90

Amplitude (mV)

20 ms

f g h Component addition

DSGC

SAC

DSGC

SAC

GABAzine

Trial

MMA

GABAzine

1 mV

Control co-release

0 5 10 15 20

Onset latency (ms)

100 mV

Figure 1 | Starburst amacrine cell-mediated cholinergic excitation and GABAergic inhibition. (a) Reconstruction of a connected pair of ON-SAC (green) and ON-DSGCs (black). (b) Paired recording revealed the obligatory co-release of ACh and GABA from ON-SACs. Overlain voltage traces show the transformation of SAC-driven synaptic excitation of an ON-DSGC to synaptic inhibition when nAChRs were blocked with Mecamylamine (MMA, 10 mM, blue traces; action potentials (APs) have been truncated for clarity). The lower traces show the somatic current-evoked excitation of the ON-SAC.

(c) Cumulative probability distributions of the amplitude of each SAC-evoked PSP recorded under the indicated conditions, from the indicated number of paired recordings. (d) In a different pair of ON-SAC and ON-DSGCs net inhibition was generated under control conditions, which was transformed to excitation by the blockade of GABAA receptors (SR-95531 (GABAzine), 10 mM, red; morphologies shown in a). (e) Cumulative probability distributions of

PSP amplitude under the indicated conditions. (f) Onset latency of pharmacologically isolated excitatory and inhibitory PSPs, recorded in GABAzine (n 7)

and MMA (n 7), respectively. Onset latency was measured from the onset of the presynaptic driving current to 5% of the peak of the postsynaptic

response. (g) Averaged pharmacologically isolated excitatory and inhibitory PSPs, the vertical dashed line indicates the onset time of the presynaptic driving current. (h) Upper overlain traces show the biphasic waveform (sum, black trace) produced by the arithmetic summation of isolated excitatory and inhibitory PSPs. The lower trace shows an averaged PSP recorded under control conditions, note the biphasic waveform.

isolated cholinergic PSPs observed here. We note, however, that previous paired recordings have revealed similar onset times of SAC-mediated cholinergic and GABAergic postsynaptic currents in DSGCs, when temporally overlapping excitatory and inhibitory synaptic currents were separated under somatic voltage-clamp by reversal potential16. Functionally, the different onset of dendrodendritic excitation and inhibition observed here would be predicted to lead to the generation of biphasic compound PSPs, if both GABA and ACh are co-released from a single ON-SAC. Indeed when pharmacologically isolated excitatory and inhibitory components were arithmetically summed, a clear biphasic compound PSP was generated (Fig. 1h). Consistent with this, SAC-evoked PSPs recorded under control conditions frequently exhibited a biphasic waveform, that was clearly evident in single trials and when consecutive responses were digitally averaged (Fig. 1d,h, respectively). Together, these data reveal that ACh and GABA are co-released from single ON-SACs.

We next enquired if the synaptic impact of ON-SACs aligned with the directional tuning of postsynaptic ON-DSGCs (Fig. 2). To do this we mapped the action potential output of ON-DSGCs generated by light bars moved across their receptive elds and

made paired recordings from ON-SACs positioned close to the edge of the dendritic arbour rst activated by light stimuli moving in a preferred direction, the preferred side, or ON-SACs positioned close to the edge of the dendritic arbour rst activated by null direction light stimuli, the null side (Fig. 2a,b inset and Fig. 2c,d, inset, Supplementary Fig. 2; light bar size 100 by 300400 mm, moved in one of 12 directions at

0.24 mm s 1; direction selectivity index (DSI) 0.930.03;

SAC-DSGC soma separation 27014 mm; n 17). When the

somata of ON-SACs were positioned close to the edge of the preferred side of ON-DSGCs, and the sites of close dendro-dendritic SAC-DSGC apposition restricted to the preferred side, net excitatory responses were generated (Fig. 2a,b,e; PSP amplitude 2.020.37 mV, integral 31.44.3

mV.s; n 6 pairs). In contrast, when the somata and sites of

dendro-dendritic apposition were focussed on the null side, responses were on average inhibitory, but demonstrated a wide range (Fig. 2c-e; PSP amplitude 0.590.8 mV;

integral 33.721.0 mV.s; n 11 pairs; signicantly different

from preferred side: P 0.035, T 2.31; integral: P 0.041,

T 2.24). Under current-clamp recording conditions, therefore,

NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683

b Pref. e

DSGCs

Preferred

Null

100 m

6 mV

5 mV

Figure 2 | The impact of ON-SACs aligns with the directional tuning of postsynaptic ON-DSGCs. (a) Reconstruction of a connected pair of ON-SAC (green) and ON-DSGCs (black), showing the sites (red symbols), and vectorial angle (red arrow, relative to SAC somata) of close dendro-dendritic appositions. Morphologies have been aligned to the preferred direction vector of ON-DSGC light responses (b inset). In all panels the preferred side of the ON-DSGC, the dendritic eld rst activated by light stimuli moving in the preferred direction, lies above the horizontal dashed line. Note that the SAC somata and all dendro-dendritic close appositions are positioned on the preferred side of the postsynaptic ON-DSGC. (b) Paired recordings from the illustrated cells (a) revealed that activation of the SAC-evoked excitatory PSPs in the ON-DSGC (grey traces, red trace is a digital average). The somatic current-evoked presynaptic SAC voltage responses are shown below (green trace). (c) Reconstruction of an ON-SAC located in the null side of an ONDSGC. The morphologies have been aligned to the ON-DSGC light response vector (d inset). (d) Paired recordings from the illustrated cells (c) revealed SAC-mediated inhibitory PSPs (grey traces, blue trace is a digital average). (e) Summary of the impact of SACs located on the preferred and null sides of postsynaptic ON-DSGCs. The morphologies of postsynaptic ON-DSGCs are shown as overlain and spatially ltered (50 mm, black) reconstructions registered to the preferred direction of light responses, shown schematically by the yellow light bars. Nested within this eld are perimeter representations of SACs (green). Arrows indicate the vector angle of SAC-DSGC close dendro-dendritic appositions coloured according to their postsynaptic impact (red excitatory, blue inhibitory PSPs). Vector length represents 1 contact per mm. All light stimuli, the direction of which are schematically illustrated, were applied under photopic conditions. Stimulus intensity was twice that of background illumination.

the balance of SAC-mediated cholinergic excitation outweighs that of inhibition on the preferred side of ON-DSGCs, while inhibition outweighs excitation on the null side (Fig. 2e). This nding parallels the targeted dendritic impact of SAC-mediated excitation and inhibition reported for ONOFF DSGCs16,26,27,36, and so reveals stereo-typed dendro-dendritic circuitry in the ON- and OFF-sublamina of the inner plexiform layer.

Cholinergic excitation controls light-evoked neuronal output. To explore the contribution of SAC-mediated cholinergic excitation to the light-evoked action potential output of DSGCs we presented moving light bar stimuli that were swept across the receptive elds of ON-DSGCs (Fig. 3; Supplementary Fig. 3). Antagonism of nAChRs dramatically reduced the robust action potential ring of ON-DSGCs evoked by light bars moved in the preferred direction and virtually eliminated the sparse action potential ring evoked by null direction stimuli, an effect evident in somatic whole-cell and cell-attached recording modes (Fig. 3ac; Supplementary Fig. 3; preferred direction: control 27.42.5 Hz; nAChR antagonist 3.10.8

Hz; Po0.0001, T 11.62; null direction: control 1.10.3 Hz;

nAChR antagonist 0.030.02 Hz; P 0.0008, T 3.77;

n 28). Analysis of whole-cell recorded light responses revealed

that nAChR antagonists signicantly reduced the membrane depolarization evoked by preferred direction light stimuli, and transformed null direction responses from weakly excitatory to inhibitory (Fig. 3a). This effect could not be accounted for by the reduction of action potential ring, as this transformation was

apparent when action potentials were digitally removed using a median lter (10 ms)46 (Fig. 3d; Supplementary Fig. 4; voltage integral of median ltered light responses: preferred direction: control 12.01.0 mV.s; nAChR antagonist 0.60.7 mV.s;

Po0.0001, T 11.2; null direction: control 4.01.1 mV.s;

nAChR antagonist 8.70.7 mV.s; Po0.0001, T 10.27;

n 28). Furthermore, when directional tuning was disrupted

by antagonism of GABAA receptors, the blockade of nAChRs powerfully decreased moving light bar-evoked action potential output in all tested directions (Supplementary Fig. 5; action potential ring rate: GABAzine 35.83.6 Hz,

GABAzine Hex 7.83.8 Hz; Po0.0001, T 10.96;

n 6; DSI: GABAzine 0.030.03, GABAzine Hex 0.140.05).

To further explore the role of cholinergic signalling in the control of the light-evoked action potential output of DSGCs we disrupted the hydrolysis of ACh by the application of the acetylcholinesterase (AChE) inhibitor ambenonium. In the presence of ambenonium the action potential ring of ON-DSGCs evoked by preferred and null direction light stimuli were powerfully facilitated and directional-selectivity disrupted (Fig. 4a,b; DSI: control 0.930.0.1;

ambenonium 0.170.03; P 0.002, Wilcoxon signed rank test;

voltage integral of median ltered light responses: preferred direction: control 8.80.8 mV.s; ambenonium 27.62.1 mV.s;

P 0.0002, T 9.85; null direction: control 2.20.6 mV.s;

ambenonium 28.83.1 mV.s; P 0.0004, T 8.17; n 6). Pre

vious work has demonstrated that the pharmacological manipulation of AChE allows investigation of the spatial relationship between

4 NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683 ARTICLE

Preferred direction

Null direction

b c Preferred direction

Preferred direction

Photoreceptors

SAC DSGC

Null

4 Control Hex

AP rate (Hz)

0 Cont. Hex MMA

d Preferred Null

APs per bin

n = 23

Bipolar cells

Hex(100 M) 10 mV

4 mV

GABA

ACh

0 0.5

Soma 0.5

Cont.

Null direction

APs per bin

Control Hex

0.1

Voltage (mV.s)

Preferred

0 0.5

Soma

0.5

0.5 s

Stimulus position (mm)

Figure 3 | Cholinergic signalling powerfully controls the light-evoked action potential output of ON-DSGCs. (a) Physiological activation of the retinal microcircuit (summarized in the inset schematic) by light bars moved across the receptive eld of an ON-DSGCs (inset morphological reconstruction) leads to the generation of powerful action potential (AP) ring when moved in a preferred direction, but sparse AP output when moved in a null direction (control, black traces, APs have been truncated for clarity). The antagonism of nAChRs leads to a reduction of preferred direction light-evoked AP ring and the generation of pure inhibitory responses when light stimuli are moved in the null direction (hexamethonium (Hex); red traces). (b) Peri-stimulus histogram of preferred and null direction light-evoked AP ring under the indicated conditions (bin size 20 mm). (c) Quantication of the reduction of preferred direction light-evoked AP ring by nAChR antagonists (mecamylamine (MMA); blue symbols). (d) Quantication of the voltage integral of median ltered (10 ms) preferred and null direction light responses under the indicated conditions (control versus Hex: preferred: Po0.0001, T 10.19;

null: Po0.0001, T 11.47; control versus MMA: preferred: Po0.0001, T 15.86; null: P 0.0038, T 6.02). All light stimuli were applied under photopic

conditions. Stimulus intensity was twice that of background illumination.

cholinergic release sites and postsynaptic nAChRs, nding that manipulation of ACh hydrolysis alters cholinergic signalling only when release sites are spatially distant to activated postsynaptic nAChRs47,48. We therefore examined if blockade of AChE controlled pharmacologically isolated unitary nAChR-mediated PSPs evoked in paired SAC-DSGC recordings. Under these conditions the application of the AChE inhibitor ambenonium signicantly enhanced unitary cholinergic excitatory transmission, but did not affect pharmacologically isolated unitary GABAergic inhibition (Fig. 4c,d; excitatory PSP integral: control 20.84.2 mV.s; ambenonium 54.311.5 mV.s;

P 0.011, T 4.49; n 5). Furthermore paired SAC-DSGC

recording revealed that the augmentation of ACh hydrolysis by the local application of exogenous AChE reversibly attenuated the amplitude of pharmacologically isolated nAChR-mediated PSPs (Supplementary Fig. 6; AChE (0.4 U per ml) dissolved in Ames solution: control 1.870.25 mV; AChE puff 1.100.17 mV;

P 0.0023, T 5.72; n 6). In contrast, the control local

application of Ames solution did not alter SAC-DSGC excitatory transmission (Supplementary Fig. 6; control 2.150.64 mV;

Ames puff 1.990.67 mV; P 0.105, T 2.09; n 5). The bi

directional control of unitary SAC-evoked cholinergic transmission by the augmentation and reduction of AChE activity is therefore consistent with a spatial separation between SAC release sites and activated postsynaptic AChRs47,48. To verify that ACh release controls preferred and null direction light responses we depleted presynaptic ACh by blocking the vesicular ACh transporter with vesamicol49. When presynaptic ACh release was depleted, preferred and null direction light responses were severely attenuated, and SAC-DSGC excitatory synaptic transmission selectively depressed (Fig. 4eh; voltage integral of median ltered light responses: preferred direction: control 14.53.4 mV.s;

vesamicol -3.31.5 mV.s; P 0.0008, T 9.05; null direction:

control 4.93.2 mV.s; vesamicol -11.11.9 mV.s; P 0.0004,

T 11.11; n 5; excitatory PSP amplitude: control 3.00.6 mV;

vesamicol 0.90.2 mV; P 0.004, T 4.47; n 7). Taken

together these data reveal that ACh release from ON-SACs powerfully controls the physiological responsiveness of ON-DSGCs through the activation of postsynaptic nAChRs, in a manner consistent with a local paracrine form of neurotransmission.

To examine if cholinergic excitation inuenced the responsiveness of ON-DSGCs to light stimuli across a wide stimulus range, we systematically varied the intensity of light stimuli and their speed of motion to generate stimulus-response relationship under control conditions and in the presence of a nAChR antagonist (Fig. 5a,b; intensity: 10 to 100% above background; speed: 0.040.9 mm s 1). Under control conditions the directional selective output responses of ON-DSGCs emerged at the light stimulus threshold for action potential ring and were maintained across a 50-fold preferred direction ring range (Fig. 5c). The antagonism of nAChRs increased the threshold light intensity, constrained the speed of stimuli that generated action potential output, and attenuated action potential ring throughout the range of light stimuli (Fig. 5a,b). Notably, this compression of dynamic range was not accompanied by a degradation of the delity of the computation of direction selectivity (Fig. 5c; DSI: control 0.9460.014; Hex

0.9900.004; n 70 trials, n 9 cells). To further explore the

light range over which SAC-mediated cholinergic signalling controlled the responsiveness of ON-DSGCs, we made recordings in preparations adapted to mesopic light conditions (background illumination and stimulus intensity reduced to 0.06 of the standard photopic levels). Under these dim light conditions antagonism of nAChRs dramatically reduced the action potential output of ON-DSGCs evoked by preferred and null direction light bars (Supplementary Fig. 7; preferred direction: control

15.12.3 Hz; nAChR antagonist 0.40.1 Hz; Po0.0001,

T 6.314; n 11; null direction: control 1.00.3 Hz; nAChR

antagonist 0.040.02 Hz; P 0.0027, T 3.961). Altogether

NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683

a Control Ambenonium (50 nM)

Control Vesamicol (10 M)

b f

Preferred Null Preferred Null

n = 6 n = 5

Control ABN

Control VES

APs per bin

500

Soma

500

Soma

500

Soma

500

Soma

500

Stimulus position (m)

c d

TTX + GABAzine

TTX + MMA + ABN (50 nM)

EPSP h

ABN (50 nM)

TTX + GABAzine

VES (10 M)

0.10

1.5

EPSP

IPSP

Norm. PSP amplitude

1.0

0.05

0.5

PSP area (mV.s)

1.5

1.0

0.1

0.5

n = 3

0.2

0 10 20 30

Time (min)

Figure 4 | Selective pre or postsynaptic manipulation of cholinergic signalling controls light responses. (a) Blockade of acetylcholinesterase activity powerfully augments action potential (AP) ring evoked by preferred and null direction light bars (APs have been truncated for clarity). (b) Peri-stimulus histograms of light-evoked AP ring under the indicated conditions (ambenonium (ABN); bin size 20 mm). (c) Ambenonium augments pharmacologically isolated SAC-mediated excitatory PSPs (recorded in tetrodotoxin (TTX); 1 mM and GABAzine (10 mM)), but not inhibitory PSPs (recorded in TTX and mecamylamine (MMA); 10 mM). (d) Pooled data showing the selective augmentation of the area of excitatory PSPs by the blockade of acetylcholinesterase activity (area measured over a 100 ms time-window). (e) Blockade of the vesicular ACh transporter attenuates preferred direction light-evoked AP output, and unmasks null direction membrane hyperpolarization. (f) Peri-stimulus histograms of light-evoked AP ring under the indicated conditions (vesamicol (VES); bin size 20 mm). (g) Vesamicol attenuates pharmacologically isolated SAC-mediated excitatory PSPs, but not inhibitory PSPs. (h) Pooled data showing the selective reduction of excitatory PSPs by the pharmacological blockade of the vesicular ACh transporter (PSP amplitude has been normalized).

All light stimuli were applied under photopic conditions. Stimulus intensity was twice that of background illumination. Data in d,h represent means.e.m.

Stimulus speed (mm s1)

1.0

Control

AP firing rate (Hz)

0.8

Direction index

0.6

50 mV

1 s

0.4

Hex

0.2

0 0.2 0.4 0.6 0.8 1.0 0 20 40 60 80 100 0 20 40 60

Stimulus intensity (%)

Observation

Figure 5 | Cholinergic signalling controls the stimulus sensitivity and dynamic range of ON-DSGCs. (a,b) Antagonism of nAChRs (hexamethonium (Hex); red symbols and traces) attenuates action potential (AP) ring evoked by preferred direction light bars moved across the receptive eld over a wide-range of speeds (a), and at the optimal speed across a range of light intensities (b; speed 0.24 mm/s). (c) Impact of cholinergic signalling on the

computation of direction selectivity, and preferred direction light-evoked AP ring over the stimulus range shown in a,b. Note the decreased stimulus sensitivity, attenuation of AP ring rate, but preservation of directional tuning in hexamethonium. All light stimuli were applied under photopic conditions. Stimulus intensity is relative to background illumination.

6 NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683 ARTICLE

a b

Preferred direction Preferred direction

Control

Hex

Control Hex

Preferred

APs per bin

0 Edge

0.5

1.0

0.1

0 0.2 Edge

0.1

Stimulus position (mm) Stimulus position (mm)

Photopic 100% Photopic 50% Mesopic 100%

Cell

0 Edge 0.5 1.0

0 Edge 0.5 1.0 Edge 0.5 1.0 Edge 0.5 1.0

0 Edge 0.5 1.0 Edge 0.5 1.0

Displacement (mm)

1.0

Cumulative prob.

0.5

0.5 s

20 mV

Figure 6 | Cholinergic signalling controls receptive eld structure. (a) The tight spatial relationship between the onset of action potential (AP) ring and the peripheral edge of the dendritic arbour of ON-DSGCs is disrupted by nAChR antagonism (control, black traces; hexamethonium (Hex); 100 mM, red traces, * indicates rst AP). The dashed reference line shows the edge of the partially reconstructed dendritic tree. Light stimuli were applied under photopic conditions (stimulus intensity 50% greater than background illumination). (b,c) Dendritic edge aligned peri-stimulus histogram of preferred direction light-evoked AP ring under control (black) and in the presence of nAChR antagonists (red; Hex; 100 mM, bin size 20 mm, photopic 50% stimuli).

(d) Cholinergic excitation controls receptive eld size across a wide range of light intensities. Graphs summarize the relationship between the mean (s.e.m.) spatial position of the rst AP evoked by preferred direction light stimuli relative to the edge of the preferred dendritic subtree under control (black) and in nAChR antagonists (red). The percentage stimulus intensity, relative to background, is shown above each graph. Each row represents the results from a single cell (4 5 trials under each condition), the grey arrows indicate ON-DSGCs that did not generate AP output in the presence of the

nAChR antagonist. (e) Cumulative probability plots of the spatial position of the rst AP evoked by preferred direction light stimuli relative to the edge of the preferred dendritic subtree in single trials, data are illustrated over the stimulus intensity range indicated in d.

these data show that cholinergic excitation acts to control the light stimulus-sensitivity and the magnitude of ON-DSGC output responses.

Cholinergic signalling drives active dendritic integration. To gain insight into how SAC-mediated cholinergic excitation is integrated in postsynaptic ganglion cells we rst determined if cholinergic signalling controlled the receptive eld structure of ON-DSGCs. Under control conditions we found a tight spatial relationship between the size of the supra-threshold receptive eld and the dendritic eld size of ON-DSGCs across a wide-range of light conditions, consistent with previous reports50 (Fig. 6; receptive eld size 92324 mm; dendritic eld

size 85123 mm, r 0.574; Po0.0001, n 66 cells). Notably,

when preferred direction light bars were swept across the receptive eld a close relationship was found between the time at which light stimuli activated the retinal circuitry which innervated the edge of the preferred dendritic subeld of ON-DSGCs and the time of occurrence of the rst light-evoked action potential, which could be converted into a spatial relationship (Fig. 6ae; control: displacement 19.83.6 mm;

n 642 trials, n 66 cells). The pharmacological blockade of

nAChRs reduced action potential ring and delayed its onset as preferred direction light bars were swept across the receptive eld, resulting in a disruption of the relationship between dendritic eld edge and action potential generation, and a dramatic constriction of supra-threshold receptive eld size (Fig. 6ae; control 92324 mm; nAChR antagonist 47140 mm;

Po0.0001; T 11.27; n 66). The blockade of SAC-mediated

cholinergic excitation disrupted the wide eld receptive eld properties of ON-DSGCs to a similar degree under photopic and mesopic light conditions (compare graphs in Fig. 6d,e), suggesting that SAC-mediated cholinergic signalling functions to powerful drive dendritic excitation across a broad physiological range of stimuli.

Previous observations from the rabbit retina have indicated that active dendritic integration is essential for the emergence of the wide receptive eld properties of ON-DSGCs43,45. We therefore used simultaneous somato-dendritic recording techniques to directly examine the dendritic mechanisms underlying the integration of SAC-mediated cholinergic excitation. Under control conditions simultaneous somato-dendritic recordings demonstrated that preferred direction light stimuli drove the robust generation of dendritic spikes, which forward propagated to the soma and axon to initiate

NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683

a b Control Hex

Preferred direction

APs

Dendritic spikes

150

100

Events

0 Cont. Hex Cont. Hex

Time (ms)

Soma first

500

Soma

500

Stimulus position (m)

e g

Appositions Timing

1.0

Cumulative prob.

SAC

DSGC

0.5

Dendritic Rx Distance

Soma 500

0 0.5 1

Distance (m)

Dendrite first

Time (ms)

Figure 7 | Cholinergic signalling controls light-evoked dendritic spike initiation. (a) Reconstruction of an ON-DSGC showing the placement of recording electrodes and the preferred direction movement of a light bar; the blue coloured section of the dendritic tree feeds to the dendritic recording site.(b) The generation of dendritic spikes (dendritic recording (blue traces), positions 1 and 2 in a), and back-propagating action potentials (positions 3 and 4 in a) were attenuated by antagonism of nAChRs (hexamethonium (Hex); 100 mM) when a preferred direction light bar was swept across the receptive eld.

Traces are aligned to the peak of somatically recorded action potentials (APs). (c) Summary of the reduction of AP output and dendritic spike generation by hexamethonium. (d) Spatial pattern of somato-dendritic spike delay under control (black symbols) and following the blockade of nAChRs (Hex; red symbols). (e) Reconstruction of simultaneously recorded ON-SAC and ON-DSGCs, showing the placement of recording electrodes. (f) ON-DSGC somatic (black traces) and dendritic (blue traces) recordings show that the presynaptic SAC drives dendritic spike generation (reconstruction in a; recordings were made in GABAzine (10 mM); ve consecutive SAC-evoked supra-threshold excitatory responses are shown). The lower overlain traces show the dendro-somatic attenuation of a sub-threshold voltage response evoked by SAC activation (green trace). (g) Quantication of the sites of SAC-DSGC close apposition in recorded DSGC dendritic subtrees (left graph, vertical line represents average dendritic recording site; 1455 mm from the soma; n 5).

Note that all detected appositions were distal to dendritic recording sites. The right graph shows that dendritic spikes preceded action potentials in all supra-threshold trials.

action potential ring (Fig. 7a,b; Supplementary Fig. 8). The pharmacological blockade of SAC-mediated cholinergic excitation dramatically attenuated light-evoked dendritic spike generation and consequential action potential output (Fig. 7ad). Analysis revealed that cholinergic excitation controlled the number of dendritic spikes evoked by light bars as they excited terminal dendritic sites which fed to the site of dendritic recording, illustrating that SAC-mediated cholinergic excitation is integrated locally in the dendritic arbour of ON-DSGCs (Fig. 7ad, terminal dendritic tree highlighted in blue in the reconstruction; dendritic spikes per episode: control 14.41.9;

Hex 2.00.8; P 0.0002, T 6.4; n 9; dendritic recording

preferred side 27522 mm from soma; n 5; null

subeld 26541 mm; n 4, Supplementary Fig. 8). To

explore the determinants of this form of sub-cellular integration, we tested if cholinergic excitation provided by a single SAC was capable of driving the initiation of dendritic spikes in ON-DSGCs. To do this we made simultaneous somato-dendritic recordings from ON-DSGCs and a third somatic recording from a presynaptic ON-SAC, in the presence of the GABAA receptor antagonist GABAzine

(Fig. 7eg; n 7). Under these conditions the activation of a

single SAC led to the generation of nAChR-mediated dendritic excitatory PSPs, which were crowned by the ring of dendritic spikes, in recordings where dendro-dendritic SAC-DSGC appositions were focussed in the recorded dendritic arbour of the ON-DSGCs at loci distal to the site of dendritic recording (Fig. 7eg; dendritic recordings 1455 mm from soma; average site of SAC dendro-dendritic appositions from DSGC soma 43726 mm; n 5). Simultaneous somatic recording

revealed that each dendritic spike preceded and drove action potential ring (dendritic spike to action potential delay 0.400.05 ms). In contrast, when dendritic recordings

were made from the dendritic subeld contralateral to the site of predominate SAC innervation, SAC-evoked cholinergic excitation drove action potential ring, which was rst recorded somatically and subsequently back-propagated to the dendritic recording site, consistent with cholinergic excitation of the contralateral dendritic tree (Supplementary Fig. 9). Taken together these data directly demonstrate that dendro-dendritic cholinergic excitation is capable of driving the generation of dendritic spikes in ON-DSGCs.

8 NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683 ARTICLE

a Control Hex (100 M)

0.6

0.4

Control Hexn = 20

APs per bin

0.2

0 0 0.2 0.4 0.6

Time (s)

1.0

Terminal spikes

Dend. spikes

Events per bin

0.5

Control Hex

0 0 20 40 60 80

Light

Amplitude (mV)

Figure 8 | Cholinergic signalling drives a cascade of dendritic spike generators. (a) Dramatic attenuation of light spot-evoked terminal dendritic (marked by *) and large-amplitude dendritic spike generation by the blockade of nAChRs (Hex; 100 mM). The inset shows ON-DSGC morphology, and the placement of recording electrodes and light spot stimuli. (b) Peri-stimulus time histogram of AP ring evoked by dendritic light spots (ON time 00.5 s)

under the indicated conditions. (c) Amplitude distributions of light spot-evoked dendritic regenerative activity under control (black) and in hexamethonium show the nAChR-mediated control of terminal and large-amplitude dendritic spike generation.

Cholinergic signalling gates dendritic spike initiation. To quantitatively examine how light-evoked dendro-dendritic cholinergic excitation controls the generation of dendritic spikes, we ashed light spots at terminal dendritic sites under control and in the presence of nAChR antagonists (Fig. 8; light spot 1055.2 mm; n 20). In conrmation of previous obser

vations45, direct dendritic recordings revealed that light spots evoked both small- and large-amplitude dendritic spikes under control conditions. Small-amplitude, terminal dendritic, spikes were compartmentalized in the dendritic tree and acted as variable triggers of large-amplitude dendritic spikes, which preceded and drove action potential output (Fig. 8a). The application of nAChR antagonists signicantly limited the initiation of dendritic spikes, by decreasing the occurrence of light spot-evoked terminal and large-amplitude dendritic spikes, and as a consequence action potential output (Fig. 8ac). In support of the central role of cholinergic excitation in this form of local dendritic integration, simultaneous somato-dendritic recordings revealed that the dendritic depolarization evoked by light spot stimuli was dramatically reduced in amplitude by the antagonism of nAChRs when recordings were made in the presence of tetrodotoxin (TTX) and GABAzine to block dendritic spike generation and SAC-mediated inhibition, respectively45 (Supplementary Fig. 10; light spot 100 mm; dendritic

recordings 24011 mm from soma; n 7, control: dendrite

16.62.2 mV, soma 4.50.7 mV; Hex: dendrite

6.81.2 mV, soma 1.70.3 mV; P 0.0008, T 6.24 and

P 0.0009, T 6.04, respectively; n 7). Taken together, these

data reveal that light spot-evoked bipolar-cell-mediated glutamatergic12,42 and SAC-mediated cholinergic excitation are integrated locally in the dendritic tree to inuence the initiation of terminal dendritic spikes, and the subsequent recruitment of large-amplitude parent dendritic spikes, which in turn, drive action potential output.

To critically evaluate the contribution of cholinergic excitation in this multi-level integrative process we attempted to recreate the postsynaptic effects of light-evoked dendritic cholinergic excitation on dendritic spike initiation when nAChRs were pharmacologically blocked, or when presynaptic ACh release was depleted (Fig. 9). When nAChRs were blocked, light spot-evoked terminal and large-amplitude dendritic spike generation was

signicantly attenuated. The generation of dendritic spikes by light-evoked bipolar cell-mediated glutamatergic excitatory input could however be rescued by the pairing of light spot stimuli with direct dendritic depolarization generated by the injection of sub-threshold steps of positive current through the dendritic recording electrode (Fig. 9a,b; light spot 107.112 mm;

dendritic current injection 138.127.5 pA; dendritic

recordings 21512 mm from soma; n 7). Notably, the

occurrence of both terminal and large-amplitude dendritic spikes were signicantly augmented by the pairing of light spot stimuli with direct dendritic depolarization, whereas dendritic depolarization alone failed to evoke small-amplitude, terminal dendritic spike generation, but infrequently reached the threshold for the generation of large-amplitude dendritic spikes in a fraction of trials (Fig. 9c,d; occurrence of terminal and large-amplitude dendritic spikes signicantly different between paired and unpaired Hex conditions; P 0.0071, T 4.0 and

P 0.0274, T 2.9, respectively; n 7). In order to more closely

replicate the actions of light-evoked ACh release from SACs we paired the local dendritic iontophoretic delivery of ACh with light spot excitation, when SAC-mediated ACh release had been depleted by blocking the ACh vesicular transporter (Fig. 9e,f; Supplementary Fig. 11). Under these conditions the pairing of ACh iontophoresis with light spot stimuli powerfully drove dendritic spike generation and action potential output, whereas light stimuli, or ACh iontophoresis alone infrequently reached the threshold for dendritic spike generation (Fig. 9e,f; occurrence of terminal and large-amplitude dendritic spikes signicantly increased in the paired condition; P 0.0047; q 6.44 (light

alone) and P 0.0012; q 7.99 (iontophoresis alone); ANOVA;

n 5; dendritic recordings 31515 mm from soma). Consistent

with this, when SAC-mediated excitation and inhibition were blocked, the pairing of light-evoked bipolar cell-mediated excitatory input with direct dendritic depolarization dramatically facilitated dendritic spike generation and neuronal output, underscoring the role of SAC-mediated inhibition in the control of dendritic spike generation45 (Supplementary Fig. 12; light spot 87.512.5 mm; dendritic recordings 13214 mm from

soma; dendritic current 22547.1 pA; dendritic spikes per

episode: unpaired 1.10.5; paired 12.82.9; P 0.003,

T 6.3; n 4). These data therefore demonstrate

NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683

Control Hex (100 M) b

0.6

Events per trial

APs per bin

0.4

inj

No light + Iinj

n = 7

0.2

Light

0 0 0.6

0.2 0.4

Time (s)

0 Cont. Hex

Dend.

Terminal

1 Terminal spikes

Dend. spikes

Events per bin

0.5

Events per trial

Soma

Light

I inj

100 ms

0.2 nA

0 0 20 40 60 80

0 Hex + Iinj

No light + Iinj

Amplitude (mV)

Vesamicol (10 M) + Acetylcholine + AcetylcholineNo light

Vesamicol

Light + ACh

No light + ACh

Events per trial

20 mV

1 ms

n = 5

Light 100 ms

Light ACh

ACh Ionto.

Figure 9 | Cholinergic signalling positively gates terminal dendritic spike generation. (a) Light spot-evoked terminal dendritic (marked by *) and large-amplitude dendritic spike generation were powerfully attenuated by nAChR antagonism (upper traces, hexamethonium; Hex). This reduction could be rescued by pairing light stimuli with sub-threshold dendritic depolarization (lower left traces). The insets show light spot-evoked dendritic spikes (blue traces) and somatically recorded activity (black traces). A section of the ON-DSGC morphology, the placement of electrodes, and the position of the light spot stimuli are also shown. (b) Peri-stimulus time histogram of light spot-evoked (ON time 00.5 s) action potential (AP) output under the indicated

conditions. (c) Amplitude distribution of dendritically recorded spikes under the indicated conditions. Note that pairing of light spot stimuli with dendritic depolarization increased the generation of both terminal and large-amplitude dendritic spikes. (d) Number of terminal and large-amplitude dendritic spikes evoked by light spot stimuli under the indicated conditions (data represent means.e.m). (e) When presynaptic ACh release was depleted (vesamicol), the pairing of light spot stimuli with the local dendritic iontophoretic delivery of ACh augmented dendritic spike generation. The insets show overlain spikes, a section of the ON-DSGC morphology, the placement of recording and iontophoresis electrodes, and the position of light spot stimuli. (f) Number of terminal and large-amplitude dendritic spikes evoked by light spot stimuli per trial under the indicated conditions (data represent means.e.m).

that SAC-mediated cholinergic and bipolar cell-mediated glutamatergic excitation drive the generation of dendritic spikes, revealing a mechanism by which dendro-dendritic ACh release acts in a spatially selective manner in the dendritic tree of DSGCs to control neuronal output.

DiscussionThe retinal microcircuitry that underlies the processing of image motion has been solved at ultrastructural resolution9,17, but a full understanding of the mechanisms underlying the computation of direction selectivity requires investigation of circuit function at a similar scale. Here we used multi-site somatic and dendritic electrophysiological recording techniques to demonstrate that the direction-selective action potential output of rabbit ON-DSGCs is powerfully controlled by SAC-mediated dendro-dendritic cholinergic excitation.

In conrmation of previous ndings obtained from ON-OFF-DSGCs16,26,3436, our paired ON-SAC to ON-DSGC recordings revealed the dendro-dendritic co-release of ACh and GABA from SACs, which drove postsynaptic excitation and inhibition, mediated by the activation of nAChR and GABAA receptors, respectively. Notably, we observed that the

postsynaptic impact of SACs was dened by their position in the receptive eld of postsynaptic DSGCs. SACs positioned on the preferred side, which made close dendro-dendritic appositions from dendrites oriented in the preferred direction of DSGCs, generated net excitatory responses in DSGCs, whereas, SACs located on the null side generated, on average, postsynaptic inhibition. These data suggest that when light stimuli enter the receptive elds of DSGCs in a preferred or null direction, the activation of SACs located on the periphery of the preferred and null side should differentially inuence the action potential output of ON-DSGCs. Consistent with this, when the direction-selective microcircuit was physiologically activated, the robust preferred direction light-evoked action potential output of ON-DSGCs was greatly attenuated by the pharmacological antagonism of nAChRs, and by the selective depletion of presynaptic ACh release from SACs. Notably, we found across a broad range of light stimuli that the antagonism of nAChRs disassociated the tight relationship between the receptive and dendritic eld areas of ON-DSGCs, and disrupted the characteristic generation of action potential ring evoked by preferred direction light bars as they entered the receptive eld.

10 NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683 ARTICLE

a b

Controlpreferred direction preferred direction

DSGC

No cholinergic excitation

DSGC

Excitation

Figure 10 | Schematic summary of the role of SAC-mediated control of active dendritic integration compartments of ON-DSGCs. (a) SAC-mediated cholinergic excitation and GABAergic inhibition gate terminal dendritic integration to control action potential output under control conditions. (b) In the absence of cholinergic excitation terminal dendritic spike generation is weakly driven by bipolar cell-mediated excitation.

Direct dendritic electrical recording, calcium imaging and computational modelling have shown that active dendritic integration underlies the wide-eld responsiveness of DSGCs13,4345. Multi-site somato-dendritic recording techniques have deconstructed this process to reveal that dendritic spikes are rst initiated in the small calibre terminal dendrites of DSGCs45, which receive excitatory glutamatergic synapses from bipolar-cells51,52. Terminal dendritic spike generation has been found to be the rst step in a multilayered active integration cascade, as these spikes act as variable triggers for the generation of dendritic spikes in larger calibre parent dendrites that, once initiated, forward propagate to the axon to drive action potential output45. Our results demonstrate that SAC-mediated cholinergic signalling is a major component of this integrative process. Simultaneous somato-dendritic recordings revealed that the generation of terminal and large-amplitude parent dendritic spikes by light spots or preferred direction moving light bars were powerfully attenuated by the antagonism of nAChRs. An effect underscored by the rescue of dendritic spike generation when cholinergic excitation was mimicked by local dendritic depolarization or the terminal dendritic application of ACh. The dendro-dendritic release of ACh therefore controls the action potential output of DSGCs by driving terminal dendritic spike generation, and enhancing the coupling between active dendritic integration sites. We suggest that SAC-mediated cholinergic excitation acts as a dendritic subunit-specic gate of bipolar-cell signalling, that when activated by moving light bars powerfully controls the action potential output of DSGCs, to dene the stimulus sensitivity and receptive eld area over which the computation of direction selectivity is executed (Fig. 10). An idea supportive of previous observations made from ON-OFF-DSGCs, which have revealed that cholinergic excitation acts to control bipolar-cell mediated

glutamatergic excitation by supplying membrane depolarization to relieve the voltage-dependent block of NMDA receptor-mediated PSPs41. Our ndings however extend this idea to demonstrate that these excitatory inputs act in concert to drive the initiation of terminal dendritic spikes, and thus launch a cascade of dendritic spike generation which ultimately results in the driving of action potential output, providing a substrate for the localized dendritic processing of light stimuli3 (Fig. 10).

In contrast, SACs located on the null side of DSGCs drove net postsynaptic inhibition, a nding that is consistent with the greater GABAA receptor-mediated synaptic conductance, and distribution of dendro-dendritic synapses on the null side of DSGCs16,17,2628. However, at these sites, we nd that dendro-dendritic cholinergic signalling plays a functional role, as the blockade of cholinergic excitation transformed null direction light-evoked voltage responses from weakly excitatory to strongly inhibitory. Consistent with this, when GABAergic inhibition was pharmacologically blocked, the antagonism of nAChRs attenuated moving light bar-evoked action potential output in all tested directions, as has been shown for ON-OFF-DSGCs23. Furthermore, when the uptake of ACh was inhibited, both null and preferred direction light responses were strongly facilitated, and direction tuning disrupted. These data reveal that a balance between the postsynaptic impact of SAC neurotransmitters is essential for the emergence of directional selectivity, questioning how action potential ring is suppressed by null direction light stimuli.

Dendritic recording and neuronal modelling has demonstrated that null direction light stimuli evoke sparse action potential output in rabbit DSGCs because of the SAC-mediated inhibitory control of dendritic spike initiation44,45. Direct recording has demonstrated that this inhibitory control is restricted to terminal dendritic sites, as null direction synaptic inhibition does not

NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683

inhibit the initiation of large-amplitude parent dendritic spikes in DSGCs generated in response to steps of direct current45. When considered together with our current work, these data suggest that the dominant SAC-mediated null side inhibition acts to clamp the terminal dendritic membrane potential at a voltage sub-threshold for the generation of terminal dendritic spikes, thereby negatively gating their generation by bipolar-cell-mediated glutamatergic and SAC-mediated cholinergic excitation (Fig. 10).

A structural substrate for the dual excitatory and inhibitory roles of SACs in the retinal circuitry is well documented16,23,26,27,2936. The dendrites of ON- and OFF-SACs form dense choline acetyl transferase and vesicular ACh transporter delineated fascicles in the inner plexiform layer, the passage of which are closely followed by the dendrites of DSGCs51,53,54. Notably this mapping is most rened for small terminal DSGC dendrites, the predominant site of SAC-DSGC synaptic connectivity22. Ultrastructural analysis of murine ON-OFF-DSGCs has, however, shown that the distribution of dendro-dendritic synapses is asymmetrical, with a B13-fold greater density established by SACs positioned on the functional determined null side of ON-OFF-DSGCs17. Our results demonstrate in the simplied circuitry of the ON-sublamina of the inner plexiform layer that the postsynaptic impact of ON-SACs is position dependent, with cholinergic excitation dominating on the preferred, and GABAergic inhibition on the null side, consistent with previous results from ON-OFF-DSGCs16,2628,36. As our paired recordings, and previous results, have revealed the obligatory co-release of ACh and GABA from SACs, these data suggest that ACh may be released from, or have postsynaptic impact at, sites other than dendro-dendritic synapses, a nding that is supported by the relatively long time to onset of cholinergic PSPs, and the effects of manipulating ACh hydrolysis47,48. Furthermore, direct dendritic recordings revealed that light stimuli, and direct activation of SACs drove localized nAChR-dependent dendritic excitation, which was integrated locally in the dendritic arbour to control dendritic spike initiation. Our ndings are therefore consistent with a model in which ACh operates in a localized paracrine fashion in the IPL17,41,55, at sites constrained by the co-fasciculation of SAC and DSGC dendrites. The development of new tools for the direct visualization of ACh release and diffusion are, however, required to denitively address this issue. Based on our ndings we conclude that the excitatory and inhibitory neurotransmitters released from the dendrites of feed-forward interneurons in the output layer of the retina are integrated together with direct bipolar-cell-mediated glutamatergic excitation to drive and control the generation of dendritic spikes in DSGCs, the initiation of which determines the light stimulus sensitivity, receptive eld size and direction selectivity of action potential output (Fig. 10).

Methods

Retinal preparation. Adult pigmented or albino rabbits of either sex (Nanowie Small Animal Production Unit) were dark-adapted for 1-2 h, anesthetized with ketamine (12 mg kg 1 of body weight, intramuscular) and xylazine (12 mg kg 1, intramuscular) then overdosed with pentobarbarbitone sodium (150 mg kg 1, intravenous), before the eyes were quickly enucleated, hemisected, and placed in carbogenated Ames solution (8.8 g l 1 Ames, 1.9 g l 1 sodium bicarbonate,0.75 g l 1 glucose) at room temperature (2224 C). Large sections of the inferior peripheral retina were separated from the sclera and pigmented epithelial layer, and stored in carbogenated Ames solution at room temperature. A single section was placed in a recording chamber perfused with Ames solution at 3537 C. The Animal Ethics Committee of the University of Queensland approved all procedures.

Whole-cell recording. Dual and triple whole-cell recordings were made from visually identied DSGCs and SACs, using infrared differential interference

contrast video microscopy, with identical current-clamp ampliers congured in bridge-mode (BVC 700A, Dagan Corporation)45. Voltage and current signals were low-pass ltered at DC to 1030 kHz and sampled at 50 kHz. DSGC recording pipettes were lled with (in mM): 135 K-gluconate; 7 NaCl; 10 HEPES; 10 phosphocreatine; 2 Na2-ATP; 0.3 Na-GTP; 2 MgCl2 and 0.35 Alexa Fluor 594 (Molecular Probes) (pH 7.37.4; KOH), and had an open tip resistance of 45 MO for somatic and 1520 MO for dendritic pipettes. All DSGC recordings were made from the resting membrane potential (soma 52.70.7 mV; n 156;

dendrite 54.90.4 mV; n 47; 136.98.8 mm from soma). SAC recording

pipettes were lled with (in mM): 110 Cs-methysulphate; 5 NaCl; 10 HEPES; 10 phosphocreatine; 2 Na2-ATP; 0.3 Na-GTP; 2 MgCl2 and 0.35 Alexa Fluor 488 (Molecular Probes) (pH 7.37.4; CsOH), and had an open tip resistance of57 MO. For SAC recordings, a period of 725 min was left to allow for the intracellular blockade of potassium conductances56, before testing connectivity with DSGCs. SAC-DSGC connectivity was tested for by the generation of threshold regenerative activity evoked by the injection of short somatic positive current steps (10 ms; 0.870.11 nA; n 42; usually terminated by a negative current step:

0.50.9 s; 0.150.01 nA; inter-trial interval 2030 s; connectivity 84%) in

the presence of the glutamate receptor antagonists (6-cyano-7-nitroquinoxaline-2,3-dione; 10 mM and DL-2-amino-5-phosphonopentanoic acid; 100 mM). No correction was made for liquid junction potentials. Drugs were typically applied by addition to the perfusion medium at determined concentrations. In some experiments ACh was delivered by iontophoresis (100 mM in iontophoresis pipette) to a site co-aligned with light spot stimuli distal (9316 mm) to the dendritic recording electrode (ejection current 50200 nA; retention

current 520 nA)45,57. Exogenous AChE (0.4 U per ml)47 dissolved in Ames

solution, or Ames solution, were applied by pressure application (0.580.03 PSI for 10 min; n 11) under visual guidance from a pipette with similar

characteristics as those used for somatic recording. At the termination of each whole-cell recording, the location of the recording pipettes and neuronal morphology were examined by uorescence microscopy, and recorded using a digital camera (QImaging) or confocal microscopy (Zeiss), and reconstructed using Neurolucida software (MBF Bioscience).

Visual stimulation and analysis. Visual stimuli were generated using custom software (developed by W.R. Taylor (Oregon Health and Science University) and R.G. Smith (University of Pennsylvania)), and presented on a 800 600 pixel

OLED screen. Images were projected and focussed on the plane of photoreceptor outer segments to physiologically activate the retinal network using a 10 water

objective lens with a numerical aperture of 0.3 (Olympus)45. Standard light stimuli were applied under photopic conditions (background illumination was maintained above the level of rod saturation atE3.5 10^11 quanta cm 2 s 1) and the visual

stimuli were standardly set at 100% of the background, in some preparations mesopic conditions were established (background and visual stimuli reduced by0.06 with a neutral density lter). Moving light bars (100 300400 mm) were

standardly moved across receptive elds in 12 directions at 30 intervals, at a speed of 0.24 mm s 1 (inter-trial interval 1217 s). A range of light spot stimuli

(50200 mm) ashed (500 ms) at determined receptive eld loci were also employed. Directional selectivity was quantied by calculating a directional selectivity index of the number of action potentials, or action potential ring rate, recorded in multiple trials (n 50.5), generated by preferred and null direction

stimuli, according to the equation: preferred - null/preferred null35. Directionally

selective output responses were represented as polar plots, and a directional vector calculated. Peri-stimulus time histograms of preferred or null direction moving light bar-evoked action potential ring were binned for each segment when the moving bar (240 mm s 1) traversed 20 mm of the receptive eld. The time delay between simultaneously recorded regenerative events was measured at peak amplitude. For analysis of the time delay between simultaneously recorded spikes as a function of visual stimulus position, regenerative events from each cell were binned according to their spike times for each segment where the moving bar traversed 20 mm of the receptive eld.

Statistical analysis. Data sets were compared using a Students t-test (two-sided), analysis of variance (Tukeys multiple comparison), the Wilcoxon signed rank test or the MannWhitney test, with the test selected according to data structure. The statistical difference between amplitude distributions was determined with the KolmogorovSmironov test. Statistical signicance was accepted at Po0.05, and normality of distribution established with the KolmogorovSmironov test with a

DallalWilksonLillie post test. No statistical methods were used to predetermine sample sizes. Data collection and analysis were not randomized or performed blind to the conditions of the experiments. Numerical values are presented as means.e.m.

Data availability. The data sets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

12 NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683 ARTICLE

References

1. Denk, W., Briggman, K. L. & Helmstaedter, M. Structural neurobiology: missing link to a mechanistic understanding of neural computation. Nat. Rev. Neurosci. 13, 351358 (2012).

2. Oyster, C. W. & Barlow, H. B. Direction-selective units in rabbit retina: distribution of preferred directions. Science 155, 841842 (1967).

3. Barlow, H. B. & Levick, W. R. The mechanism of directionally selective units in rabbits retina. J. Physiol. 178, 477504 (1965).

4. Borst, A. & Euler, T. Seeing things in motion: models, circuits, and mechanisms. Neuron 71, 974994 (2011).

5. Masland, R. H. The neuronal organization of the retina. Neuron 76, 266280 (2012).

6. Vaney, D. I., Sivyer, B. & Taylor, W. R. Direction selectivity in the retina: symmetry and asymmetry in structure and function. Nat. Rev. Neurosci. 13, 194208 (2012).

7. Beier, K. T. et al. Transsynaptic tracing with vesicular stomatitis virus reveals novel retinal circuitry. J. Neurosci. 33, 3551 (2013).

8. Cruz-Martin, A. et al. A dedicated circuit links direction-selective retinal ganglion cells to the primary visual cortex. Nature 507, 358361 (2014).

9. Sun, L. O. et al. Functional assembly of accessory optic system circuitry critical for compensatory eye movements. Neuron 86, 971984 (2015).

10. Barlow, H. B. & Hill, R. M. Selective sensitivity to direction of movement in ganglion cells of the rabbit retina. Science 139, 412414 (1963).

11. Chen, M., Lee, S., Park, S. J., Looger, L. L. & Zhou, Z. J. Receptive eld properties of bipolar cell axon terminals in direction-selective sublaminas of the mouse retina. J. Neurophysiol. 112, 19501962 (2014).

12. Park, S. J., Kim, I. J., Looger, L. L., Demb, J. B. & Borghuis, B. G. Excitatory synaptic inputs to mouse on-off direction-selective retinal ganglion cells lack direction tuning. J. Neurosci. 34, 39763981 (2014).

13. Yonehara, K. et al. The rst stage of cardinal direction selectivity is localized to the dendrites of retinal ganglion cells. Neuron 79, 10781085 (2013).

14. Euler, T., Detwiler, P. B. & Denk, W. Directionally selective calcium signals in dendrites of starburst amacrine cells. Nature 418, 845852 (2002).

15. Fried, S. I., Munch, T. A. & Werblin, F. S. Mechanisms and circuitry underlying directional selectivity in the retina. Nature 420, 411414 (2002).

16. Lee, S., Kim, K. & Zhou, Z. J. Role of ACh-GABA cotransmission in detecting image motion and motion direction. Neuron 68, 11591172 (2010).

17. Briggman, K. L., Helmstaedter, M. & Denk, W. Wiring specicity in the direction-selectivity circuit of the retina. Nature 471, 183188 (2011).

18. Yoshida, K. et al. A key role of starburst amacrine cells in originating retinal directional selectivity and optokinetic eye movement. Neuron 30, 771780 (2001).

19. Lefebvre, J. L., Kostadinov, D., Chen, W. V., Maniatis, T. & Sanes, J. R. Protocadherins mediate dendritic self-avoidance in the mammalian nervous system. Nature 488, 517521 (2012).

20. Miller, R. F. & Bloomeld, S. A. Electroanatomy of a unique amacrine cell in the rabbit retina. Proc. Natl Acad. Sci. USA 80, 30693073 (1983).

21. Tukker, J. J., Taylor, W. R. & Smith, R. G. Direction selectivity in a model of the starburst amacrine cell. Vis. Neurosci. 21, 611625 (2004).

22. Famiglietti, E. V. Synaptic organization of starburst amacrine cells in rabbit retina: analysis of serial thin sections by electron microscopy and graphic reconstruction. J. Comp. Neurol. 309, 4070 (1991).

23. He, S. & Masland, R. H. Retinal direction selectivity after targeted laser ablation of starburst amacrine cells. Nature 389, 378382 (1997).

24. Vlasits, A. L. et al. A role for synaptic input distribution in a dendritic computation of motion direction in the retina. Neuron 89, 13171330 (2016).

25. Massey, S. C., Linn, D. M., Kittila, C. A. & Mirza, W. Contributions of GABAA receptors and GABAC receptors to acetylcholine release and directional selectivity in the rabbit retina. Vis. Neurosci. 14, 939948 (1997).

26. Yonehara, K. et al. Spatially asymmetric reorganization of inhibition establishes a motion-sensitive circuit. Nature 469, 407410 (2011).

27. Wei, W., Hamby, A. M., Zhou, K. & Feller, M. B. Development of asymmetric inhibition underlying direction selectivity in the retina. Nature 469, 402406 (2011).

28. Morrie, R. D. & Feller, M. B. An asymmetric increase in inhibitory synapse number underlies the development of a direction selective circuit in the retina.J. Neurosci. 35, 92819286 (2015).29. Ariel, M. & Daw, N. W. Pharmacological analysis of directionally sensitive rabbit retinal ganglion cells. J. Physiol. 324, 161185 (1982).

30. Kittila, C. A. & Massey, S. C. Pharmacology of directionally selective ganglion cells in the rabbit retina. J. Neurophysiol. 77, 675689 (1997).

31. Brecha, N., Johnson, D., Peichl, L. & Wassle, H. Cholinergic amacrine cells of the rabbit retina contain glutamate decarboxylase and gamma-aminobutyrate immunoreactivity. Proc. Natl Acad. Sci. USA 85, 61876191 (1988).

32. Vaney, D. I. & Young, H. M. GABA-like immunoreactivity in NADPH-diaphorase amacrine cells of the rabbit retina. Brain Res. 474, 380385 (1988).

33. OMalley, D. M. & Masland, R. H. Co-release of acetylcholine and gamma-aminobutyric acid by a retinal neuron. Proc. Natl Acad. Sci. USA 86, 34143418 (1989).

34. Feller, M. B., Wellis, D. P., Stellwagen, D., Werblin, F. S. & Shatz, C. J. Requirement for cholinergic synaptic transmission in the propagation of spontaneous retinal waves. Science 272, 11821187 (1996).

35. Fried, S. I., Munch, T. A. & Werblin, F. S. Directional selectivity is formed at multiple levels by laterally offset inhibition in the rabbit retina. Neuron 46, 117127 (2005).

36. Pei, Z. et al. Conditional knock-out of vesicular GABA transporter gene from starburst amacrine cells reveals the contributions of multiple synaptic mechanisms underlying direction selectivity in the retina. J. Neurosci. 35, 1321913232 (2015).

37. Lipin, M. Y., Taylor, W. R. & Smith, R. G. Inhibitory input to the direction-selective ganglion cell is saturated at low contrast. J. Neurophysiol. 114, 927941 (2015).

38. Weng, S., Sun, W. & He, S. Identication of ON-OFF direction-selective ganglion cells in the mouse retina. J. Physiol. 562, 915923 (2005).

39. Chiao, C. C. & Masland, R. H. Starburst cells nondirectionally facilitate the responses of direction-selective retinal ganglion cells. J. Neurosci. 22, 1050910513 (2002).

40. Grzywacz, N. M., Amthor, F. R. & Merwine, D. K. Necessity of acetylcholine for retinal directionally selective responses to drifting gratings in rabbit. J. Physiol. 512, 575581 (1998).

41. Sethuramanujam, S. et al. A central role for mixed acetylcholine/GABA transmission in direction coding in the retina. Neuron 90, 12431256 (2016).

42. Poleg-Polsky, A. & Diamond, J. S. NMDA receptors multiplicatively scale visual signals and enhance directional motion discrimination in retinal ganglion cells. Neuron 89, 12771290 (2016).

43. Oesch, N., Euler, T. & Taylor, W. R. Direction-selective dendritic action potentials in rabbit retina. Neuron 47, 739750 (2005).

44. Schachter, M. J., Oesch, N., Smith, R. G. & Taylor, W. R. Dendritic spikes amplify the synaptic signal to enhance detection of motion in a simulation of the direction-selective ganglion cell. PLoS Comput. Biol. 6, 1000899 (2010).

45. Sivyer, B. & Williams, S. R. Direction selectivity is computed by active dendritic integration in retinal ganglion cells. Nat. Neurosci. 16, 18481856 (2013).46. Longordo, F., To, M. S., Ikeda, K. & Stuart, G. J. Sublinear integration underlies binocular processing in primary visual cortex. Nat. Neurosci. 16, 714723 (2013).

47. Bennett, C., Arroyo, S., Berns, D. & Hestrin, S. Mechanisms generating dual-component nicotinic EPSCs in cortical interneurons. J. Neurosci. 32, 1728717296 (2012).

48. Lamotte dIncamps, B., Krejci, E. & Ascher, P. Mechanisms shaping the slow nicotinic synaptic current at the motoneuron-renshaw cell synapse. J. Neurosci. 32, 84138423 (2012).

49. Zhou, F. M., Liang, Y. & Dani, J. A. Endogenous nicotinic cholinergic activity regulates dopamine release in the striatum. Nat. Neurosci. 4, 12241229 (2001).

50. Yang, G. & Masland, R. H. Receptive elds and dendritic structure of directionally selective retinal ganglion cells. J. Neurosci. 14, 52675280 (1994).

51. Brandon, C. Cholinergic neurons in the rabbit retina: dendritic branching and ultrastructural connectivity. Brain Res. 426, 119130 (1987).

52. Famiglietti, E. V. Synaptic organization of complex ganglion cells in rabbit retina: type and arrangement of inputs to directionally selective and local-edge-detector cells. J. Comp. Neurol. 484, 357391 (2005).

53. Vaney, D. I. & Pow, D. V. The dendritic architecture of the cholinergic plexus in the rabbit retina: selective labeling by glycine accumulation in the presence of sarcosine. J. Comp. Neurol. 421, 113 (2000).

54. Dong, W., Sun, W., Zhang, Y., Chen, X. & He, S. Dendritic relationship between starburst amacrine cells and direction-selective ganglion cells in the rabbit retina. J. Physiol. 556, 1117 (2004).

55. Ford, K. J., Felix, A. L. & Feller, M. B. Cellular mechanisms underlying spatiotemporal features of cholinergic retinal waves. J. Neurosci. 32, 850863 (2012).

56. Taylor, W. R. & Wassle, H. Receptive eld properties of starburst cholinergic amacrine cells in the rabbit retina. Eur. J. Neurosci. 7, 23082321 (1995).

57. Brombas, A., Fletcher, L. N. & Williams, S. R. Activity-dependent modulation of layer 1 inhibitory neocortical circuits by acetylcholine. J. Neurosci. 34, 19321941 (2014).

Acknowledgements

This work was supported by grants from the Australian Research Council (FT100100502), National Health and Medical Research Council (APP1082257) and the Hand Heart Pocket Foundation to S.R.W. We are grateful to Rowan Tweedale and Ben Sivyer for their constructive comments.

NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15683

Author contributions

S.R.W. conceived the project, S.R.W. and A.B. designed the experiments, A.B. and S.K.-d.C. conducted experiments, and A.B., S.K.-d.C., E.J.C.-W. and S.R.W. analysed the data. S.R.W. wrote the paper with input from all authors.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Brombas, A. et al. Dendro-dendritic cholinergic excitation controls dendritic spike initiation in retinal ganglion cells. Nat. Commun. 8, 15683 doi: 10.1038/ncomms15683 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the articles Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/ http://creativecommons.org/licenses/by/4.0/

Web End =licenses/by/4.0/

r The Author(s) 2017

14 NATURE COMMUNICATIONS | 8:15683 | DOI: 10.1038/ncomms15683 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

Word count: 11054

Show less

Abstract

Translate

The retina processes visual images to compute features such as the direction of image motion. Starburst amacrine cells (SACs), axonless feed-forward interneurons, are essential components of the retinal direction-selective circuitry. Recent work has highlighted that SAC-mediated dendro-dendritic inhibition controls the action potential output of direction-selective ganglion cells (DSGCs) by vetoing dendritic spike initiation. However, SACs co-release GABA and the excitatory neurotransmitter acetylcholine at dendritic sites. Here we use direct dendritic recordings to show that preferred direction light stimuli evoke SAC-mediated acetylcholine release, which powerfully controls the stimulus sensitivity, receptive field size and action potential output of ON-DSGCs by acting as an excitatory drive for the initiation of dendritic spikes. Consistent with this, paired recordings reveal that the activation of single ON-SACs drove dendritic spike generation, because of predominate cholinergic excitation received on the preferred side of ON-DSGCs. Thus, dendro-dendritic release of neurotransmitters from SACs bi-directionally gate dendritic spike initiation to control the directionally selective action potential output of retinal ganglion cells.

Details

Title

Dendro-dendritic cholinergic excitation controls dendritic spike initiation in retinal ganglion cells

Author

Brombas, A; Kalita-de Croft, S; Cooper-williams, E J; Williams, S R

Pages

15683

Publication year

2017

Publication date

Jun 2017

Publisher

Nature Publishing Group

e-ISSN

20411723

Source type

Scholarly Journal

Language of publication

English

DOI

https://doi.org/10.1038/ncomms15683

ProQuest document ID

1906452720

Dendro-dendritic cholinergic excitation controls dendritic spike initiation in retinal ganglion cells

Jump to:

Full Text

Abstract

Details

Suggested sources