Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 17, 355-79

Exploring the Retinal Connectome

Affiliations

Exploring the Retinal Connectome

James R Anderson et al. Mol Vis.

Abstract

Purpose: A connectome is a comprehensive description of synaptic connectivity for a neural domain. Our goal was to produce a connectome data set for the inner plexiform layer of the mammalian retina. This paper describes our first retinal connectome, validates the method, and provides key initial findings.

Methods: We acquired and assembled a 16.5 terabyte connectome data set RC1 for the rabbit retina at ≈ 2 nm resolution using automated transmission electron microscope imaging, automated mosaicking, and automated volume registration. RC1 represents a column of tissue 0.25 mm in diameter, spanning the inner nuclear, inner plexiform, and ganglion cell layers. To enhance ultrastructural tracing, we included molecular markers for 4-aminobutyrate (GABA), glutamate, glycine, taurine, glutamine, and the in vivo activity marker, 1-amino-4-guanidobutane. This enabled us to distinguish GABAergic and glycinergic amacrine cells; to identify ON bipolar cells coupled to glycinergic cells; and to discriminate different kinds of bipolar, amacrine, and ganglion cells based on their molecular signatures and activity. The data set was explored and annotated with Viking, our multiuser navigation tool. Annotations were exported to additional applications to render cells, visualize network graphs, and query the database.

Results: Exploration of RC1 showed that the 2 nm resolution readily recapitulated well known connections and revealed several new features of retinal organization: (1) The well known AII amacrine cell pathway displayed more complexity than previously reported, with no less than 17 distinct signaling modes, including ribbon synapse inputs from OFF bipolar cells, wide-field ON cone bipolar cells and rod bipolar cells, and extensive input from cone-pathway amacrine cells. (2) The axons of most cone bipolar cells formed a distinct signal integration compartment, with ON cone bipolar cell axonal synapses targeting diverse cell types. Both ON and OFF bipolar cells receive axonal veto synapses. (3) Chains of conventional synapses were very common, with intercalated glycinergic-GABAergic chains and very long chains associated with starburst amacrine cells. Glycinergic amacrine cells clearly play a major role in ON-OFF crossover inhibition. (4) Molecular and excitation mapping clearly segregates ultrastructurally defined bipolar cell groups into different response clusters. (5) Finally, low-resolution electron or optical imaging cannot reliably map synaptic connections by process geometry, as adjacency without synaptic contact is abundant in the retina. Only direct visualization of synapses and gap junctions suffices.

Conclusions: Connectome assembly and analysis using conventional transmission electron microscopy is now practical for network discovery. Our surveys of volume RC1 demonstrate that previously studied systems such as the AII amacrine cell network involve more network motifs than previously known. The AII network, primarily considered a scotopic pathway, clearly derives ribbon synapse input from photopic ON and OFF cone bipolar cell networks and extensive photopic GABAergic amacrine cell inputs. Further, bipolar cells show extensive inputs and outputs along their axons, similar to multistratified nonmammalian bipolar cells. Physiologic evidence of significant ON-OFF channel crossover is strongly supported by our anatomic data, showing alternating glycine-to-GABA paths. Long chains of amacrine cell networks likely arise from homocellular GABAergic synapses between starburst amacrine cells. Deeper analysis of RC1 offers the opportunity for more complete descriptions of specific networks.

Figures

Figure 1
Figure 1
The vertical bars represent section structure of the 0.25 mm diameter column of 341 transmission electron microscope (TEM) data set slices in volume RC1, imaged at 2 nm resolution. The horizontal bars represent capstone and individual intercalated computational molecular phenotyping (CMP) sections for molecular tagging. CMP images were captured at 70 nm/pixel and upsampled to 2.18 nm/pixel in Viking. The CMP skips in the TEM sequence were intentional and created no problems in process tracking. The gaps indicate unplanned capture skips in due to defects that prevented imaging. A block refacing event at section 306 caused a significant (350–400 nm) loss in the ganglion cell layer. At bottom, a single grid carrying three imaged sections is shown. The gold spot on each section indicates the area captured, each spot averaging over 1,000 individual captures at a magnification of 5,000×.
Figure 2
Figure 2
Connectome RC1 data sets were visualized by fusing transmission electron microscope (TEM) images and computational molecular phenotyping (CMP) signals. A: TEM section 001 is a near-horizontal plane section through the inner nuclear layer (INL) of the retina, visualized with 4-aminobutyrate(GABA).glycine.glutamate (γGE) → red.green.blue (RGB) transparency mapping, displaying retinal neurons, and a dark gold alpha (α) channel derived from taurine and glutamine (τQ) channels marking retinal glia (γGE.τQ) → RGB.α (see Methods). GABA+ (red) neurons are amacrine cells, while glycine+ (green) neurons are either amacrine or an ON cone bipolar cell subset. Glutamate+ (blue) neurons are largely bipolar cells. The image width is 243 μm. B: TEM section 030 is a connectome slice roughly 2.5 μm deeper in the INL, visualized with a GABA.glycine → magenta.green transparency (γG=MG; see Methods). The circled cells represent 12 validated AII amacrine cells, 8 visible in section 030 (solid circles) and 4 originating in a plane beneath section 030 (dashed circles). The image width is 243 μm. C: TEM section 184 with orange GABA (γ) overlay (see Methods) shows that all bipolar cell terminals are GABA- (boxes), as are lobular appendages of AII amacrine cells (circle), a descending portion of AII amacrine cell C4835, and the radial fibers of Müller cells (asterisks). Numerous GABA+ processes and a weakly labeled ganglion cell dendrite (arrow) are present. The image scale is 5 μm.
Figure 3
Figure 3
The computational molecular phenotyping (CMP) matrix for volume RC1 is bounded by CMP data sets and intercalated every thirty sections with ultrathin CMP sections that map molecular tags onto transmission electron microscope (TEM) data. Each row of 2 or 3 fields contains a TEM slice with its associated index number, one or more optical CMP channels composed of one to four molecular tags, and for fields in the inner nuclear and ganglion cell layer, overlay images of the CMP data registered onto the TEM channel. Each disc is 243 μm in diameter. The matrix was assembled from 32x down-scaled TEM data sets (70 nm/pixel). This represents a threefold oversampling of the optical data. The abbreviations and color key for the figure are: B, 1-amino-4-guanidobutane, color=cyan; E, glutamate, color=blue; G glycine, color=green; γ, GABA, color=red (slices 001, 371) or magenta (slices 62, 184, 312); τ taurine, color=gold (slice 001), red (slice 371), or orange (slices 122, 244); γG → γ magenta: G green; γGE → γ red, G green, E blue; γGEτ → γ red, G green, E blue + gold τ alpha channel overlay mask; γBE → γ red, B green, E blue; τQE → τ red, Q green, E blue.
Figure 4
Figure 4
A fragment of the mammalian AII amacrine cell network is visualized by rendering and transmission electron microscope (TEM). A: Viking-rendered AII amacrine cells (C476, C514, C2610) and rod (B518), OFF (C478) and ON (C1724) bipolar cells form a local network. Each bipolar cell was chosen to mark the center of the cone OFF, cone ON and rod driven zones of the inner plexiform layer. The small red, blue and yellow details represent postsynaptic, presynaptic and gap junction contact sites. They are scaled to true size, so most of them are below the figure’s resolution. Only the largest are visible (scale, 20 μm). B: Rod BC B518 (blue) presynaptic (r) to AII amacrine cell C476 (green) and γ+ AC C4942 (orange); γ+ ACs C4941 and C4942 are presynaptic (arrows) to B518. C: AII amacrine cell C514 (green) to OFF cone BC C478 (blue) synapses (arrows). C514 makes conventional synapses onto C478 at its terminal swelling and fine inter-varicosity processes (box, 6 sections away). D: AII amacrine cell C514 (green) and γ+ AC C5285 (orange) are both presynaptic (arrows) to OFF GC C5150 (magenta). E: Heterocellular coupling (between black arrows) between ON cone BC C1724 (blue) and AII amacrine cell C514 (green). The inset (width 169 nm) is a high resolution tilt TEM image of the gap junction. F: Homocellular coupling (arrows) occurs among AII amacrine cells C514, C476, and C3679. The scales for panels B-F are 500 nm.
Figure 5
Figure 5
Numerous synaptic connections converge on AII amacrine cell C514. A: The central image is a 3D VikingPlot rendering of C514 (scale, 10 μm). Surrounding the cell are instances of different synaptic connections made by C514. In each panel, green profiles are C514, orange profiles are γ+ ACs, azure profiles are BCs, blue profiles are GCs, red profiles are γ-, G- and glutamate+. Arrows indicate direction of synaptic signaling and double arrows indicate gap junctions. B: C514 is postsynaptic to large γ-/G- axosomatic synapses likely deriving from TH1 (tyrosine hydroxylase immunopositive type 1) cells. However, the architecture of the synapse is of a fast conventional transmitter, likely glutamate (see Figure 7 and Figure 8). C: C514 is postsynaptic to γ+ / peptidergic processes in the OFF sublayer at points where dense-core, peptide vesicles form fusion complexes. D: C514 is postsynaptic to γ+ / peptide processes in the OFF sublayer at a conventional inhibitory synapse. E: C514 is postsynaptic to conventional, non-peptide γ+ processes in the OFF sublayer. F: C514 is both presynaptic and postsynaptic to an OFF cone bipolar cell. G: C514 is presynaptic to an OFF ganglion cell. H: C514 is presynaptic to an OFF bipolar cell. I: C514 is coupled to an ON cone bipolar cell. J: C514 is postsynaptic to an ON cone bipolar cell. K: C514 is postsynaptic to a γ+ amacrine cell. L: C514 is postsynaptic to a γ+ type AI amacrine cell. M: C514 is postsynaptic to a rod bipolar cell and coupled to another AII amacrine cell. The scales for panels B-M are 500 nm.
Figure 6
Figure 6
Twelve different neuronal classes generate at least seventeen distinct input-output motifs involving AII amacrine cells: 1, 2 represent the dopamine and glutamate inputs from TH1 axonal cells (AxCs); 3 and 4 are outputs to OFF ganglion cells (GCs) and OFF γ+ cone amacrine cells (AC); 5,6 are inputs from dual GABA/peptide amacrine cells; 7,8 is output to another class of γ+ amacrine cells; 9,10 are outputs to and inputs from OFF cone CBa bipolar cells (BCs); 11 is input from ON cone γ+ amacrine cells; 12 is coupling to several classes of ON cone CBb bipolar cells; 13 and 14 are coupling and ribbon inputs from CBwb BCs; 15 is input from AI amacrine cells; 16 is input from rod bipolar cells; 17 is coupling with other AII amacrine cells.
Figure 7
Figure 7
TH+ (tyrosine hydroxylase immunopositive) cells have glutamatergic, not GABAergic signatures. The nine panels show nine TH+ cells from a single rabbit retina (A-I), probed for TH, glutamate and GABA in serial 200 nm sections. Each panel shows four mappings: upper left TH (yellow) + glutamate (blue), upper right TH (yellow) + GABA (red), lower left glutamate alone (cyan), lower right GABA alone (yellow). The location of each TH+ cell is circled. Each TH+ cell has a glutamate signal higher than the surrounding amacrine cell somas and equivalent to that of a ganglion cell. TH+ cells have no measurable GABA signal. Scale, 10 μm.
Figure 8
Figure 8
AII amacrine cell 514 in RC1 displays axosomatic synaptic input. A: Section 58 (z 58) shows two axosomatic synapses (circled, Scales, 10 and 1 μm). B: Section 61 (z 61), with GABA overlay in orange, shows that the axosomatic synaptic terminal has the same GABA negative signal as the AII amacrine cell but is flanked by orange GABA+ processes (Scale, 1 μm). C: A serial section series from section 57 (z 57) through 67 (z 67), omitting section 61 shown in panel B, shows that axosomatic synapses are formed at z 57 & z 58. Several sections show large dense-core vesicles (circle in z 67) characteristic of TH cells (Scale, 1 μm). D: Goniometric tilt re-imaging of the oblique synaptic contact in panel C z 58. A 55° tilt aligns the axosomatic contact membrane, and clearly shows a characteristic 10 nm synaptic gap and polarity (arrow) targeting AII amacrine cell 514.
Figure 9
Figure 9
Gap junctions between identified cells are visualized by transmission electron microscope (TEM) in the RC1 data set and after re-imaging at 0.3 nm resolution. Panels A-G are native Viking images showing putative gap junctions (box areas). Each panel is 3570 nm wide. The numbers denote the location of the image in the RC1 image volume (section number, x location, y location). Panels H-N are the same locations re-imaged at 40,000× on Kodak 4498 Electron Microscopy Film and digitized at 2540 dpi and 16-bits. Each panel is 602 nm wide. Panels O-U are validated gap junctions scaled from the boxed regions in H-N. Each panel is 150 nm wide.
Figure 10
Figure 10
Gap junctions can be displayed as transmission electron microscope (TEM) image density profiles for homocellular (black lines) and heterocellular (dotted lines) pairings. A central 12–13 nm wide pentalaminar zone with three sharp dense bands representing membrane protein density is common to all retinal gap junctions. But the flanking cytoplasmic compartments differ markedly. Homocellular gap junctions between pairs of AII amacrine cells show little protein density near the membrane. Conversely, heterocellular gap junctions between AII amacrine cells and cone bipolar cells show thick bands of protein density extending about 10 nm and over 15 nm from the bipolar cell and amacrine cell faces, respectively.
Figure 11
Figure 11
RC1 contains novel retinal networks. A: A Viking rendering of γ+ amacrine cell C5303 shows that it is postsynaptic to ON cone bipolar cells at axonal ribbon sites (circles C and D), presynaptic to ON cone bipolar cell C483, and co-stratifies with ON starburst amacrine cell C4890 (scale 20 μm). The circles indicate corresponding transmission electron microscope (TEM) images. B: Axonal ribbons (r) from ON cone BC C180 target AC neurites as the axon bifurcates in mid-inner plexiform layer. C: ON cone BC C177 makes axonal nanoribbon contacts onto γ+ amacrine cell C5303. D: Axonal ribbon synapses from C166 target cell C5303. C166 receives an axonal veto synapse from a yet unidentified amacrine cell (arrow). E: A large γ+ amacrine cell makes an axonal veto synapse onto an OFF cone bipolar cell axon. F: An axonal cistern contact is formed between ON cone bipolar cell C168 onto an amacrine cell process. E denotes glutamate; γ denotes GABA; and question mark denotes unknown. The letter colors match the profiles in the image. The scales for images B-F are 500 nm.
Figure 12
Figure 12
Axonal ribbon synapses (black circles) and veto synapses (white circles) are distributed along the axons of 105 reconstructed bipolar cells. Grey lines are OFF cone bipolar cells terminating high in the inner plexiform layer. Green lines are G+ ON cone bipolar cells. Magenta lines are rod bipolar cells with extensive AI and AII amacrine cell contacts. Each line indicates the length of the axon from its point of entry to its terminal expansion level in the inner plexiform layer (IPL). The physical cells are longer as we show only the axon, not the entire terminal arbor. The break in the lower panel axons represents the approximate position of the lower limit of identified OFF bipolar cell processes. Importantly, some bona fide ON bipolar cells axons are shorter than the longest OFF bipolar cell axons and terminals and they co-mingle.
Figure 13
Figure 13
AI and AII amacrine cells display complex networks. A: Transmission electron microscope (TEM) section 062 shows γ+ AI AC 4943 and neighboring BCs and ACs (green) with a magenta GABA overlay. B: γ+ AI amacrine cell 4943 (red) spans the width of the RC1 volume and some of the cells associated with it are rod bipolar cells 518, 5923, 8586; OFF cone bipolar cell 5539; G+ AII amacrine cell, 476; G+ OFF amacrine cell, 7188. The circles over the proximal dendrites of amacrine cell 4943 denote sites of multiple amacrine cell synaptic inputs. C: Rod bipolar cell 8586 synapses onto AI AC 4943. D: AI AC 4943 onto synapses rod bipolar cell 5923. E: G+ AC 7188 makes a conventional synapse on AI AC 4943. F: AI AC 4943 receives serial conventional synapses. E denotes glutamate; G denotes glycine. The letter colors match the profiles in the image. The main panel is from section 168 and the yellow panel insert is from section 165.
Figure 14
Figure 14
The rabbit inner plexiform layer contains synaptic chains up to six stages long. The chain starts with OFF cone bipolar cells (1) targeting two amacrine cells (2) that both converge on γ+ amacrine cell C6011. C6011 then targets two different classes of OFF cone bipolar cells and (3) a G+ AII amacrine cell, which then targets (4) γ+ amacrine cell C174, ultimately pre-synaptic (5) to another OFF cone bipolar cell that drives (6) a retinal ganglion cell.
Figure 15
Figure 15
The ON starburst amacrine cell stratum of the inner plexiform layer also contains long synaptic chains. A: Viking screen capture at locus x 59627, y 34285, z 240 shows four amacrine cells (A1, A2, A3, A4) forming a concatenated chain (yellow circles) in the stratum occupied by the dendrites of ON starburst amacrine cell C4890 (magenta). C4890 is also postsynaptic to AC1 (orange circle) in this and other sections. The synaptic chain was re-imaged at high-resolution (0.5 nm/pixel) and goniometric tilt in panels B-D. B: The AC1 → AC2 synapse viewed with 55° tilt. C: The AC2 → AC3 synapse viewed with 20° tilt. D: The AC3 → AC4 synapse viewed with 20° tilt. The scale for panel A is 1000 nm; for panels B-D it is 500 nm.
Figure 16
Figure 16
AGB mapping allows visualization of bipolar cell light responses. A: Slice 001 transmission electron microscope (TEM) with an overlay of glycine:AGB → magenta.green mapping shows a collection of mapped bipolar cells. B: Slice 001 with glycine:AGB → magenta.green mapping alone shows only the molecular signatures of the cells. C: Slice 001 with greyscale glycine intensity mapping reveals both glycinergic amacrine cells and bipolar cells coupled to AII amacrine cells. D: Slice 001 with greyscale AGB intensity mapping displays the light responses of all bipolar cells. Six identified bipolar cells are circled: one G+ WF (wide-field) bipolar cell (170), two OFF bipolar cells (173, 175), two rod bipolar cells (5595, 5614) and one non-coupled ON layer bipolar cell (5292).
Figure 17
Figure 17
Bivariate glycine (abscissa) and photopic light-stimulated AGB signals (ordinate) for the validated bipolar cells shown in Figure 14 form unique clusters. OFF cone bipolar cells (black) have some of the strongest light-driven responses, while the mean response of most coupled ON cone bipolar cells (orange) is weaker. WF bipolar cells form a small subgroup within the entire G+ ON bipolar cell cluster (blue). Rod bipolar cells show no significant response. Conversely, three non-coupled, G- cone BCs terminating high in the ON layer have extremely strong responses (white). The stimulus regime was a 3 Hz pulse train of 3 yellow/black pulse cycles followed by one blue/black pulse cycle with a 50% duty cycle over 90 min.
Figure 18
Figure 18
Neuronal processes are often apposed without synaptic contacts. A: A dendrite (purple) from ON ganglion cell C7594 courses through the inner plexiform layer and is physically apposed to many cells with which it never makes synaptic contact, such as an amacrine cell dendrite (AC, orange). B: An enlarged view of the apposition shown in panel A. The amacrine cell membrane is directly apposed to ganglion cell C7594 with no intervening glial processes (M). The image is centered on x 71300, y 46622, z 279 in RC1, the apposition spans sections 275–286 (770–990 nm), and is at least 1450 nm long in the XY plane. C: An apposition between an ON cone bipolar cell (BC, azure) and ganglion cell C7594 at a more distal location in RC1. E denotes glutamate; γ denotes GABA; τ denotes taurine. The letter colors match the profiles in the image. The bipolar cell makes a ribbon monad onto a very small amacrine cell dendrite, but is never presynaptic to the ganglion cell. The image is centered on x 73722, y 53700, z 258 in RC1, spans sections 256–269 (910–1170 nm), and is at least 2200 nm long in the XY plane based on serial tracking. Scales, 1000 nm.

Similar articles

See all similar articles

Cited by 68 PubMed Central articles

See all "Cited by" articles

References

    1. Sporns O, Tononi G, Kötter R. The Human Connectome: A structural description of the human brain. PLOS Comput Biol. 2005;1:e42. - PMC - PubMed
    1. Lichtman JW, Livet J, Sanes JR. A technicolour approach to the connectome. Nat Rev Neurosci. 2008;9:417–22. - PMC - PubMed
    1. Anderson JR, Jones BW, Yang JH, Shaw MV, Watt CB, Koshevoy P, Spaltenstein J, Jurrus E. U V K, Whitaker RT, Mastronarde D, Tasdizen T, Marc RE. A computational framework for ultrastructural mapping of neural circuitry. PLoS Biol. 2009;7:e1000074. - PMC - PubMed
    1. Anderson JR, Mohammed S, Grimm B, Jones BW, Koshevoy P, Tasdizen T, Whitaker R, Marc RE. The Viking viewer for connectomics: scalable multi-user annotation and summarization of large volume data sets. J Microsc. 2010;241:13–28. - PMC - PubMed
    1. Marc RE, Murry RF, Basinger SF. Pattern recognition of amino acid signatures in retinal neurons. J Neurosci. 1995;15:5106–29. - PMC - PubMed

Publication types

LinkOut - more resources

Feedback