In the visual system, rod photoreceptors register dim light signals,

Transcription

1 Direct rod input to cone BCs and direct cone input to rod BCs challenge the traditional view of mammalian BC circuitry Ji-Jie Pang a, Fan Gao a, Janis Lem b, Debra E. Bramblett c, David L. Paul d, and Samuel M. Wu a,1 a Cullen Eye Institute, Baylor College of Medicine, Houston, TX 77030; b Department of Ophthalmology, Programs in Genetics, Neuroscience, Cell and Molecular and Developmental Biology, Tufts University School of Medicine, Boston, MA 02111; c Department of Medical Education, Texas Tech University Health Sciences Center, El Paso, TX 79905; and d Department of Neurobiology, Harvard Medical School, Boston, MA Edited by John E. Dowling, Harvard University, Cambridge, MA, and approved November 18, 2009 (received for review June 29, 2009) Bipolar cells are the central neurons of the retina that transmit visual signals from rod and cone photoreceptors to third-order neurons in the inner retina and the brain. A dogma set forth by early anatomical studies is that bipolar cells in mammalian retinas receive segregated rod/cone synaptic inputs (either from rods or from cones), and here, we present evidence that challenges this traditional view. By analyzing light-evoked cation currents from morphologically identified depolarizing bipolar cells (DBCs) in the wild-type and three pathway-specific knockout mice (rod transducin knockout [Trα / ], connexin36 knockout [Cx36 / ], and transcription factor beta4 knockout [Bhlhb4 / ]), we show that a subpopulation of rod DBCs (DBC R2 s) receives substantial input directly from cones and a subpopulation of cone DBCs (DBC C1 s) receives substantial input directly from rods. These results provide evidence of the existence of functional rod-dbc C and cone-dbc R synaptic pathways in the mouse retina as well as the previously proposed rod hyperpolarizing bipolar-cells pathway. This is grounds for revising the mammalian rod/cone bipolar cell dogma. light-evoked cation and chloride currents axon terminal stratification connexin36 wild-type and mutant mice depolarizing bipolar cell In the visual system, rod photoreceptors register dim light signals, and cone photoreceptors encode brighter light signals (1). Bipolar cells (BCs) are the second-order neurons in the retina that receive light-elicited signals from rod and cone photoreceptors and transmit them to amacrine cells (ACs) and ganglion cells (GCs) in the inner retina (1, 2). Early anatomical studies have shown that mammalian rods make synaptic contacts with only one type of bipolar cell, the rod depolarizing bipolar cell (DBC R ), whereas cones make synaptic contacts with eight to nine types of cone depolarizing (DBC C s) or hyperpolarizing bipolar cells (HBC C s) (3 5). Additionally, DBC R s do not make output synapses directly on GCs, the output neurons of the retina, but on the AII amacrine cells (AIIACs), which make electrical synapses (with connexin36 at least at the AIIAC side) (6, 7) with DBC C s (that send signals to ON GCs) and inhibitory glycinergic synapses with HBC C s and OFF GCs (8 10). Therefore, in addition to direct cone synaptic inputs, DBC C s receive rod-mediated signals from AIIAC-DBC C electrical synapses, and HBC C s receive rod-mediated signals from the AIIAC-HBC C chemical synapses (11, 12). This AIIAC-mediated rod/cone signal mixing is named the primary rod-to-cone signaling pathway (13). Furthermore, rods and cones are electrically coupled with each other, possibly through connexin36-mediated gap junctions (6, 14, 15), and such rod/cone-signal mixing at the photoreceptor level is named the secondary rod-cone pathway (13). The rod and cone bipolar cell-signaling circuitry described above has been considered for many years as the general organizational plan (to a certain degree as the dogma) for all mammals (3, 5, 16). Evidence from recent studies, however, begins to challenge this view. In the rabbit retina, for example, when the rod-dbc R synapses are blocked by L-AP4, rod inputs to OFF GCs persist, indicative of an alternative rod-hbc-off GC synaptic pathway (17). Studies on normal and coneless transgenic mice and rabbits indicate that rods send signals directly to certain types of HBCs (13, 18). Moreover, recent electron microscopic analysis suggests that rods in the mouse retina make chemical synapses on some HBC C s and DBC C s (19, 20). These results suggest that functional pathways mediating DBC and HBC responses in mammalian retinas may be more complex than the general plan set forth by earlier anatomical studies. It is crucial to systematically investigate rod and cone inputs to various types of BCs and to determine the synaptic pathways by which rod and cone signals are transmitted to BCs in the mammalian retina. It is of great interest to determine whether or not tertiary rod-cone interaction pathways (direct rod-to-cone BC and direct cone-torod BC synapses) are functional in mammalian retina. In this study, we examine rod and cone synaptic inputs to various morphologically identified (by Lucifer yellow-dye filling) DBCs in dark-adapted mouse retinas. In addition to studying rod/cone inputs to DBCs by using the response sensitivity and paired light protocols in wild-type mice, we take advantage of several pathway-specific knockout-mouse models to verify the relative contributions of rod and cone as well as the primary, secondary, and tertiary rod/cone pathways to DBC light responses. The mouse models include mice that lack rod response (rod transducin knockout [Trα / ]) (21), mice that lack the connexin36 gap-junction protein (Cx36 / ) (6), and mice that lack DBC R s (Bhlhb4 / ) (22). Results obtained suggest that subpopulations of DBC R s receive direct synaptic inputs from cones and subpopulations of DBC C s receive direct synaptic inputs from rods. They provide clear physiological evidence for functional tertiary rod/cone pathways as well as grounds for revising the rod/cone bipolar-cell dogma in the mammalian retina. Results DBCs with Different Morphology and Rod/Cone Inputs in the Wild- Type Mouse Retina. Fig. 1A shows the morphology (revealed by Lucifer yellow [Fig. 1Aa]), light-evoked currents at various holding potentials (Fig. 1Ab), and cation current (ΔI C recorded at E Cl ) evoked by a pair of light steps (500 nm, 3.5, 0.5 s in duration and 1 s apart [Fig. 1Ac]) of the four types of DBCs in darkadapted wild-type mouse retinal slices. Photocurrents from a rod and an M-cone recorded with suction electrodes in dark-adapted mouse retinal slices are shown in Fig. 1B ([Fig. 1Ba] photocurrents elicited by 500-nm light steps of different intensities and Author contributions: J.-J.P., F.G., and S.M.W. designed research; J.-J.P., F.G., and S.M.W. performed research; J.-J.L., D.E.B., D.L.P., and S.M.W. contributed new reagents/analytic tools; J.-J.P., F.G., D.L.P., and S.M.W. analyzed data; and J.-J.P., J.L., D.E.B., D.L.P., and S.M.W. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1 To whom correspondence should be addressed. swu@bcm.edu. NEUROSCIENCE PNAS January 5, 2010 vol. 107 no

2 Fig. 1. Morphology and light responses of DBCs in the wild-type mouse retina. (Aa) a stacked confocal fluorescent image of a DBC C2, a DBC C1, a DBC R2,anda DBC R1 (columns 1 4, respectively) in mouse retinal slices. INL, inner nuclear layer; IPL, inner plexiform layer (with 10 divisions). (Ab) light-evoked current responses to 500-nm light steps (0.5 s) of different intensities (marked as log-unit attenuation) at various holding potentials. (Ac) light-evoked cation current (ΔI C ) recorded at E Cl to a pair of light steps (500 nm, 3.5, 0.5 s in duration and 1 s apart). (B) a, photocurrents elicited by 500-nm light steps of various intensities recorded with suction electrodes from a dark-adapted rod (upper trace) and a dark-adapted M-cone (lower trace). b, photocurrents from the same rod (upper trace) and M-cone (lower trace) elicited by a pair of light steps (500 nm, 3.5, 0.5 s in duration and 1 s apart). (C) Response-intensity (R-Log I) relations of the rod (n = 8; green) and M-cone photocurrents (n = 2; green) and light-evoked cation currents (ΔI C ) of the DBC C2 s(n = 3), DBC C1 s(n = 6), DBC R2 s (n = 5), and DBC R1 s(n = 3). (Error bars, SDs.) [Fig. 1Bb] photocurrents elicited by a pair of light steps [500 nm, 3.5, 0.5 s in duration and 1s apart]). Rod photocurrent is about 2 log units more sensitive to the 500-nm light step than the M-cone photocurrent and about 4 log units more sensitive than the S-cone photocurrent (23), but we have not recorded from S-cones in this study; the rod exhibits one continuous response to the 500-nm light-step pair because of its slow response recovery time, whereas the M-cone shows two separate responses to the same pair of light steps. The response-intensity (R-Log I) relations of ΔI C of the four DBCs are plotted along with the R-Log I relations of the rod and M-cone (green symbols) in Fig. 1C. The dynamic ranges (defined as the range of light intensity that elicits responses between 5% and 95% of the cell's maximum response) (24) of responses to 500-nm light steps of the rod, M-cone, and S- cone as well as the ΔI C (red) and ΔI Cl (black) of the four types of DBCs are plotted as solid thick lines in Fig. 2. From the polarity of light-evoked cation current (ΔI C recorded at E Cl ), cells 1 4inFig.1A are DBCs (inward ΔI C ). By comparing ΔI C sensitivity (R-Log I dynamic range) and response waveform to light-step pairs of 82 DBCs with the corresponding values of the rod and cone photocurrents, we found that BCs with axon terminals branching within 55 70% of inner plexiform layer (IPL) (such as the DBC in the first column of Fig. 1A) are DBCs with M-cone sensitivity and cone-response waveform; they have two separate responses of similar amplitude to the light-step pair, which is similar to the cone response to the light-step pair in Fig. 1B, and they are named DBC C2 (25). BCs with axon terminals branching within 60 75% of IPL (such as the DBC in the second column of Fig. 1A)are DBCs with mixed rod/m-cone sensitivity and mixed rod/coneresponse waveform; they have two distinguishable responses to the light-step pair with the second response being smaller than the first response, and they are named DBC C1 (25). Two types of BCs with rod bipolar morphology (globular axon terminals in the proximal half of the IPL) (4, 25) have been identified. One type of BC has globular axon terminals endings near 75 90% of IPL depth (such as the DBC in the third column of Fig. 1A) with rod sensitivity but mixed rod and M-cone response waveform (two distinguishable responses to the light-step pair with the second response being smaller than the first response. The other type of BC has globular axon terminals ending at the ganglion cell layer (near 100% of IPL depth, such as the DBC in the fourth column of Fig. 1A) with rod sensitivity and rod-response waveform (one continuous response to the light-step pair similar to the rod response to the light-step pair in Fig. 1B), and they are named DBC R2 and DBC R1, respectively. Moreover, our previous study shows that DBC R1 and DBC R2 have different ΔI Cl (25), and the ΔI Cl dynamic ranges of the four types of DBCs are shown as black bars in Fig. 2. To verify the anatomical features of the four types of DBCs in Fig. 1 with known morphological markers, we immunolabeled 17 retinal slices (each contains a DBC filled with Lucifer yellow after physiological experiments) with antibodies against PKCα and Choline acetyltransferase (ChAT), because anti-pkcα labels Pang et al.

3 Fig. 2. Dynamic ranges of photoreceptors and DBCs to 500-nm light steps in wild-type and mutant mice. Average dynamic ranges (the light-intensity range from threshold [5% of the maximum response] to saturation [95% of the maximum response]) of the rod, M-cone, and S-cone photocurrents and of the ΔI C and ΔI Cl of the DBC C2,DBC C1,DBC R2 and DBC R1 in response to 500-nm light stimuli. Red lines, photocurrents and ΔI C ; black lines, ΔI Cl ; thick solid lines, ΔI C and ΔI Cl in wild-type mice; dotted lines, ΔI C and ΔI Cl in Cx36 / mice; dashed lines, ΔI C and ΔI Cl in Bhlhb4 / mice; thin solid lines, ΔI C and ΔI Cl in Trα / mice. DBC R s in mammalian retina (26) and anti-chat labels cholinergic neurons as well as two distinct bands in IPL (30% and 65% of the mouse IPL depth) (27). By examining the ΔI C polarity, sensitivity, and response waveform to light-step pairs as well as the morphology of the 17 cells (the same way as in Fig. 1), we found that four are DBC C2 s, five are DBC C1 s, five are DBC R2 s, and three are DBC R1 s. Additionally, the physiological and morphological signatures are very consistent within each type. For the four groups (DBC C2 s, DBC C1 s, DBC R2 s, and DBC R1 s), the ΔI C thresholds are near 5.0, 7.9, 7.8, and 8.0, respectively, the axon terminal endings are near 60%, 70%, 85%, and 100% of the IPL depth, respectively, and responses to lightstep pairs exhibit the same double/single-peak patterns as the corresponding DBC types in Fig. 1. Figure 3 shows one example of each of the four DBC groups labeled with Lucifer yellow (red), anti-pkcα (green), and anti-chat (blue). It is evident that the DBC R2 and DBC R1 (Figs. 3 C and D) are PKCα positive, because the soma and especially, the axons and the globular axon terminals appear yellow (red + green). However, DBC C2 and DBC C1 (Figs. 3 C and D) are PKCα negative (soma, axon, and axon terminals are red). Please note that the globular axon terminals look smaller than those in Figs. 1 and 4, because the retinal slices here were fixed. Moreover, the two ChAT-positive bands (blue) in the IPL confirm the axon-terminal ending positions in the IPL of the four DBC types as described in Fig. 1. By comparing the morphology of the DBCs in this study with the mouse BCs described previously (types 1 9 and RBs) (4, 28), we found that DBC C2 s are similar to type 5, DBC C1 s are similar to types 6 and 7, and DBC R2 s and DBC R1 s are similar to RBs. It is evident that DBC C1 s and DBC R2 s are two distinct BC types, because the former is PKCα negative and the latter is PKCα positive; additionally, the axon terminal endings of the former are near to the lower ChAT band, whereas those of the latter are proximal to the lower ChAT band in the IPL. Pathway-Specific Mutant Mice Verify Rod/Cone Input Pathways to Various Types of DBCs. Figure 4A shows a cell with DBC C1 morphology (column 1), a cell with DBC R1 morphology (column 2), and a cell with DBC R2 morphology (column 3) recorded from Cx36 / mouse retinal slices as well as a cell with DBC C1 morphology recorded from a Bhlhb4 / mouse retinal slice (column 4), and a cell with DBC R2 morphology recorded from a Trα / mouse retinal slice (column 5). The response-intensity (R-Log I) relations of the rod photocurrents of the wild-type Cx36 /, Bhlhb4 /, and Trα / mice are shown in the left panel of Fig. 4B, and the ΔI C -Log Fig. 3. Immunolabeling of DBCs with anti-pkcα and anti-chat. Anti-PKCα (green) and anti-chat (blue) labeling of retinal slices containing a Luciferyellow red-filled DBC C2 (A), a DBC C1 (B), a DBC R2 (C), and a DBC R1 (D). The DBCs were identified by their characteristic light responses and morphology as described in Fig. 1. The PKCα-positive cells are DBC R s, and the ChAT-positive cells are cholinergic amacrine cells with two ChAT-positive bands near 30% and 65% of the IPL depth. I relations of the six types of DBCs in Fig. 4A as well as the wildtype DBC R1, DBC R2, and DBC C1 are given in the right panel of Fig. 4B. The dynamic ranges of ΔI C and ΔI Cl from the six DBCs in the mutant mice are given as dotted (Cx36 / ), dashed (Bhlhb4 / ), and thin-solid (Trα / ) lines in Fig. 2. It is evident that rod photocurrent sensitivity and dynamic ranges of the Cx36 / and Bhlhb4 / mice are indistinguishable from the wild-type rods, and the rod photocurrent in the Trα / mouse is absent. The DBC C1,DBC R1, and DBC R2 from Cx36 / mice exhibit the same ΔI C sensitivity to 500-nm lights (R-Log I relations and dynamic ranges) and response waveforms to light-step pairs as their counterparts in the wild-type mouse (Figs. 1 and 4B). The DBC R1 result is consistent with the notion that the DBC R1 s receive inputs only directly from rods. The DBC R2 result suggests that their cone input is not mediated by the connexin36-dependent rod-cone coupling but rather by direct cone-dbc R2 synapses. The DBC C1 from Cx36 / and Bhlhb4 / mice also display the same ΔI C characteristics as their wild-type counterparts, suggesting that the rod input to DBC C1 s is not mediated by the connexin36-dependent rod-cone coupling or the rod DBC R -AIIAC-DBC C1 pathway. The DBC R2 and DBC C1 from the Trα / mouse exhibit ΔI C s about 2 log units less sensitive than the DBC R2 and DBC C1 ΔI C in wild-type mice, but their ΔI C s are robust when elicited by bright 500-nm lights with two response peaks to the light-step pair. These results are consistent with the notion that both DBC R2 s and DBC C1 s receive substantial inputs from both rods and cones and their response sensitivity reduces by about 2 log units when the rod input is absent. NEUROSCIENCE Pang et al. PNAS January 5, 2010 vol. 107 no

4 Fig. 4. Morphology and light responses of DBCs in the mutant mouse retinas. Stacked confocal fluorescent images and light responses of DBC C1 in a Cx36 / retinal slice (column 1), a DBC R1 in a Cx36 / retinal slice (column 2), a DBC R2 in a Cx36 / retinal slice (column 3), a DBC C1 in a Bhlhb4 / retinal slice (column 4), a DBC R2 in a Trα / retinal slice (column 5), and a DBC C1 in a Trα / retinal slice (column 6). (A) Stacked confocal images of Lucifer yellow-filled DBCs in retinal slices. IPL, inner plexiform layer. (B) Current responses to a 500-nm, 0.5-s light step of different intensities (marked as log-unit attenuation) at various holding potentials. (Left) Response-intensity (R-Log I) relations of rod (n = 8) and M-cone (n = 2) photocurrents in dark-adapted wild-type mice and rod photocurrents in the Cx36 / (n =3),Bhlhb4 / (n = 4), and Trα / (n = 3) mice. (Right) Response-intensity (R-Log I) relations of the light-evoked cation currents (ΔI C ) of DBC R1, DBC R2, and DBC C1 in wild-type mice (compare Fig. 1Cb with corresponding R-Log I relations of the mutant DBCs) and of the DBC R1 in Cx36 / mouse (n = 3), the DBC R2 in Cx36 / mouse (n = 4), the DBC C1 in Bhlhb4 / mouse (n = 4), the DBC R2 in Trα / mouse (n = 5), and the DBC C1 in Trα / mouse (n = 3). (C) Light-evoked cation current (ΔI C ) recorded at E Cl to a pair of light steps (500 nm, 0.5-s long and 1 s apart) of different intensities (marked as log-unit attenuation). Discussion We present evidence in this article that a subpopulation of depolarizing rod bipolar cells (DBC R2 s) receives substantial inputs from cones and a subpopulation of depolarizing cone bipolar cells (DBC C1 s) receives substantial inputs from rods in the wild-type mouse retina. These conclusions are derived from the sensitivities and dynamic ranges of the cells cation current responses (ΔI C )to 500-nm light steps (Figs. 1C and 2) compared with the corresponding responses of rods and cones to the same 500-nm lights. DBCs receiving rod inputs (DBC R1 s, DBC R2 s, and DBC C1 s) have response thresholds about 1.2 log units lower than rods (Fig. 2), consistent with the notion that about 20 rods converge signals to a DBC (19). Moreover, ΔI C -response waveforms to light-step pairs also support our conclusion that both DBC R2 s and DBC C1 s receive mixed rod/cone inputs. Both types of cells exhibit two distinguishable responses to the light-step pair with the second response smaller than the first response, which is a mixture of the rod response with one continuous response and the cone response with two responses of equal amplitudesee (Fig. 1B). Despite the similarity in response sensitivity and waveform, DBC R2 s and DBC C1 s are two distinct types of BCs because of three reasons: (i) DBC R2 s have globular axon terminals and DBC C1 s have branching axon terminals; (ii) axon-terminal endings of DBC R2 s are more proximal than DBC C1 axon-terminal endings; and (iii) DBC R2 s are PKCα (rod BC-specific) (26) positive and DBC C1 s are not. DBC R2 s exhibit the same morphology (with globular axon terminals) as the DBC R1 s, except with slightly shorter axons; the globular axon terminals end near 75 90% of IPL depth as opposed to the axon terminals that end near 100% IPL depth for DBC R1 s. In a previous study, we found that DBC R1 s and DBC R2 s exhibit nearly identical ΔI C sensitivity but different ΔI Cl (25). In this study, the light-step pair protocol clearly distinguishes the two cell types by their ΔI C response waveforms. Moreover, we show in Fig. 4 that DBC R2 s in mice without rod responses (Trα / mice) have robust cone-mediated ΔI C, whereas DBC R1 sintrα / mice have virtually no ΔI C. These results again suggest that the two types of DBC R s (morphologically distinguishable by axon-terminal ending levels in IPL) are physiologically distinct; the one has pure rod input and the other has mixed rod/cone input, and they have different ΔI Cl s. In the connexin36 knockout mice, we have shown that DBC R1 s, DBC R2 s, and DBC C1 s exhibit response sensitivities, dynamic ranges, and responses to light-step pairs very similar to their counterparts in the wild-type mouse. These results indicate that the cone inputs to DBC R2 s and rod inputs to DBC C1 s are unlikely to be dominated by the connexin36-mediated rod-cone coupling (secondary pathway). Rod inputs to DBC C1 s are also unlikely to Pang et al.

5 be dominated by the connexin36-dependent rod DBC R -AIIAC- DBC C1 pathway (primary pathway). Therefore, the tertiary pathways (direct cone-dbc R2 pathway and direct rod-dbc C1 pathway) are likely to be the dominating synaptic channels for the rod/cone cross-over in these DBCs. It is important to note that our results from Cx36 / mice do not rule out the possible contributions of the primary pathway to the DBC C1 response and the contributions of the primary and secondary pathways to the DBC R2 and DBC C1 responses. Rather, our results suggest that these connexin36-dependent synapses play a minor role in mediating the cone-to-dbc R2 and rod-to-dbc C1 signals. Similarly, our DBC C1 result in the Bhlhb4 / mouse also supports the idea that the rod-dbc C1 direct (tertiary) pathway contributes much more to the rod-to-dbc C1 signals than the rod DBC R -AIIAC (primary) pathway. It is worth noting that rod-cone coupling is relatively weak (about times weaker than rod-rod and conecone coupling) (29) and that photoreceptor-bc chemical synapses are of high gain (30, 31). This may also contribute to the reason why direct rod-dbc C1 and cone-dbc R2 chemical synapses (when present) contribute more to rod-cone cross-over signals in bipolar cells than rod-cone coupling. In this study, we used three lines of mutant mice, the Trα /, Cx36 /, and Bhlhb4 / mice. One major concern commonly associated with using mutant mice for studying neural circuitry is whether or not the mutation or gene deletion results in changes or reorganization of the neural network other than the changes caused by the specific gene products. In the retina, many gene mutations result in neuronal degeneration, which leads to reorganization of the surviving neurons nearby to accommodate for the degenerative changes, a process called retinal remodeling (or rewiring) (32). However, we believe that the three mutant mouse strains used in this study do not exhibit significant retinal rewiring, at least during the age window that we used (6 16 wk of age), because of three reasons. (i) All three mutant mouse strains have been extensively studied and used by many research groups. There is little detectable neuronal degeneration (6, 21, 22), and thus, degeneration-induced rewiring should not occur in these mice. (ii) The function of the retinal network and individual cells in these mutant mice has been examined, and there is no evidence suggesting that reorganization of any kind occurs in these mice (6, 21, 22, 33). In fact, we have shown in a previous publication that the light-response characteristics of AIIACs in Cx36 / and Bhlhb4 / mice agree almost completely with corresponding response characteristics isolated by pharmacological tools in the wild-type mouse (34). (iii) All mouse strains (wild-type and mutant) used in this study were within the same age window (6 16 wk). This not only gives us age-matched information from all mouse strains, but it also lowers the possibility of age-related retinal remodeling in some of the mutant strains. The input and output synapses of the four types of DBCs described in this article are summarized in Fig. 5. The primary, secondary, and tertiary rod/cone pathways are labeled as red, black, and blue arrows (chemical synapses) and/or zigzags (electrical synapses), respectively. Our results provide direct physiological evidence that challenges the dogma set forward by earlier anatomical studies that mammalian bipolar cells receive segregated rod/cone synaptic inputs (3, 16). In addition to the tertiary rod-hbc C pathway proposed by previous studies (13, 33, 35), our data suggest direct rod-dbc C1 and direct cone-dbc R2 tertiary pathways in the mouse retina. Recent electron microscopic studies reveal chemical synapses made from mouse rods to DBC C s (20), and our data suggest that these synapses are functional and responsible for a substantial share of the DBCs photoreceptor inputs. Although no ultrastructural evidence is available for direct chemical synapses from cones to DBC R s, we clearly show significant cone inputs to DBC R2 s (wild-type mouse results in Fig. 1 and Trα / results in Fig. 4, column 5), and the cone-dbc R input is likely to be direct (Cx36 / data in Fig. 4, column 3). Future electron and confocal microscopic studies on cone-dbc R synapses will provide further support for this observation. Materials and Methods Animals. The wild-type mouse used in this study was C57BL/6J from Jackson Laboratory. Generation of the Cx36 /, Bhlhb4 /, and Trα / mice was described in previous publications (6, 21, 22). All animals were handled in accordance with Baylor College of Medicine's policies on the treatment of laboratory animals. Dissection and preparation of living retinal slices followed essentially the procedures described in previous publications (25, 36). Oxygenated Ames solution (adjusted to ph 7.3) was introduced continuously to the recording chamber by a gravity superfusion system, and the medium was maintained at 34 C by a temperature control unit (TC 324B; Warner Instruments). Light Stimulus. A photostimulator was used to deliver light spots (diameter = 600 1,200 μm) to the retina through the epi-illuminator of the microscope. The intensity of unattenuated (log I = 0) 500-nm light was photons μm 2 sec 1. The number of photoisomerizations per rod per second (Rh*rod -1 sec 1 ) was calculated from a rod cross-section of 0.5 μm 2 (31). The NEUROSCIENCE Fig. 5. Schematic diagram of synaptic connections of the input and output synapses of the four types of DBCs in the mouse retina. R, rod; MC, M-cone; DBC C2, type 2 cone-depolarizing bipolar cell; DBC C1, type 1 cone-depolarizing bipolar cell; DBC R2, type 2 rod-depolarizing bipolar cell; DBC R1, type 1 rod-depolarizing bipolar cell; AC M1, M-cone depolarizing amacrine cell; AII, AII amacrine cell; A17/S1, A17 amacrine cell; ONGC, ON ganglion cell; green, rods and rod BCs; blue, M-cones and M-cone BCs; light orange, GABAergic ACs; dark orange, glycinergic ACs; gray, GCs; arrows,chemical synapses (+,sign-preserving;, signinverting); zigzags, electrical synapses; red, primary rod and cone pathways; black, secondary rod/cone pathways; blue, tertiary rod/cone pathways; PRL, photoreceptor layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer (A, sublamina a; B, sublamina b); GCL, ganglion cell layer. Pang et al. PNAS January 5, 2010 vol. 107 no

6 peak amplitude of light-evoked current responses was plotted against lightstimulus intensity, and data points were fitted by the Hill equation (Eq. 1): R=R max ¼ I N = I N þ σ N ¼ 0:5 1 þ Tanh 1:15N Log I Log σ where R is the current response amplitude, Rmax is the maximum response amplitude, σ is the light intensity that elicits a half-maximal response, N is the Hill coefficient, Tanh is the hyperbolic tangent function, and Log is the logarithmic function of base 10. In this study, we used the R-Log I plot for our analysis, and the light-intensity span (dynamic range [DR] is the range of intensity that elicits responses between 5% and 95% of Rmax) of a cell equals to 2.56/N (24). We define response threshold as the intensity of light that elicits 5% of Rmax. Electrophysiology. Voltage-clamp recordings were made with an Axopatch 200A amplifier connected to a DigiData 1200 interface and pclamp 6.1 software (Axon Instruments). Whole-cell voltage-clamp recordings were made with patch electrodes made with Narishige or Sutter patch electrode pullers that were of 5 7 MΩ tip resistance when filled with an internal solution containing 118 mm Cs methanesulfonate, 12 mm CsCl, 5 mm EGTA, 0.5 mm CaCl 2, 4 mm ATP, 0.3 mm GTP, 10 mm Tris, and 0.8 mm Lucifer yellow; the solution was adjusted to ph 7.2 with CsOH. The chloride equilibrium potential, E Cl, with this internal solution was about 60 mv. Morphology. Cell morphology was visualized in retinal slices through the use of Lucifer-yellow fluorescence with a confocal microscope (Zeiss 510). Images were acquired with a 40 water-immersion objective (numerical aperture = 1.20) using the 458-nm excitation line of an argon laser and a long-pass 505-nm emission filter. Consecutive optical sections were superimposed to form a single image using the Zeiss LSM-PC software, and these compressed image stacks were further processed in Adobe Photoshop 6.0 to improve the signal-to-noise ratio. The level at which dendritic processes stratified in the IPL was characterized in retinal vertical sections by the distance from the processes to the distal margin (0%) of the IPL. Immunocytochemistry. Retinal slices containing Lucifer yellow-filled DBCs were fixed in fresh 4% paraformaldehyde in PBS (ph 7.4) for min at room temperature. They were then incubated in primary antibodies in the presence of 1% donkey serum/pbs with 0.5%Triton X-100/0.1% sodium azide for 5 10 days at 4 C. The mouse anti-pkcα (BD Transduction Lab; dilution = 1:500) and goat anti-chat (Chemicon International; dilution = 1:100) were used. After extensive washing with PBS containing 0.5%Triton X-100/0.1% sodium azide, the tissues were incubated overnight with immunofluorescent secondary antibodies. After extensive rinsing, the tissues were mounted with Vectashield (Vector Laboratories), and immunofluorescence was visualized by the confocal microscope. ACKNOWLEDGMENTS. We thank Roy Jacoby and Cameron Cowan for critically reading this manuscript. This work was supported by grants from the National Institutes of Health (EY04446, EY014127, and EY12008), National Institutes of Health Vision Core (EY02520), the Retina Research Foundation (Houston), Massachusetts Lions, and Research to Prevent Blindness, Inc. (Challenge Grant to Tufts University School of Medicine). 1. Dowling JE (1987) The Retina, an Approachable Part of the Brain (Harvard University Press, Cambridge, MA). 2. Werblin FS, Dowling JE (1969) Organization of the retina of the mudpuppy, Necturus maculosus. II. Intracellular recording. J Neurophysiol 32: Boycott B, Wässle H (1999) Parallel processing in the mammalian retina: The Proctor Lecture. Invest Ophthalmol Vis Sci 40: Ghosh KK, Bujan S, Haverkamp S, Feigenspan A, Wässle H (2004) Types of bipolar cells in the mouse retina. J Comp Neurol 469: Wässle H (2004) Parallel processing in the mammalian retina. Nat Rev Neurosci 5: Deans MR, Volgyi B, Goodenough DA, Bloomfield SA, Paul DL (2002) Connexin36 is essential for transmission of rod-mediated visual signals in the mammalian retina. Neuron 36: Demb JB, Pugh EN (2002) Connexin36 forms synapses essential for night vision. Neuron 36: Kolb H, Famiglietti EV (1974) Rod and cone pathways in the inner plexiform layer of cat retina. Science 186: Bolz J, Wässle H, Thier P (1984) Pharmacological modulation of on and off ganglion cells in the cat retina. Neuroscience 12: Menger N, Pow DV, Wässle H (1998) Glycinergic amacrine cells of the rat retina. J Comp Neurol 401: Chun MH, Han SH, Chung JW, Wässle H (1993) Electron microscopic analysis of the rod pathway of the rat retina. J Comp Neurol 332: Pang JJ, Gao F, Wu SM (2003) Light-evoked excitatory and inhibitory synaptic inputs to ON and OFF alpha ganglion cells in the mouse retina. J Neurosci 23: Soucy E, Wang Y, Nirenberg S, Nathans J, Meister M (1998) A novel signaling pathway from rod photoreceptors to ganglion cells in mammalian retina. Neuron 21: Zhang J, Wu SM (2004) Connexin35/36 gap junction proteins are expressed in photoreceptors of the tiger salamander retina. J Comp Neurol 470: Dang L, et al. (2004) Connexin 36 in photoreceptor cells: Studies on transgenic rod-less and cone-less mouse retinas. Mol Vis 10: Kolb H (1994) The architecture of functional neural circuits in the vertebrate retina. The Proctor Lecture. Invest Ophthalmol Vis Sci 35: DeVries SH,Baylor DA (1995)An alternative pathway for signal flow from rod photoreceptors to ganglion cells in mammalian retina. Proc Natl Acad Sci USA 92: Li W, Keung JW, Massey SC (2004) Direct synaptic connections between rods and OFF cone bipolar cells in the rabbit retina. J Comp Neurol 474: Tsukamoto Y, Morigiwa K, Ueda M, Sterling P (2001) Microcircuits for night vision in mouse retina. J Neurosci 21: Tsukamoto Y, et al. (2007) A novel connection between rods and ON cone bipolar cells revealed by ectopic metabotropic glutamate receptor 7 (mglur7) in mglur6- deficient mouse retinas. J Neurosci 27: Calvert PD, et al. (2000) Phototransduction in transgenic mice after targeted deletion of the rod transducin alpha-subunit. Proc Natl Acad Sci USA 97: Bramblett DE, Pennesi ME, Wu SM, Tsai MJ (2004) The transcription factor Bhlhb4 is required for rod bipolar cell maturation. Neuron 43: Nikonov SS, Kholodenko R, Lem J, Pugh EN Jr (2006) Physiological features of the S- and M-cone photoreceptors of wild-type mice from single-cell recordings. J Gen Physiol 127: Thibos LN, Werblin FS (1978) The response properties of the steady antagonistic surround in the mudpuppy retina. J Physiol 278: Pang JJ, Gao F, Wu SM (2004) Light-evoked current responses in rod bipolar cells, cone depolarizing bipolar cells and AII amacrine cells in dark-adapted mouse retina. J Physiol 558: Haverkamp S, Wässle H (2000) Immunocytochemical analysis of the mouse retina. J Comp Neurol 424: Zhang J, Yang Z, Wu SM (2005) Development of cholinergic amacrine cells is visual activity-dependent in the postnatal mouse retina. J Comp Neurol 484: Wässle H, Puller C, Müller F, Haverkamp S (2009) Cone contacts, mosaics, and territories of bipolar cells in the mouse retina. J Neurosci 29: Attwell D, Wilson M, Wu SM (1984) A quantitative analysis of interactions between photoreceptors in the salamander (Ambystoma) retina. J Physiol 352: Yang XL, Wu SM (1997) Response sensitivity and voltage gain of the rod- and conebipolar cell synapses in dark-adapted tiger salamander retina. J Neurophysiol 78: Field GD, Rieke F (2002) Nonlinear signal transfer from mouse rods to bipolar cells and implications for visual sensitivity. Neuron 34: Marc RE, Jones BW, Watt CB, Strettoi E (2003) Neural remodeling in retinal degeneration. Prog Retin Eye Res 22: Abd-El-Barr MM, et al. (2009) Genetic dissection of rod and cone pathways in the dark-adapted mouse retina. J Neurophysiol 102: Pang JJ, et al. (2007) Relative contributions of rod and cone bipolar cell inputs to AII amacrine cell light responses in the mouse retina. J Physiol 580: Haverkamp S, et al. (2008) Type 4 OFF cone bipolar cells of the mouse retina express calsenilin and contact cones as well as rods. J Comp Neurol 507: Wu SM (1987) Synaptic connections between neurons in living slices of the larval tiger salamander retina. J Neurosci Methods 20: Pang et al.