From retinal circuitry to eye diseases -- in memory of Henk Spekreijse

Abstract

This article summarizes our recent works on stratum-by-stratum structure-function rules for synaptic contacts between retinal bipolar cells and third-order retinal neurons in the inner plexiform layer. These rules were derived from large-scale voltage clamp recordings of various types of bipolar cells in the tiger salamander retina, and they appear applicable to bipolar cells in the mouse and other mammalian species. This review also gives a brief account of how we used pathway-specific knockout mouse models to dissect rod and cone signaling channels in the mammalian retina. Furthermore, studies on cellular and genetic mechanisms underlying several neurodegenerative retinal disorders are described.

My first interaction with Henk Spekreijse was at the 1990 Taniguchi Symposium on Visual Science in Lake Biwa, Japan. The five-day symposium was full of fun, scientific exchange and ample opportunity to know each of the participants. Henk was selected to represent our group to address Mr. Taniguchi in the formal reception, not only because he was a well-known leader in the field, but also, according to our host, that he was “the most handsome man in the group”.

Soon after the Symposium, as the Editor-in-Chief of Vision Research, Henk invited me to be a member of the editorial board, and in 1992, he asked me to be the Neurobiology section editor, a post I held until 2003. During the years in the Vision Research editorial board, I got to know Henk very well. What impressed me most was the breadth of his research areas in addition to his high quality, in depth and quantitative approach to science. He was an ideal person to be in charge of Vision Research, which under his leadership, enjoyed a broad and dynamic coverage of all major aspects of visual science. In 1995, in recognizing that the journal needed more submissions of articles in molecular biology and neuroscience, Henk, with the help of Paul Carton of Elsevier, Joanne Angle of ARVO and myself, started the pre-ARVO symposia in Ft. Lauderdale every year. These symposia and the accompanying Vision Research special issues, have served as important forums on current hot topics in visual neuroscience/disorders at the molecular, cellular and systems levels, and substantially increased the scope and visibility of the journal.

Henk will be remembered as one of the most influential vision researchers in his generation. It is hard to find anyone with his diverse scientific contributions, vast success in establishing Dutch biomedical research institutions and international collaborations, and high-quality leadership in editorial endeavor over two decades (he was one of the two distributing editors of Vision Research since the early 80s and Editor-in-Chief from 1991 to 2004). In his research, Henk enthusiastically strived to bridge basic science findings with ophthalmologic and neurological practice. In this article, I wish to share this view of his by briefly summarizing some of the research in my lab on basic retinal synaptic circuitry and eye disease models.

1. Bipolar cells in the tiger salamander retina exemplify general structure-function rules of the vertebrate retina

The retina is the first neural station of the visual system. It absorbs light, encodes visual information into electrical signals and transmits them to the brain. Most vertebrate retinas have very similar structural and functional organizations, and thus experimental results obtained from different species can be integrated with very few species-specific variations, into a general description of retinal architecture and functional pathways.

Rod and cone photoreceptors respond to light with membrane hyperpolarization, where rods register dim light signals and cones register bright signals. These signals are further processed by a complex but orderly network of electrical and chemical synapses along various bipolar cell (BC) pathways, which carry visual information from the outer retina to the inner retina (Dowling, 1987). BCs may be ON-center with OFF-surrounds (named ON-center BCs or depolarizing BCs (DBCs)), or OFF-center with ON-surrounds (named OFF-center BCs or hyperpolarizing BCs (HBCs)). BCs are further classified according to their rod/cone inputs. In mammals, anatomical evidence reveals that rods and cones make synaptic connections with rod and cone BCs separately, and one type of rod BC and 9-10 types of cone BCs have been identified (Boycott & Wassle, 1999). BCs in amphibian retinas contact both rods and cones, but some are clearly rod-dominated where others are cone-dominated (Yang & Wu, 1997).

BCs in most vertebrates share many common functional and morphological features. Cross-species studies on retinal BCs in recent years lead us to propose that there are 6 major functional types of BCs: the rod (or rod-dominated), cone (or cone-dominated) and mixed (rod/cone) depolarizing and hyperpolarizing bipolar cells (DBC_R, DBC_C, DBC_M, HBC_R, HBC_C and HBC_M). Each exhibits a characteristic set of light response attributes and axon terminal morphology. Based on a large-scale patch clamp study of the salamander BCs (Pang et al., 2004b), several rules for the function-morphology relationships of retinal BCs have been revealed: (1) BCs with axon terminals in the distal half of IPL (sublamina A) are HBCs and those in the proximal half (sublamina B) are DBCs (Nelson et al., 1978). (2) BCs with axon terminals in strata 1, 2 and 10 are rod-dominated, those in strata 4-8 are cone-dominated, and those in strata 3 and 9 exhibit mixed rod/cone dominance. (3) Light-evoked chloride currents (ΔI_C1) in rod-dominated bipolar cells are sustained ON currents whereas those in cone-dominated bipolar cells are transient ON-OFF currents. ΔI_Cl in all BCs are outward, and thus they are synergistic to light-evoked cation currents (ΔI_C) in HBCs and antagonistic to ΔI_C in DBCs. (4) Bipolar cells with axon terminals stratified in multiple strata exhibit combined light response properties of the narrowly monostratified cells in the same strata. (5) Bipolar cells with pyramidally-branching or globular axons exhibit light response properties very similar to those of narrowly monostratified cells whose axon terminals stratified in the same stratum as the axon terminal endings of the pyramidally-branching or globular cells.

2. Functional analysis of retinal circuitry in wildtype and pathway-specific knockout mice

In the mouse retina, although most BCs exhibit diffuse or globular axon terminals, they obey the general function-morphology rules set forth by the salamander BCs. For example, rod DBCs (similar to DBC_Rs in the salamander) have globular axon terminal endings in stratum 10, DBCs with mixed rod/cone inputs (DBC_C1s, similar to DBC_Ms in the salamander) bear axons that terminate in strata 8-9 and axon terminals endings of cone DBCs (DBC_C2s, similar to DBC_Cs in the salamander) are found in strata 6-8 (Pang et al., 2004a).

We also studied the relative rod/cone inputs to DBC synaptic pathways by comparing the light evoked excitatory and inhibitory currents with the thresholds and dynamic ranges of rods and cones to 500nm light. On average, the sensitivity of a DBC_R to 500nm light is about 20 times higher than that of a rod, mainly due to synaptic convergence. DBC_Cs with mixed rod/cone inputs have similar response thresholds as DBC_Rs but wider dynamic range, and DBC_Cs with only cone inputs have a threshold about two log units higher than DBC_Rs (Pang et al., 2004a).

In the mammalian retina, it has been shown that the AII amacrine cells (AIIACs) serve as a communication hub for the BC synaptic pathways: they receive chemical synaptic inputs from DBC_Rs, make electrical synapses with DBC_Cs (DBC_C1s) and inhibitory chemical synapses with HBC_Cs and OFF ganglion cells. We therefore examined synaptic inputs mediating AIIAC light responses (Pang et al., 2007). We found that the light-evoked current response of AIIACs in the wildtype mouse retina is almost completely mediated by DBC inputs, as most of the AIIAC light response as well as the spontaneous postsynaptic currents (sPSCs) are suppressed by 20 μM L-AP4. The residual response in L-AP4 is a small transient current at light offset, which is probably mediated by the HBC-AIIAC synapses in IPL sublamina A (Tsukamoto et al. 2001). Additionally, sIPSCs in AIIACs in the presence L-AP4 may reflect inhibitory inputs from GABAergic/glycinergic amacrine cells (e.g. A17 amacrine cells).

In order to separate the DBC_R and DBC_C1s inputs to AIIACs, we recorded AIIAC light responses in the presence of CNQX, which blocks the DBC_R input but not the DBC_C1 input. In addition, we used two knockout mice, the Bhlhb4 -/- (which lacks DBC_Rs) and the connexin36 (Cx36)-/- mice to verify the pharmacological results. These experiments reveal that the light-evoked current response of AIIACs in the mouse retina is mediated by two parallel DBC inputs: a DNQX-resistant component mediated by cone DBCs through an electrical synapse and a DNQX-sensitive component mediated by rod DBCs. This scheme is supported by the finding that the dynamic range of the AIIAC light response in the Bhlhb4 -/- mouse resembles that of the DNQX-resistant component and the dynamic range of the AIIAC light response in the Cx36 -/- mouse resembles that of the DNQX-sensitive component. By comparing the light responses of the DBC_Rs with the DNQX-sensitive AIIAC component and light responses of the DBC_C1s with the DNQX-resistant AIIAC component, we found that the input-output relations of the DBC_R→AIIAC chemical synapse is nonlinear with a voltage gain of 5 and it saturates for DBC_R signals larger than 5.5 mV. The DBC_C1→AIIAC electrical synapse is approximately linear with a voltage gain of 0.92 (Pang et al., 2007).

The sensitivity of an AIIAC is more than 1000 times higher than that of a rod, suggesting that AIIAC responses are pooled through a coupled network of about 40 AIIACs. Interactions of rod and cone signals in dark-adapted mouse retina appear asymmetrical: rod signals spread into the cone system more efficiently than cone signals into the rod system. The mouse synaptic circuitry allows small rod signals to be highly amplified, and effectively transmitted to the cone system via rod/cone and AIIAC/DBC_C coupling (Pang et al, 2004a).

We also studied BC outputs to retinal ganglion cells in the mouse retina. Three types of alpha ganglion cells (αGCs) were identified. (1) ONαGCs exhibit no spike activity in darkness, increased spikes in light, sustained inward ΔI_C, sustained outward ΔI_Cl of varying amplitude, a large soma (20-25 μm in diameter) with alpha-cell-like dendritic field about 180-350 μm wide stratifying near stratum 7 of the IPL. (2) Transient OFFαGCs (tOFFαGCs) exhibit no spike activity in darkness, transient increased spikes at light offset, small sustained outward ΔI_C in light, a large transient inward ΔI_C at light offset, a sustained outward ΔI_Cl, and a morphology similar to the ONαGCs except for that their dendrites stratified near stratum 3 of the IPL. (3) Sustained OFFαGCs (sOFFαGCs) exhibit maintained spike activity of 5-10 Hz in darkness, sustained decrease of spikes in light, sustained outward ΔI_C, sustained outward ΔI_Cl, and a morphology similar to the tOFFαGCs. By comparing the response thresholds and dynamic ranges of αGCs with those of the BCs and ACs, our data suggest that ONαGCs receive excitatory inputs (ΔI_C) primarily from DBC_C1s, and inhibitory inputs (ΔI_Cl) from ACs with sustained, mixed rod/cone light responses. tOFFαGCs receive excitatory inputs primarily from S-cone dominated HBC_Cs, and inhibitory inputs from ACs with sustained, mixed cone dominated light responses, and sOFFαGCs receive excitatory inputs primarily from M-cone dominated HBC_Cs, and inhibitory inputs from AIIACs through direct feedforward inhibitory synapses (Pang et al., 2003).

We found an abrupt voltage-dependent increase of sIPSCs that occurred at holding potentials above −40 mV in all OFFαGCs, but not in ONαGCs. We interpret this as evidence of electrical coupling between OFFαGCs and presynaptic amacrine cells. The depolarizing current needed to maintain the positive holding potential in the OFFαGC leaks into amacrine cells through gap junctions (Xin & Bloomfield, 1997), that facilitate the release of GABAergic or glycinergic vesicles to the recorded OFFαGC. This is consistent with anatomical evidence on reciprocal electrical synapses between OFFαGCs and GABAergic AC_M2 in mammalian retinas (Jacoby et al., 1996;Bloomfield & Xin, 1997;Dacey & Brace, 1992).

3. Mouse models for studying retinal diseases

During the past several years, our lab has used several lines of mutant and knockout mice to study retinal function and mechanisms of retinal degeneration. These mice can be divided into two broad categories. The first is pathway-specific knockout mice, in which specific molecules or types of cells in the synaptic circuitry are deleted. These mice are useful for studying functional pathways and synaptic connectivity in the retina. Examples are the connexin36-/- and Bhlh4-/- mice, which are used to dissect rod and cone DBC inputs to AIIACs ((Pang et al., 2007), described in the last section). Others such as the connexin57-/- (Cx57 is the gap junction protein in HCs) mouse are used to determine how horizontal cell coupling affects the center-surround receptive fields of various types of retinal bipolar cells and ganglion cells.

The second type of mouse model is associated with retinal degeneration in eye and neurological disorders. We have characterized mechanisms of retinal degeneration in the Spinocerebellar Ataxia type 7 (SCA7) mutant knockin and the Bardet-Biedl Syndrome 4 (BBS4) knockout mice, as two examples of disease-specific mouse models. Additionally, we studied how photoreceptors and bipolar cells dysfunction in mice with deletions of phototransduction proteins or transcription factors. These include the GCAP1/2-/-, Beta2 (NeuroD)-/- and Beta4 (Bhlh4)-/- mice.

SCA7 is an autosomal dominant neurodegenerative disease caused by the expansion of a translated CAG repeat. SCA7 patients suffer from progressive loss of motor coordination, speech impairment, swallow difficulties and cone-rod dystrophy (Zoghbi, 2000). We used an ataxin-7 knockin model in which 266 CAG repeats were introduced into the mouse Sca7 locus (Sca7^266Q/5Q mice). These mice were born with normal eyes, but their eye receded and develop ptosis with age. Our ERG results showed that cone dysfunction preceded rod dysfunction, and M-cones were affected before Scones in the Sca7^266Q/5Q mice, a pattern similar to ERG data obtained from SCA7 patients. Additionally, like infantile SCA7 patients, Sca7^266Q/5Q mice showed progressive shortening of photoreceptor outer segments, possibly due to downregulation of photoreceptor-specific genes such as those for photopigments (Bcp, Gcp and Rho) and outer segment structure proteins (Rom1 and Prph2) (Yoo et al., 2003).

Bardet-Biedl Syndrome (BBS) is an oligogenic syndrome whose clinical manifestations include retinal degeneration, renal abnormalities, obesity and polydactylia. Increasing evidence suggests that the main etiopathophysiology of this syndrome is impaired basal body and ciliary function, two organelles critical for Intraflagellar Transport (IFT). We examined the BBS4 knockout (Bbs4-/-) mice and found that they recapitulated the human disease phenotype as they possess an age-dependent retinal degeneration, as measured by fundus photography and histology (Eichers et al., 2006). Additionally, using ERG, we were able to characterize the retinal degeneration as a cone-rod dystrophy (Eichers et al., 2006). To understand the functions of the BBS4 protein, we examined its roles in regulating IFT in mouse photoreceptors (Eichers et al., 2006). We found that BBS4 is required for a specific transport process from the inner segments to the outer segments of photoreceptors. This process is responsible for the translocation of phototransduction proteins, such as cone opsins, rhodopsin, transducin and arrestin, but not for transport of structural proteins such as rom-1 and peripherin/rds. Loss of this transport leads to cone and rod photoreceptor degeneration. The second possible function of the BBS4 protein is that it is involved in synaptic transmission from the photoreceptor to the second-order retinal neurons, perhaps by interacting with synaptic ribbons in the photoreceptor synaptic terminals. Our combined use of electrophysiology and molecular biology allowed us to gain much insight into the mechanisms of this devastating blinding disease.

Guanylate cyclase activating proteins (GCAPs) are Ca⁺² regulated mediators of photoreceptor response recovery, and several mutants in the GCAP1 genes have been associated with autosomal dominant cone-rod and cone dystrophy in human patients (Newbold et al., 2002). Our immunocytochemical results showed that both GCAP1 and GCAP2 are expressed in mouse rods, and GCAP1 is expressed in mouse M- and S-cones (Howes et al., 2002). Our electrophysiological studies indicate that the recovery kinetics of rod and cone responses are significantly slower in the GCAP1/2-/- mice. Expression of GCAP1 on a null background restore rod and cone response recovery kinetics, suggesting that GCAP1, in the absence of GCAP2, is capable of mediating normal rod and cone response kinetics. Interestingly, preliminary data suggests that GCAP2, in the absence of GCAP1 is also able to rescue the rod recovery kinetics in the GCAP1/2-/- mice. This would suggest that these 2 proteins may play supplemental or redundant roles in the recovery of the rod photoreceptor response. The two-fold slower recovery in GCAP-/-mice compared with the WT mice is much less than the delay in GRK1-/- and RGS9-/-mice, suggesting that the default pathway for cGMP synthesis in the absence of activation enzyme is faster than the default pathways for rhodopsin and transducin inactivation (Pennesi et al., 2003;Pennesi et al., 2006).

The basic helix loop helix (bHLH) transcription factors collectively mediate cellular differentiation in most body tissues including the retina (Murre et al., 1989;Jan & Jan, 1993;Cepko, 1999). We examined the roles of two bHLH transcription factors, Beta2/NeuroD and Bhlhb4, in retinal development and in pathogenesis in the adult retina, by using mice lacking these genes. In Beta2/NeuroD-/- mice the cell loss was most prominent in the outer nuclear layer (ONL), which contains the photoreceptors, normally 10-12 cells thick (Carter-Dawson & LaVail, 1979). In Beta2/NeuroD-/- mice at 2 months of age, the thickness of the ONL was reduced to 5-6 cells, and in 18-month-old null mice the ONL was completely devoid of photoreceptors. In contrast, the ONL of Bhlhb4-/-retinas were normal in thickness and cell count. However, there was a notable difference in thickness of the INL, and our immunohistochemical data showed that the rod bipolar cells are missing, and that the death of these cells occurred during the postnatal development of the retina (Bramblett et al., 2004). Although no visual disorders in humans have been linked to the Beta2/NeuroD or Bhlhb4 loci, heterozygous mutations in BETA2/NeuroD are associated with the development of both type 1 and type 2 diabetes mellitus in humans, but have not been implicated in retinal degeneration (Malecki et al., 1999;Iwata et al., 1999). The loss of Bhlhb4 leads to a ERG phenotype that is similar to that found in humans with congenital stationary night blindness (CSNB) (Dryja, 2000). Bhlhb4 is unlikely as the determinate of X-linked CSNB because it has been mapped to the distal end of human chromosome 20 (Bramblett et al., 2002). However, variations of CSNB, such as the autosomal dominant and recessive forms, exist (Fitzgerald et al., 2001), and Bhlhb4 may be associated with these disorders.

By analyzing retinal light responses in normal, mutant and diseased mouse models, we begin to put together a description of how specific genes, proteins, and synaptic pathways mediate visual function and dysfunction in the mammalian retina. The combined use of physiological, anatomical and genetic methods proves to be a powerful approach, as Henk had suggested, not only for our understanding of the visual system, but also for unraveling the mystery of the entire brain.