Comparisons of structural and functional abnormalities in mouse b-wave mutants

Abstract

In the most simplistic view, the retinal circuit can be divided into vertical excitatory pathways that use glutamate as their neurotransmitter and lateral inhibitory pathways in the outer and inner synaptic layers that modulate excitation via glycine and GABA. Within the vertical excitatory pathways, the visual signal is initiated in the rod, cone or both photoreceptors, depending on the adaptation state of the retina. This signal is transmitted to the rest of the retina through the bipolar cells, which can be subdivided based on: the photoreceptor that provides their input, their dendritic and axonal morphology, and the polarity of their response evoked by a luminance increment, e.g. depolarizing or hyperpolarizing responses. The polarity of this response is controlled by the type of glutamatergic postsynaptic receptor that is expressed on their dendritic terminals. Hyperpolarizing bipolar cells express AMPA/kainate receptors, whereas depolarizing bipolar cells (DBCs) express the metabotropic glutamate receptor 6 (Grm6). The electroretinogram (ERG) is a non-invasive method used to assess overall retinal function. The initiation of the visual signal in the photoreceptors is reflected in the ERG a-wave and the ensuing depolarization of DBCs in the b-wave. When there is failure of signal transmission from photoreceptors to DBCs or signalling within DBCs, the ERG a-wave is present, while the b-wave is absent or significantly reduced. This ERG phenotype has been found in the human population and is referred to as congenital stationary night blindness. Until recently, it had been assumed that the absence of a b-wave was indicative of a lack of signalling through the On pathway, leaving the Off pathway unaffected. Here we review recent findings that demonstrate that many mouse mutants share a no b-wave ERG phenotype but their retinal morphology and RGC responses differ significantly, suggesting very different effects of the underlying mutations on output from the DBCs to the rest of the retinal circuit.

Opening remarks

In the retina the visual signal is initiated when light interacts with opsin in the outer segments of the rod and cone photoreceptors. This signal causes a series of changes within the G-protein-mediated photoreceptor signalling pathway Calvert et al. 2006; Sarfare & Pittler, 2007). These include: a decrease in intracellular cGMP, closure of a cGMP-gated cation channel in the outer segments, an intensity-dependent hyperpolarization in membrane polarization and a concomitant decrease in glutamate release.

Like other CNS neurons, neurotransmitter (i.e. glutamate) release from photoreceptors depends on calcium influx, which in photoreceptors is mediated by L-type voltage-dependent calcium channels (L-VDCCs) (Ball & Gregg, 2002; Barnes & Kelly, 2002). The magnitude of the calcium influx, which depends on membrane polarization and stimulus intensity, creates a graded change in glutamate release (i.e. luminance increments reduce glutamate release) (Trifonov, 1968; Penn & Hagins, 1969; Baylor & Fuortes, 1970; Dowling & Ripps, 1973; Trifonov & Byzov, 1977). Photoreceptors contact two classes of BCs, hyperpolarizing bipolar cells (HBCs) and depolarizing bipolar cells (DBCs), named for the polarity of their response to a decrease in glutamate release (Saito & Kaneko, 1983; Attwell et al. 1987; Nawy & Jahr, 1991; Masu et al. 1995).

The activity of photoreceptors and DBCs has been assessed non-invasively in many species with the electroretinogram (ERG), a gross potential recorded at the corneal surface whose components reflect light-evoked retinal activity. The normal ERG waveform has two primary components, the a- and b-waves. The negative going a-wave represents the hyperpolarization of photoreceptors in response to a luminance increment (Robson & Frishman, 1998; Xu et al. 2003). Under dark-adapted conditions, the response of rod photoreceptors dominates the a-wave, while under light-adapted conditions this component reflects cone photoreceptor responses (note, in the mouse ERG the a-wave at light-adapted levels is absent). The b-wave reflects the depolarizing response of the DBCs and is dominated by rod or cone DBCs depending on the adaptation state of the retina (Robson & Frishman, 1998; Sharma et al. 2005). In addition to its use in human clinical assessments, the ERG also has been instrumental in identifying spontaneous animal mutants and characterizing genetically manipulated mice with defects in outer retinal signalling (Peachey & Ball, 2003). Numerous mutants with a-wave abnormalities have been identified and subsequent identification of the underlying gene defect has been important in defining proteins critical to the phototransduction cascade. Further, the ERG has identified a different abnormal phenotype: a normal a-wave but the absence or reduction of the b-wave. Human patients with this ERG phenotype have a disease referred to as congenital stationary night blindness (CSNB) (Miyake et al. 1986). The first spontaneous mouse mutant described with this functional phenotype was named nob for no b-wave (Pardue et al. 1998). Subsequently, additional no b-wave mutants were identified and the nomenclature extended using the chronology of their discovery (nob, nob2, nob3, nob4). Once the gene responsible for the phenotype was identified, the official nomenclature incorporated both the gene and allele name (phenotype) for each mutant (for example: Nyx^nob (Pardue et al. 1998), Cacna1f^nob2 (Chang et al. 2006), Grm6^nob3 (Maddox et al. 2008) and Grm6^nob4 (Pinto et al. 2007). In this review, we will refer to specific mouse mutants using this official nomenclature. When discussing no b-wave mutants in general terms we will use that terminology to differentiate them from nob mice (Nyx^nob). At times we will use the collective terminology Grm6 mutants.

The no b-wave ERG phenotype indicates that phototransduction is normal, but that synaptic transmission between photoreceptors and DBCs or signalling within DBCs is defective. Mouse (and human) mutants with this ERG phenotype fall into two groups that can be distinguished based on the severity of the b-wave loss: those with a complete loss of the b-wave (cCSNB) and those that retain a residual b-wave and oscillatory potentials, indicative of inner retinal activity (iCSNB) (Miyake et al. 1986). As the mutations were identified, it became evident that the iCSNB phenotype was caused by mutations in genes expressed presynaptically, in the photoreceptor terminals that control glutamate release. In contrast, cCSNB involved mutations in genes expressed postsynaptically, in the DBCs that control membrane depolarization. Below we discuss the results of analyses that have characterized retinal structure and function in several no b-wave mutant mouse models. We discuss how these results have advanced our understanding of: (1) the genes that control neurotransmitter release from photoreceptors and synaptogenesis in the outer plexiform layer (OPL); (2) the postsynaptic mechanisms that control DBC depolarization; and (3) the potential role of DBC output to the inner retinal circuit on the responses of retinal ganglion cells (RGCs).

Presynaptic no b-wave mouse mutants: Ca²⁺ influx through L-type VDCCs and the control of glutamate release from photoreceptors

Glutamate release from photoreceptors relies on Ca²⁺ influx through L-VDCCs. L-VDCCs are heteromultimeric proteins consisting of an α₁ subunit that forms the pore of the Ca²⁺ channel, and auxiliary β, α₂δ and γ subunits that modulate the Ca²⁺ current, regulate channel activation and inactivation and, finally, control the proper assembly of the channel and its localization to the membrane (Catterall, 2000). Immunohistochemical data indicate that rod and cone photoreceptors express L-VDCCs that are comprised of α_1F and α_1D subunits in cones and α_1F only in rods (Morgans et al. 2005). While these two subunits result in channels with similar biophysical and pharmacological properties, their absence results in dramatically different functional outcomes. Knockout and mutant α_1F mice exhibit a significantly reduced ERG b-wave under both light- and dark-adapted conditions, consistent with this subunit's expression and function at both rod and cone photoreceptor terminals (Mansergh et al. 2005; Chang et al. 2006). Knockout α_1D mice have normal ERGs (Wu et al. 2007) and the dark-adapted ERG of the α_1F/α_1D double mutants shows no significant difference from the single α_1F mutant (McCall et al. 2008). Thus, the source of the residual ERG b-wave in α_1F mutant mice may involve other mechanisms, including expression of other L-VDCC α subunits. In addition to mutations in α₁ subunits, loss of either the β₂ or α₂/δ₄ L-VDDC subunit also disrupts the ERG b-wave (Ball et al. 2002; Wycisk et al. 2006a,b). Further, a similar ERG defect indicating disruption in synaptic transmission in the OPL, occurs in mice with mutations in: (1) a component of the ribbon synapse, bassoon (Dick et al. 2003); (2) a molecule thought to modulate the L-VDCC, CaBP4 (Haeseleer et al. 2004); and (3) extracellular or intracellular matrix molecules (dystrophin; Rs1h) (Pillers et al. 1999; Johnson et al. 2006).

Postsynaptic no b-wave mouse mutants: defects in the Grm6 signalling cascade of DBCs

Synaptic transmission also requires the correct detection of the changes in the concentration of neurotransmitter in the synaptic cleft by the postsynaptic cells. In DBCs, glutamate binds to Grm6 (Nakajima et al. 1993) and initiates a G-protein signalling cascade that ends with the closure of a non-specific cation conductance (Slaughter & Miller, 1981). The intermediate molecular components involved in this DBC signalling cascade remain relatively unresolved (see review by Snellman et al. 2008). Six postsynaptic no b-wave mutants have provided or verified the identity of four proteins critical to signal transduction in DBCs. Three disrupt Grm6 itself: Grm6^Tm1Nak (Masu et al. 1995); Grm6^nob4 (Pinto et al. 2007) and Grm6^nob3 (Maddox et al. 2008). In fact, it was the evaluation of the Grm6 knockout mouse (Grm6^Tm1Nak) that verified that Grm6 was the glutamate receptor responsible for DBC signalling. Two mutations disrupt G-proteins Gαo and Gβ5 (Gnao1^tm1Lbi and Gnb5^tmCk) (Dhingra et al. 2000, 2002; Rao et al. 2007) and, again, it was the absence of a b-wave that provided convincing evidence that these G-protein subunits were intermediates in the DBC signalling cascade. The last mutation, Nyx^nob, disrupts the expression of nyctalopin, a protein whose function remains enigmatic (Gregg et al. 2003, 2007). Two lines of evidence place nyctalopin as an integral protein in the DBC signalling cascade. First, Nyx^nob DBCs do not respond to exogenous agonist (glutamate) application (Gregg et al. 2007). Second, expression of an enhanced yellow fluorescent protein (EYFP)–yctalopin fusion protein co-localizes with Grm6 on the DBC dendritic terminals, and more importantly, rescues the ERG b-wave, abnormalities in RGC responses and anatomical defects in their axon terminals in the dorsal Lateral Geniculate Nucleus (dLGN) (Demas et al. 2006; Gregg et al. 2007). While Gαo, Gβ5 and nyctalopin are now established elements within the DBC signalling cascade and cation channel gating, their exact positions within the cascade remain to be determined, as does the identity of the channel itself.

Pre- and postsynaptic no b-wave mutations lead to disparate morphological consequences

Early in postnatal development, the OPL thins due to disruption of its synaptic architecture in all presynaptic no b-ave mutants. In all cases, the synaptic ribbons in the rod photoreceptors are abnormal or absent and the dendrites of horizontal cells and the DBCs extend into the outer nuclear layer, and in some cases ectopic synapses appear to form with photoreceptors (Ball et al. 2002; Dick et al. 2003; Dick & Foroud, 2003; Chang et al. 2006). In contrast, retinal structure is normal at the light microscopic level in four postsynaptic mutants: Grm6^Tm1Nak (Masu et al. 1995; Tagawa et al. 1999); Grm6^nob4 (Pinto et al. 2007); Gnao1^tm1Lbi (Dhingra et al. 2000); and Nyx^nob (Pardue et al. 1998; Ball et al. 2003). In Nyx^nob and Gnao1^tm1Lbi mice, the models studied at the EM level, both pre- and postsynaptic structures in the OPL, are normal (Dhingra et al. 2000; Pardue et al. 2001). There appears to be one exception to this general rule: abnormal OPL morphology has been observed in Gβ5 knockout mice, but this could result from altered expression of the long isoform of Gβ5 in photoreceptors rather than a change in the short isoform in DBCs (Rao et al. 2007).

Altered outer retinal neurotransmission results in disparate RGC phenotypes

The impact of outer retinal defects on retinal output in presynaptic no b-wave mutants has been explored by assessing the spontaneous and visually evoked response properties of their RGCs. Some very surprising results have been reported. In vitro assessments, using multi-electrode arrays, have been reported for Grm6^Tm1Nak, Grm6^nob4, Nyx^nob and Grm6^nob3 (Renteria et al. 2006; Demas et al. 2006; Pinto et al. 2007; Maddox et al. 2008). In vivo, RGC responses have been studied directly using extracellular recordings in the optic nerve for Cacna1f^nob2, Grm6^nob4, Nyx^nob and Grm6^nob3 (Chang et al. 2006; Demas et al. 2006; Pinto et al. 2007; Maddox et al. 2008). RGC output in Grm6^tm1Nak and Cacna1f^tm1Tbh has been studied by evaluating visually evoked responses in the superior colliculus (SC) (Sugihara et al. 1997; Mansergh et al. 2005). A broad generality can be made at this time; the response properties of RGCs in no b-wave mutants with presynaptic defects have only subtle defects, while RGCs in no b-wave mutants with postsynaptic defects are severely altered (Demas et al. 2006; Renteria et al. 2006; Chang et al. 2006; Pinto et al. 2007); Maddox et al. 2008; Heflin, McCall & Gregg, unpublished observations).

RGC responses in presynaptic no b-wave mutants

In vivo RGC response properties have been analysed for two presynaptic mutants, Cacna1f^nob2 (Chang et al. 2006) and Cacn2b^tm1Rgg (McCall, Vessey & Gregg, unpublished observations) mice and their responses are similar to WT. As only preliminary data have been collected for the Cacnb2^tm1Rgg mice, specific descriptions will focus on Cacna1f^nob2 RGCs. At both light- and dark-adapted levels, Cacna1f^nob2 RGCs have receptive fields (RFs) with centre/surround organization and can be classified as either On- or Off-centre with spot stimuli restricted within the RF (Chang et al. 2006). In Cacna1f^nob2 On-centre RGCs, the most prominent defect under light-adapted conditions is a decrease in the spontaneous activity, as well as a compression of the dynamic range of their luminance contrast compared to WT. Cacna1f^nob2 Off-centre RGCs are indistinguishable from WT. Of interest is the reported loss of all visually evoked responses in the SC of the Cacna1f^tm1Tbh knockout mouse, implying the absence of visually evoked input from all RGC axons (Mansergh et al. 2005). Responses have not been evaluated in the SC of Cacna1f^nob2 mice, but a lack of visually evoked activity seems unlikely given their robust RGC responses (Chang et al. 2006). In both mutants, the expression of the Cacna1f protein is eliminated, leaving the reason for the disparate result between the two alleles unclear. One potential explanation is that the two mutant mice are on different genetic backgrounds (SV129 for Cacna1f^tm11Tbh and C57Bl/6J for Cacna1f^nob2). Differences in phenotype resulting from background effects are not without precedence, and have been associated with human CACNA1F mutations (CSNB2: Bech-Hansen et al. 1998; cone dystrophy and Aland Island Eye Disease: Jalkanen et al. 2006, 2007). To directly address this difference, one of the mutants must be moved onto the other background strain by backcrossing for 8–10 generations.

The subtlety of the changes in the responses of Cacna1f^nob2 and Cacnb2^tm1Rgg RGCs suggests that the view of an immutable L-VDCC subunit combination controlling glutamate release from photoreceptors may be too simplistic. Instead, these changes may reflect the presence of other subunits and/or subunit shuffling that might occur in the absence of expression of the primary subunit. Support for this hypothesis comes from the observation that there are differences in the characteristics of Ca²⁺ currents in mouse BCs (Berntson et al. 2003). It is possible that in the absence of the L-VDCC subunits that glutamate is released by a different mechanism and that the changes we observe in RGC responses result from the absence of Cacna1f expression in the inner plexiform layer (IPL) and concomitant alterations in synaptic function (Pan, 2001; Pan et al. 2001; Cui et al. 2003; Ma & Pan, 2003). What is required to begin to address these questions is a strategy in which L-VDCC subunit expression is selectively eliminated in either the OPL or the IPL.

RGC responses in postsynaptic no b-wave mutants

Of the six no b-wave mutants with defects in the DBC signalling cascade, RGC responses have been examined in all three Grm6 mutants and in Nyx^nob. The Gnao1^tm1Lbi mutant survives for approximately 3 weeks after birth, complicating RGC analyses.

One change in the RGC response properties across all of these no b-wave mutants is consistent with their ERG phenotype. All have severely compromised signalling though the On pathway, although RGC assessment techniques and stimuli vary (Demas et al. 2006; Renteria et al. 2006; Pinto et al. 2007; Gregg et al. 2007; Maddox et al. 2008). Further, unlike recordings from WT RGCs, there are many spontaneously active RGCs that are unresponsive to visual stimuli in all of the postsynaptic no b-wave mutants.

Visually responsive RGCs within the Grm6 mutants fall into two groups. In the first, a response is evoked at the onset of a full-field luminance increment. In vivo, these RGCs require bright full-field stimulation and have poorly defined RFs with no centre/surround organization. In vitro, some Grm6 knockout On RGCs are observed with defined RF centres (Renteria et al. 2006). Regardless of recording technique or stimulus configuration, the response to the onset to the luminance increment in these RGCs is significantly delayed, and peak firing rates are significantly lower than WT RGCs or Grm6 mutant Off-centre RGCs (Pinto et al. 2007; Maddox et al. 2008). In the second group of Grm6 mutant RGCs, a response is evoked at the offset of a luminance increment. In vivo, these RGCs have robust responses and RFs with centre/surround organization and therefore can be referred to as Off-centre RGCs. In addition, the onset of a dark spot restricted within their RF centre elicits a robust excitatory response. The latency of the response onset of Grm6^nob4 and Grm6^nob3 Off-centre RGCs is the same as WT RGCs. While Grm6^nob4 and Grm6^nob3 RGCs share these general response characteristics, there are differences between the alleles.

When a full-field stimulus is used to characterize the response of WT RGCs, 51% of WT Off-centre RGCs maintain their RF centre response, responding only at luminance decrement (sign-conserving Off-centre RGCs). The other 49%, previously identified as Off-centre using spots restricted to their RF centre, alter their RF centre response and respond at both luminance increment and decrement (sign-altering Off-centre RGCs; (Sagdullaev & McCall, 2005). In addition, a small proportion of sign-altering Off-centre RGCs (31%) actually lose their Off component.

The proportion of sign-conserving versus sign-altering Grm6^nob4 Off-centre RGCs is similar to WT (48 and 52% sign-conserving and -altering, respectively (Pinto et al. 2007). Among Grm6^nob3 Off-centre RGCs, there are significantly fewer sign-altering cells (88 and 12% sign-conserving and -altering, respectively) compared to both Grm6^nob or WT (Maddox et al. 2008). Whenever an On response is present in sign-altering Grm6 mutant Off-centre RGCs, the onset latency is significantly delayed compared to either their Off response counterpart or to WT RGCs (Renteria et al. 2006; Pinto et al. 2007). These data indicate that there are subtle differences in the inner retina circuitry of Grm6^nob3 and Grm6^nob4 mice. Although the exact mechanism is currently unknown, it could be the consequence of an alteration in input between the On and Off pathways, a mechanism referred to as cross-talk, and for which there is mounting evidence (Zaghloul et al. 2003; Sagdullaev & McCall, 2005; Pang et al. 2007; Molnar & Werblin, 2007). If this were the case, one Grm6 mutation might eliminate signalling through the On pathway, which would both eliminate On RGC responses and reduce On input to the Off pathway, which could produce results similar to Grm6^nob3 (Maddox et al. 2008). Examination of the currents underlying the On responses in these mutants should help to resolve some of these questions.

While the ERG phenotypes of Nyx^nob and Grm6 mutant mice are similar, some of their RGC response properties are very different. Notably, a novel phenotype is seen only in Nyx^nob RGCs. The spontaneous activity of Nyx^nob RGCs has a significant rhythmic bursting component with a fundamental frequency at about 4 Hz (Demas et al. 2006; Gregg et al. 2007). In addition, all Nyx^nob RGCs lack classic centre/surround RF organization (Helfin, McCall & Gregg, unpublished observations). Similar to Grm6 mutants, there is an increase in the percentage of visually unresponsive Nyx^nob RGCs (Demas et al. 2006; Pinto et al. 2007; Gregg et al. 2007; Maddox et al. 2008). In addition, in vivo, all Nyx^nob RGCs require full-field stimulation and when On responses are evoked they have significant onset delays.

One hypothesis to account for the difference between Grm6 mutant and Nyx^nob mutant RGC responses is that nyctalopin is expressed in both the OPL and the IPL (Bayley & Morgans, 2007), whereas Grm6 is expressed only in the OPL. However, this explanation is inconsistent with the restoration of both a normal ERG b-wave and RGCs responses when a tagged nyctalopin fusion protein is expressed on the tips of Nyx^nob DBC dendrites (Demas et al. 2006; Gregg et al. 2007). Another hypothesis is that the state of the Grm6 gated cation channel differs between these mutants and creates the difference in downstream visual processing. In Nyx^nob DBCs, the cation channel is closed or absent (Gregg et al. 2007), and either condition may result in more hyperpolarized DBCs compared to WT. The state of the cation channel in Grm6 mutant DBCs is unknown. However, taken together the state of the channel in Nyx^nob DBCs and the near-normal Off-centre RGC responses in Grm6 mutants predicts that the channel is present and Grm6 mutant DBCs have resting membrane potentials similar to WT. How a difference in the state of the cation channel translates into the vastly different RGC response properties in Grm6 mutant and in Nyx^nob mutant mice is unknown and will require analyses of the DBC cation channels in all of these mutants.

A similar scenario could account for the allelic variance that we observe between Grm6^nob3 and Grm6^nob4 mutant RGCs. One Grm6 mutation could result in the absence of the protein and in the other, its presence in an altered conformation. If this alters the resting membrane potential of the DBCs in the two Grm6 mutants, then one mutation might create tonic glutamate release from DBC axon terminals and the other the absence of release. An absence of release would eliminate cross-talk between the parallel On and Off pathways, while tonic release could create persistent cross-talk and unusual signalling from the On to the Off pathway. This scenario is, however, very speculative. How the differences in the responses across RGCs in Nyx^nob and the various Grm6 mutants arise await an understanding of the states of their cation channels and/or the function of nyctalopin in this pathway.

Concluding remarks

The results from the analyses of no b-wave mutants have advanced our understanding of retinal structure and function in several areas: (1) synaptic development in the OPL; (2) the state of the cation channel in DBCs; and (3) the impact of changes in DBC output on the responses of RGCs.

The presynaptic no b-wave mutants described here (Cacna1f^nob2 and Cacnb2^tm1Rgg) share morphological abnormalities in common with those in several other mutants with disrupted control of glutamate release in the OPL (bassoon, CaBP4, dystrophin, Rsh, Cacna2d4 (Ball et al. 2002; Dick et al. 2003; Haeseleer et al. 2004; Wycisk et al. 2006a,b). Synaptogenesis in the OPL of these mutants is markedly abnormal. There are few if any invaginating synapses formed and the dendrites of the DBCs and horizontal cells extend abnormally and may form ectopic synapses. In light of this morphological disruption it is remarkable that the impact on RGC responses is so subtle. This suggests that the primary circuitry responsible for centre/surround RF organization may reside in the inner retina, a conclusion reached using a knockout mouse lacking gap junctions between horizontal cells (Dedek et al. 2008). Further, the abnormal OPL synaptic organization, but normal RF centre/surround organization, may represent an important consideration in the restoration of function in retinas with retinitis pigmentosa (RP), as it would appear that restoration of normal ordered synaptic connections may not be required.

The differences across the postsynaptic no b-wave mutants described here (Grm6^nob3, Grm6^nob4 and Nyx^nob), indicate that while the ERG is a powerful analytical tool, the impact of mutations that give rise to the same ERG defect may not be representative of the effect on retinal output. This is dramatically highlighted by the difference in RGC RF organization in mice with mutations in Grm6 and Nyx.

Humans with mutations in these various genes have clear deficits, particularly under low light conditions. However, these individuals do have functional vision. Assuming humans have the same variation in RGC output as observed across Grm6^Tm1Nak, Grm6^nob3, Grm6^nob4 and Nyx^nob mouse mutants, this may represent significant plasticity within the visual system and an ability to construct relatively good spatial maps in the absence of anything resembling normal RF centre/surround organization. This result bodes well for therapies aimed at restoring light-mediated input to the retina in diseases such as RP and macular degeneration.