1. Neuroscience
Download icon

Sensitivity and kinetics of signal transmission at the first visual synapse differentially impact visually-guided behavior

  1. Ignacio Sarria
  2. Johan Pahlberg
  3. Yan Cao
  4. Alexander V Kolesnikov
  5. Vladimir J Kefalov
  6. Alapakkam P Sampath
  7. Kirill A Martemyanov  Is a corresponding author
  1. The Scripps Research Institute, United States
  2. University of California, Los Angeles, United States
  3. Washington University in St.Louis, United States
Research Article
  • Cited 3
  • Views 1,083
  • Annotations
Cite as: eLife 2015;4:e06358 doi: 10.7554/eLife.06358

Abstract

In the retina, synaptic transmission between photoreceptors and downstream ON-bipolar neurons (ON-BCs) is mediated by a GPCR pathway, which plays an essential role in vision. However, the mechanisms that control signal transmission at this synapse and its relevance to behavior remain poorly understood. In this study we used a genetic system to titrate the rate of GPCR signaling in ON-BC dendrites by varying the concentration of key RGS proteins and measuring the impact on transmission of signal between photoreceptors and ON-BC neurons using electroretinography and single cell recordings. We found that sensitivity, onset timing, and the maximal amplitude of light-evoked responses in rod- and cone-driven ON-BCs are determined by different RGS concentrations. We further show that changes in RGS concentration differentially impact visually guided-behavior mediated by rod and cone ON pathways. These findings illustrate that neuronal circuit properties can be modulated by adjusting parameters of GPCR-based neurotransmission at individual synapses.

https://doi.org/10.7554/eLife.06358.001

eLife digest

At the back of the eye, a structure called the retina contains several types of cell that convert light into the electrical signals that the brain interprets to produce vision. Cells called rods and cones detect the light, and then signal to other neurons in the retina that relay this information to the brain. Rods and cones are specialized to respond best to different visual features: cones detect color and can track rapid movement; whereas rods are more sensitive to low light levels and so enable night vision.

All rods and cones communicate with particular types of neuron called an ‘ON bipolar cell’: rods send their information to rod-specific ON bipolar cells and cones to cone ON-bipolar cells. To maintain the differences in how visual features are detected, the signals sent by the rod or cone cells need to be tuned separately. Previous studies showed that bipolar cells rely on the action of proteins called RGSs to control how information is passed from rods and cones to ON bipolar cells. However, how the RGS proteins produce their effects is not well understood, and neither is their impact on vision or behavior.

Sarria et al. used a genetic approach to create mice that progressively lost RGS proteins from their retina over the course of several weeks. Recording the nerve impulses produced by the bipolar cells as light shone on the retina revealed that RGS depletion affects these neurons in three ways: how sensitive they are to the signals sent by the rod and cone cells, how quickly they respond to a signal, and the size of the electrical response that they produce.

Sarria et al. then investigated how these changes affected the behavior of the mice. To test the response of the rod cells, the mice performed tasks in dim light. This revealed that it was only when the sensitivity of the bipolar cells decreased that the mice performed worse. However, in a task involving fast-moving objects that investigated the response of cone cells, only changes to the speed of the response affected vision. Therefore, the RGS protein has different effects on the signals from rod cells and cone cells. These findings will be useful for understanding how different light sensitive cells in the retina communicate their signals to extract important visual features, allowing us to both see well at night and track rapid changes in scenery on a bright sunny day.

https://doi.org/10.7554/eLife.06358.002

Introduction

Signaling via G protein coupled receptors (GPCR) mediates neurotransmitter actions and shapes many essential neuronal processes including vision, motor control, and cognition (Wettschureck and Offermanns, 2005). In addition to tremendous diversity in neurotransmitter receptors and the neuronal responses they modulate, GPCR pathways can be tuned in terms of response magnitude, sensitivity, and kinetics, oftentimes creating characteristic signature profiles specifically suited to meet the signaling demands of a neuronal circuit (Marder, 2012; Nusbaum and Blitz, 2012). However, the molecular mechanisms underlying the formation of specific signaling patterns, their relationship to synaptic processing, and ultimately to behavior remain poorly understood.

Signaling in GPCR pathways is initiated as the active receptor stimulates GDP/GTP exchange on G protein α subunits, triggering their dissociation from Gβγ subunits. These dissociated G proteins form active species that regulate the activity of second messenger enzymes, and either directly or indirectly modulate the opening of a number of ion channels (Cabrera-Vera et al., 2003; McCudden et al., 2005). It is now well recognized that the predominant role in controlling GPCR signaling belongs to the Regulator of G protein Signaling (RGS) proteins. RGS proteins stimulate GTP hydrolysis thus promoting subunit re-association and resulting in the termination of G protein signaling on a physiological timescale (Ross and Wilkie, 2000; Hollinger and Hepler, 2002). Studies with knockout mice indicate that the elimination of RGS impacts key parameters of GPCR-driven signaling including sensitivity, temporal characteristics, and response amplitudes (Chen et al., 2000; Heximer et al., 2003; Rahman et al., 2003; Fu et al., 2006; Posokhova et al., 2010). These changes are often paralleled by distinct behavioral changes, for example, motor deficits, alterations in drug sensitivity, memory and reward behavior (Traynor and Neubig, 2005; McCoy and Hepler, 2009; Xie and Martemyanov, 2011).

Interestingly, the effect of RGS protein loss appears to be cell-type specific. For example, the elimination of the dominant RGS in rod photoreceptors profoundly slows deactivation kinetics while minimally affecting response amplitude or sensitivity (Chen et al., 2000; Krispel et al., 2006), whereas in hippocampal pyramidal neurons, moderate changes in response kinetics are accompanied by increased sensitivity (Xie et al., 2010). Altogether, these findings suggest that changes in relative RGS activity may filter the GPCR-generated responses, tuning response parameters to control signaling and ultimately behavior. Although intuitive, this hypothesis remains untested, as the constitutive elimination of RGS proteins typically results in an all-or-nothing impact on GPCR pathways, and compensatory changes during development further confound the interpretation of behavioral changes.

GPCR pathways play an especially critical role in enabling vision at low light levels. Single-photon responses generated by rod photoreceptors are transmitted to the downstream rod ON-bipolar cells (ON-BC). The high gain and rapid response of visual signals puts exquisite demands on both the timing and sensitivity of signal transmission at this synapse to enable rod reliable vision (Okawa and Sampath, 2007; Pahlberg and Sampath, 2011). The transient suppression of the glutamate release from photoexcited rods is sensed by the postsynaptic GPCR, mGluR6 (Snellman et al., 2008; Morgans et al., 2010). In darkness, mGluR6 activates a G protein, Gαo, which maintains the TRPM1 effector channels closed. The opening of TRPM1 channels and resulting generation of the depolarizing response requires deactivation of Gαo (Dhingra et al., 2000; Sampath and Rieke, 2004). Recent studies have indicated two RGS proteins, RGS7 and RGS11 play central role in this process (Morgans et al., 2007; Mojumder et al., 2009; Cao et al., 2012; Shim et al., 2012). In humans, loss of function mutations in mGluR6 or TRPM1 prevent the depolarizing activity of ON-BC and result in congenital stationary night blindness (Dryja et al., 2005; Audo et al., 2009; van Genderen et al., 2009). Similarly, knockout of mGluR6 (Masu et al., 1995), TRPM1 (Morgans et al., 2009; Shen et al., 2009; Koike et al., 2010), Gαo (Dhingra et al., 2000), or RGS7/RGS11(Cao et al., 2012; Shim et al., 2012) in mice also disrupts synaptic transmission between rods and ON-BC, completely abrogating responses of ON-BCs to light flashes. The total loss of signal transmission in these models has largely prevented dissection of the mechanisms that connect the fine-tuning of the GPCR signaling cascade to relevant physiological functions. Furthermore, cone photoreceptors are active under bright illumination and also signal through ON-BC neurons forming a distinct circuit. However, contributions of mGluR6 cascade elements to signal transmission at cone synapses, and their implications for vision, remain poorly understood.

In this study we used an inducible system to examine the effect of progressive RGS loss in mature and differentiated retinas on the ability of ON-BC neurons to process signals generated by rod and cone photoreceptors and its impact on visually-guided behavior. We describe how graded reductions in RGS concentration affect ON-BC light-evoked responses and relate these observations to the performance in visually-guided behavioral tasks. Our findings illustrate how changes in RGS activity may tune GPCR response properties to adapt circuit function to specific behavioral demands.

Results

Conditional knockout mouse model produces a gradual reduction in RGS concentration in mature, differentiated retinas

While elimination of either RGS7 or RGS11 alone does not substantially alter depolarizing activity of ON-BCs (Mojumder et al., 2009; Chen et al., 2010; Zhang et al., 2010), simultaneous loss of both proteins completely abolishes the responses of these cells to light flashes (Cao et al., 2012; Shim et al., 2012). In an effort to achieve an intermediate state of RGS expression we studied mice where elimination of one RGS is combined with the haploinsufficiency of the other (e.g., Rgs11−/−: Rgs7+/− or Rgs11+/−: Rgs7−/−). Evaluation of these mouse models by both electroretinography (ERG) and single cell recordings from rod ON-BCs revealed largely normal responses, that showed only minor delays in the onset time of the ERG b-wave (Figure 1; Table 1). Thus, normal depolarizing activity of ON-BCs is supported by a relatively low concentration of the RGS proteins.

Effects of RGS haploinsufficiency on ON-BC properties.

(A) Representative electroretinography (ERG) traces of RGS11 knockout (Rgs11−/−), RGS7 knockout (Rgs7−/−) alone or in combination with haploinsufficiency of the other (Rgs11−/−, Rgs7+/− and Rgs11+/−, Rgs7−/−). Dashed line shows position of the response peak recorded at the dimmest flash. (B) Quantification of ERG b-wave onset timing across increasing light intensities. Some genotypes showed minor delay in the onset of b-wave at moderate to high light intensities (>1 cd*s/m2). (C) Dependence of ERG b-wave amplitude on flash intensity (>1 cd*s/m2 flashes [t-test, p > 0.5, n = 4–5]). Maximal amplitudes or sensitivities of b-waves were not affected in any of the genotypes. (D) Analysis of rod ON-BC light responses by single cell recordings. Responses of rod ON-BCs to flashes of light that varied in strength by factors of 2, from 0.5–23 R*/rod.

https://doi.org/10.7554/eLife.06358.003
Table 1

ERG b-wave parameters extracted from fitting of scotopic and photopic phases of maximal b-wave amplitudes elicited by varying flash strengths

https://doi.org/10.7554/eLife.06358.004
GenotypeI0.5, scotopic (cd*s/m2)Rmax, (scotopic) (µV)I0.5, photopic (cd*s/m2)I0.5, photopic (µV)
WT0.0009 ± 0.0001587 ± 173.5 ± 1.2457 ± 30
RGS11−/−0.001 ± 0.0001580 ± 444.4 ± 1.4419 ± 32
RGS7−/−0.0008 ± 0.0001615 ± 272.4 ± 1.1510 ± 22
RGS11−/− 7+/−0.0008 ± 0.0001628 ± 512.5 ± 1.4506 ± 25
RGS11+/− 7−/−0.0007 ± 0.0001593 ± 323.1 ± 1.0439 ± 16

To achieve a greater reduction in RGS concentration we developed a model for inducible RGS7 elimination. A conditional Rgs7 allele (Rgs7flx/flx) was introduced in the Rgs11−/− background (cDKO) and the mice were further crossed with a ubiquitous driver line expressing inducible Cre-recombinase (cDKO:Cre+; Figure 2A). In the resulting model, RGS elimination can be induced postnatally by the oral administration of tamoxifen. We first characterized changes in RGS7 expression following tamoxifen administration in cDKO:Cre+ mice. Immunostaining of retinal cross-sections revealed a progressive loss of the RGS7-positive signal from the ON-BC synapses in the outer plexiform layer (Figure 2B). A reduction was evident in both the number of RGS7 puncta and staining intensity reaching minimal levels by day 35. During this timeframe, no changes in the postsynaptic concentration of mGluR6 were detected (Figure 2B). We further quantified the decline in RGS7 protein by quantitative Western blotting. Similar to the immunohistochemical analysis, we detected a progressive decrease in the RGS7-positive band (Figure 2C). Quantitative analysis found immunostaining and Western blotting data to be in good agreement, both showing monophasic decays in the RGS7 levels (Figure 2D). We further analyzed RGS7 concentration upon tamoxifen induction separately for synapses between rods and cones, and their respective ON-BCs (Figure 2B,D; Figure 2—figure supplement 1). Before tamoxifen administration (0 day), RGS7 levels in rod and cone ON-BC synapses were comparable, showing no more than a 5 ± 2% difference in staining intensity between the two. The time course of RGS7 loss induced by tamoxifen was also not different between rod and cone ON-BC synapses: half of RGS7 in rod ON-BC was found at 7.4 ± 1.5 days, vs 7.8 ± 1.3 days in cone ON-BCs. At day 35 following tamoxifen administration, RGS7 expression was reduced to 18 ± 3% in rod ON-BCs and to 17 ± 3% in cone ON-BC synapses. Importantly, the cytoarchitecture at the ON-BC dendritic tips of cDKO mice remained unaltered despite the RGS7 loss. Postsynaptic accumulations of both TRPM1 and mGluR6 and the alignment of pre- and post- synaptic compartments were both preserved (Figure 2E; Figure 2—figure supplement 2). Prior to tamoxifen administration both cDKO:Cre- and cDKO:Cre+ mice were indistinguishable from the previously characterized Rgs11−/− background strain in both RGS7 expression and localization (Figure 3A). Furthermore, no changes in RGS7 concentration or its postsynaptic accumulation were detected upon tamoxifen administration in cDKO littermates lacking Cre recombinase expression (Figure 3A). Consistent with these observations, ERG responses recorded in cDKO:Cre- and cDKO:Cre+ mice were indistinguishable from their Rgs11−/− controls and tamoxifen administration failed to alter ERG responses in cDKO:Cre- animals (Figure 3B). Thus, the observed effects are associated specifically with genomic editing of the Rgs7 locus.

Figure 2 with 2 supplements see all
Progressive RGS7 loss in the ON-BC neurons of a conditional mouse knockout model (cDKO).

(A) Schematic representation of a breeding strategy for the conditional inactivation of RGS7 on Rgs11−/− background. To induce recombination and RGS7 loss mice were administered tamoxifen for 5 days followed by testing. (B) Immunohistochemical analysis of RGS7 expression and localization in the outer plexiform layer of mouse retinas following tamoxifen administration. Retina sections were stained with RGS7 (green) and mGluR6 (red) antibodies from cDKO:Cre+ mice at 12, 17, 22, 28, and 35 days after the start of tamoxifen administration. Retinas lacking RGS7 (Rgs7−/−) were used as a control. Two independent experiments yielding similar data were conducted. Note progressive loss of RGS7 immunoreactivity from postsynaptic puncta denoted by mGluR6. (C) Analysis of changes in RGS7 expression in total retina lysates by Western blotting. cDKO:Cre+ mice were treated with tamoxifen and retinas were collected at indicated time points. To generate retina lysates containing various RGS7 protein content, varying amounts of RGS11 knockout lysates were spiked into lysates isolated from RGS7 and RGS11 double knockout retinas (DKO). To determine the level of RGS7 reduction, a calibration curve was plotted from densities of varying amounts of RGS7 protein and used to determine the protein content in retina extracts obtained from tamoxifen-treated mice. Retinas from three separate animals were used in these experiments. (D) Quantification of RGS7 protein content at different time points following tamoxifen treatment. Values representing density of RGS7 band (red) and fluorescence intensity in mGluR6-positive puncta determined separately for rods (green) and cones (blue) were normalized to respective values observed before tamoxifen treatment (day 0). Averaged RGS7 percentages from western and IHC experiments are fitted with a single exponential decay function. Error bars are SEM values. (E) Intact synaptic morphology and retina cytoarchitecture at 28 days post-tamoxifen treatment. Retina cross-section from Rgs11−/− and cDKO:Cre+ mice were co-stained with postsynaptic marker mGluR6 and pre-synaptic marker CtBP2 and their direct apposition was noted. Immunostaining with TRPM1 (green) reveals intact morphology of ON-BC, their dendritic branching and distribution of postsynaptic puncta. Figure 2—figure supplement 1: assignment of rod vs cone synapses by immunohistochemistry. Figure 2—figure supplement 2: intact synaptic morphology and retina cytoarchitecture at 28 days post-tamoxifen treatment.

https://doi.org/10.7554/eLife.06358.005
Expression of Cre recombinase without tamoxifen induction and tamoxifen administration without Cre expression do not alter RGS7 expression, localization and light response of mice.

(A) Immunohistochemical analysis of RGS7 (green) and PKCα (red) expression and localization in the outer plexiform layer of Rgs11−/−, cDKO:Cre-, and cDKO:Cre+ mouse retinas. (B) Representative ERG traces elicited from scotopic (left) and photopic (right) flashes in mice described above show unaltered retina electrophysiological responses.

https://doi.org/10.7554/eLife.06358.008

To obtain a more quantitative picture, we determined the absolute expression levels of RGS7 and RGS11 referencing them to the levels of mGluR6. Using recombinant protein standards and quantitative Western blotting, we found that RGS7, RGS11, and mGluR6 are present at 20 ± 2, 23 ± 1, 31 ± 1 fmol per 10 mg of total protein, respectively (Figure 4). Based on immunohistochemical analysis, RGS11 (Cao et al., 2009), RGS7 (Cao et al., 2012; Figure 4—figure supplement 1) and mGluR6 (Masu et al., 1995) are found exclusively, in the outer plexiform layer of the retina, indicating that the values that we obtained reflect protein content specifically at the ON-BC synapses. Considering that mGluR6 receptors form obligate dimers that function as a single unit, there is about threefold molar excess of RGS proteins (∼43 fmol) relative to mGluR6 dimer (∼15 fmol) in ON-BCs. In summary, these data illustrate that RGS7 loss beyond its stoichiometry with mGluR6 can be induced postnatally in a graded manner without affecting synaptic morphology or retinal cytoarchitechture.

Figure 4 with 1 supplement see all
Quantification of RGS7, RGS11 and mGluR6 proteins in the retina.

(A) Representative Western blots of retina lysates analyzed alongside with recombinant standards. (B) Quantification of protein content. A calibration curve was plotted from densities of recombinant protein standards (closed circles) and used to determine the protein content (open circles) in retina extracts obtained from four separate mice. (C) Comparison of absolute RGS7, RGS11 and mGluR6 protein levels in the retinas. The experiment was performed 3 times. Error bars indicate the SEM. Figure 2—figure supplement 1: distribution of RGS7 immunofluorescence in the retina.

https://doi.org/10.7554/eLife.06358.009

Progressive loss of RGS7 changes the kinetics and sensitivity of light-evoked responses in rod ON-BCs

To determine how progressive loss of RGS affects signal transmission in the rod pathway, we examined retinal responses to light in dark-adapted live anesthetized mice by ERG. In cDKO:Cre+ mice prior to tamoxifen administration, the delivery of very dim (scotopic) flashes, that activate rods but not cones, elicited a robust b-wave, whose generation requires the activity of rod ON-BCs (Figure 5A). These responses were indistinguishable from those observed in Rgs11−/− strain (Figure 3B). Administration of tamoxifen to cDKO:Cre+ mice dramatically changed these responses, progressively reducing their amplitude and slowing their time course (Figure 5A; Table 2). We next analyzed changes in response properties across a range of flash strengths. As flash strengths increase, the contribution of cones to the overall response becomes appreciable. However, rod- and cone- driven b-waves saturate at distinct light intensities leading to a characteristic biphasic shape of the intensity-response relationship (Figure 5B). Accordingly, the first phase of the curve is commonly considered to consist of relatively pure rod-driven component that can be extracted mathematically (Figure 5C; Equation 1). Analysis of this scotopic ERG b-wave showed marked time-dependent changes in both maximal response amplitude and sensitivity (Table 2). At 35 days following tamoxifen administration, the flash responses were barely distinguishable from the baseline seen upon constitutive elimination of both RGS7 and RGS11, with an approximately10-fold reduction in b-wave maximal amplitude and nearly a 100-fold decrease in its sensitivity. The responses further showed a time-dependent slowing in their onset, as evidenced by a progressive increase in b-waves time-to-peak and visible deceleration of the b-wave slopes (Figure 5D). Importantly, we observed no tamoxifen-induced changes in ERG b-wave characteristics in littermates lacking Cre recombinase (Figure 3B), indicating that these changes are caused specifically by the progressive loss in RGS7 expression.

The effect of progressive RGS loss on rod ON-BC responses as revealed by scotopic ERG.

(A) Representative ERG responses elicited with dim flashes (0.001 cd*s/m2) from cDKO:Cre+ mice at 0, 12, 17, 22, 28, and 35 days post-tamoxifen administration. (B) Maximal ERG b-wave amplitudes plotted against their eliciting flash intensities. The ERG b-waves exhibit biphasic saturation pattern, with first one around 0.05 cd*s/m2, corresponding to activity of rod ON-BC, and a second one around 100 cd*s/m2, corresponding to activity of cone ON-BC. (C) Rod component of the ERG response extracted from fitting the first phase of the response in panel B. (D) Analysis of the changes in the onset of ERG b-wave elicited by half-saturating flashes. Time to peak values are plotted as a function of days after tamoxifen treatment. Errors bars are SEM.

https://doi.org/10.7554/eLife.06358.011
Table 2

Effect of tamoxifen treatment on scotopic ERG b-wave parameters

https://doi.org/10.7554/eLife.06358.012
Days post tamoxifenI0.5, scotopic (cd*s/m2)Rmax, scotopic (µV)Time to peak (ms) at I0.5
00.0007 ± 0.0001560 ± 10100 ± 6
120.001 ± 0.0002390 ± 10130 ± 5
170.005 ± 0.0001190 ± 8180 ± 10
220.008 ± 0.0003160 ± 5260 ± 21
280.03 ± 0.00170 ± 3280 ± 25
350.06 ± 0.00650 ± 4260 ± 22
DKON.A0N.A
  1. Values are mean + SEM, n = 5 mice for all timepoints.

To determine how the en masse activity of ON-BCs seen in ERGs is set by the responses of single cells, we recorded the light-evoked activity of rod ON-BCs in dark-adapted retinal slices (Figure 6). Whole-cell patch clamp recordings revealed some variability in ON-BCs responses suggesting that efficiency of tamoxifen-induced recombination varied on a cell-by-cell basis (Table 3). For example, we noticed that some cells were very susceptible to tamoxifen treatment and completely lost responsiveness to light, perhaps due to stochastic hyperactivation of Cre-recombinase leading to rapid elimination of RGS proteins. Because the proportion of unresponsive cells remained relatively stable (∼31–46%) across different days following tamoxifen administration, and because elimination of RGS proteins result in complete abrogation of ON-BC responses to flashes (Cao et al., 2012), these cells were excluded from the analysis. Analysis of the remaining light-sensitive cells revealed a progressive increase in time-to-peak and concomitant decrease in maximal amplitudes following tamoxifen administration (Figure 6A; Table 3). We next compared the average half-maximal flash strengths across all rod ON-BCs recorded on days 17, 22, 28, and 35 and found their progressive increase, reflecting a reduction in sensitivity. No changes in noise variance were detected, suggesting that reduction in RGS concentration also effectively decreases the signal-to-noise ratio (Figure 6B; Table 3). Interestingly, the sensitivity reduction determined from patch clamp recordings matched well that obtained by ERG, which summates the responses of all cells across the retina (Figure 6C). In summary, we observed that gradual titration of RGS levels affects kinetics, sensitivity, and amplitude of rod ON-BC light responses.

Single cell recordings from rod ON-BC recapitulate key changes in average light-evoked responses measured by ERG.

(A) Responses of rod ON-BCs to flashes of increasing strength. Shown are representative response families for 10 ms flashes (arrow) measured at an increasing number of days following tamoxifen administration. Flash strengths yielded 0.70, 1.4, 2.9, 5.7, 11, and 23 R*/rod for control (Rgs7+/+:Rgs11−/−), 0.58, 1.0, 2.2, 4.3, 8.6 and 17 R*/rod for day 17; 1.0, 2.2, 4.3, 8.7, 17, and 35 R*/rod for day 22, and 16, 32, 65, 130, 260, 520, and 1000 R*/rod for day 35. Note with time following tamoxifen administration the progressive slowing and reduction in maximum amplitude of the light-evoked response (see Table 3). (B) Average normalized response-intensity relationships for all cells recorded for control (day 0) as well as days 17, 22, 28, and 35 following tamoxifen administration. Data were sampled at 1 kHz and filtered at 50 Hz. (C) Comparison of half-saturation responses from ON-BC measured by single-cell recordings (top) (n = 4–11) and ERG (bottom) (n = 5) from cDKO:Cre+ mice at different days after tamoxifen administration. Data points plotted are half-maximal values from each tamoxifen time point fitted by sigmoidal dose-response (variable slope) equation with the least squares method. The time required for tamoxifen to reduce the sensitivity of ON-BC by half is not different when obtained by ERG (28 ± 1 days) vs single cell patch-clamp (27 ± 0.3 days). Errors bars are SEM.

https://doi.org/10.7554/eLife.06358.013
Table 3

Parameters of rod ON-BC light responses by single cell recordings

https://doi.org/10.7554/eLife.06358.014
Days post tamoxifenI0.5, (R*/rod)*Rmax (pA)Noise Variance (pA)2Unresponsive (µV)
01.6 ± 0.10 (4)340 ± 331.2 ± 0.20/4
124.8 ± 1.6 (9)200 ± 522.1 ± 0.44/13
226.4 ± 0.10 (8)200 ± 311.6 ± 0.25/13
2869 ± 31 (6)130 ± 451.3 ± 0.15/11
35160 ± 34 (13)68 ± 142.1 ± 0.411/24
  1. *

    Mean + SEM, number of cells is indicated in parenthesis.

  2. Unresponsive cells were those where a flash did not evoke a response.

Decrease in RGS expression correlates with deterioration of the cone-to-cone ON-BC pathway function

Because ON-BCs making synapses with cone photoreceptors utilize a similar mGluR6-TRPM1 signaling pathway for generating depolarizing responses, we examined how changes in RGS concentration affected the light-evoked responses in cone ON-BCs (Figure 7, Table 4). Given low abundance of cones in murine retinas and the functional heterogeneity of cone ON-BC that they connect to, we relied exclusively on ERG for these studies. Prior to tamoxifen administration, bright (photopic) flashes produced a characteristic two-component ERG waveform in cDKO mice. In contrast, after tamoxifen administration we observed a progressive reduction in the b-wave amplitudes and kinetics without detectable changes in the a-wave (Figure 7A). We performed dose-response studies analyzing the dependence of photopic b-wave amplitude on flash intensity across all treatment groups. As in the case of rods, the component of the response that reflects cone contribution (Figure 7B) was derived mathematically from the biphasic response profile (Figure 5C). The resulting traces revealed changes in cone-generated responses (Figure 7B) with a progressive decrease in maximal response amplitude and sensitivity. To verify that recorded responses reflected specifically changes in cone ON-BC signaling, we performed ERGs when rod function was suppressed by exposure to background light (Figure 7C; Figure 7—figure supplement 1). Light-evoked responses collected on this rod-suppressing background showed a progressive decline in b-wave amplitude and kinetics, just as for dark-adapted conditions. Quantification of the parameters derived from the analysis of intensity-response relationships indicated that maximal amplitude, sensitivity, and onset timing of the response similarly declined with time after tamoxifen administration (Figure 7D,E). Together, these results indicate that down-regulation of RGS proteins in the cone synapses lead to changes in cone ON-BC light-evoked responses, that are similar to changes observed in rod synapses.

Figure 7 with 1 supplement see all
The effect of progressive RGS loss on cone ON-BC responses as revealed by photopic ERG.

(A) Representative dark-adapted ERG responses elicited with photopic (100 cd*s/m2) flashes from cDKO:Cre+ mice at 0, 12, 17, 22, 28, and 35 days post-tamoxifen administration. (B) Cone-generated component of the ERG response extracted from fitting the second phase of the response in Figure 5B. (C) Representative light-adapted ERG responses elicited with photopic flashes (10 cd*s/m2, left) and (100 cd*s/m2, right). Steady background light of 50 cd/m2 was applied to saturate rods. (D) ON-BC dose-response plot of maximal ERG b-wave amplitudes recorded with background light and plotted against their eliciting flash intensities. (E) Analysis of the changes in the ERG b-wave onset timing elicited by half-saturating flash intensities on a light background. Errors bars are SEM. Figure 2—figure supplement 1: isolation of the ERG b-wave peak.

https://doi.org/10.7554/eLife.06358.015
Table 4

Effect of tamoxifen treatment on light-adapted photopic ERG b-wave parameters

https://doi.org/10.7554/eLife.06358.017
Days post tamoxifenI0.5, photopic (cd*s/m2)Rmax, photopicTime to peak (ms) at I0.5
05 ± 1204 ± 3135 ± 5
1210 ± 2111 ± 2242 ± 4
1723 ± 370 ± 763 ± 6
2265 ± 465 ± 669 ± 10
28190 ± 2955 ± 473 ± 5
35260 ± 4453 ± 374 ± 6
DKON.A0N.A
  1. Values are mean + SEM, n = 5 mice for all timepoints.

Loss of RGS differentially affects visually-guided behavior under scotopic and photopic conditions

To assess how changing responses in rod ON-BCs affects dim light vision we developed a behavioral water maze assay. The assay measures the ability of mice to locate an escape platform under varying luminance levels. Under photopic conditions, trained wild-type mice reached the visible platform in ∼10 s (Figure 8A). Video rate analysis of tracks indicated that under these conditions mice swim directly to the visible platform taking the shortest path (Figure 8A). In contrast, when the platform was hidden mice randomly swam and appeared to encounter the platform by chance, increasing the time to escape by nearly fivefold. Thus, mouse vision in this test can be evaluated by measuring the time to escape on the platform.

Evaluation of scotopic vision using water maze behavioral test.

(A) Development and validation of visually-guided behavioral task for the assessment of mouse vision. Mice were trained to find a randomly placed visible escape platform in a water maze. Their tracks were recorded and used to determine times to escape from water. Under bright photopic light environment, mice with constitutive elimination of RGS7 and RGS11 (DKO) find visible (7 ± 1 s), but not hidden (45 ± 4 s) platform as readily as wild-type mice (7 ± 1 s). Error bars are SEM (*p < 0.05, t-test, n = 4–5). Escape latencies thus served as a measure of their visual abilities. The room luminance was then decreased in a step-wise fashion followed by the re-assessment of mouse performance. DKO animals performed significantly worse than wild-type (WT) mice in the task at luminance levels ≤0.1 cd/m2 (*p < 0.05, t-test, n = 5 for both groups). (B) Behavioral performance of the conditional knockout mice (cDKO) mice in the visual task upon induction of RGS7 loss by tamoxifen administration. Mice were assessed under pure scotopic conditions equating to luminance of 0.005 cd/m2. Performances of cDKO:Cre+ and cDKO:Cre- littermates were compared across days after tamoxifen treatment. At days 28 and 35 cDKO:Cre+ mice performed comparably to DKO mice, taking 28 ± 4 and 42 ± 7 s (vs 36 ± 4 s for DKO) to find the platform (p > 0.05, One-Way ANOVA, n = 5 for both groups). In comparison, performance of cDKO:Cre+ mice lagged that of cDKO:Cre- littermates at day 28 (28 ± 4 vs 10 ± 2 s) and day 35 (42 ± 7 vs 14 ± 3 s) (*p < 0.05, t-test for each time point, n = 5 for each group). (C) Elimination of RGS7 or RGS11 along with RGS7 haploinsufficiency does not affect mouse performance in the visual discrimination water maze task. Behavioral performance of WT, Rgs7−/−, and Rgs7+/−:Rgs11−/− mice was assessed in the water maze based behavioral task across various luminance levels covering the photopic and scotopic ranges. Performance was compared among all genotypes at each specific luminance level.

https://doi.org/10.7554/eLife.06358.018

To validate the test, we evaluated the performance of the constitutive RGS7/11 double knockouts (DKO), which completely lack an ERG b-wave and thus are considered night blind. Under photopic conditions, the escape latency of DKO mice with the visible platform was no different from that of wild-type animals, indicating that they see the platform normally (Figure 8A). However, when the maze luminance was decreased to the scotopic range, the performance of DKO mice dropped sharply and their escape latencies became statistically equivalent to random encounters with the hidden platform, indicating a complete loss of scotopic vision (Figure 8A). Thus, RGS expression in rod ON-BCs is absolutely required for high sensitivity scotopic vision.

We next tested the performance of the cDKO mice under scotopic conditions. Before tamoxifen administration escape times of cDKO:Cre+ mice were substantially shorter than in DKO and no different from these of cDKO:Cre- littermates or WT mice, indicating that their scotopic vision remained intact (Figure 8B). This performance did not decline up to 22 days after the beginning of tamoxifen treatment. However, a marked increase in escape time was observed in cDKO:Cre+ mice at 28 days, and by 35 days they displayed similar difficulty in visually locating the platform as the constitutive DKO animals, indicating that they developed night blindness (Figure 8B). Administration of tamoxifen to RGS7 haploinsufficient or Rgs7−/− mice did not alter the performance compared to wild-type animals at any light levels (Figure 8C).

To understand better the relevance of RGS function in ON-BCs to behavior, we also assessed visual function of mice in a virtual environment using the optokinetic reflex (OKR). The optomotor response test is based on the ability of mice to respond reflexively to computer-generated rotating sine-wave gratings, which form a virtual cylinder around them (Prusky et al., 2004). Previous studies have established an essential contribution of ON-pathway to visual contrast sensitivity under both scotopic and photopic conditions reflecting the critical role of both rod and cone pathways to this behavior (Schiller et al., 1986; Iwakabe et al., 1997; Kolesnikov et al., 2011). To assess scotopic contrast sensitivity change in cDKO mice we used the optimal mouse scotopic temporal frequency of 0.75 Hz in combination with a moderate stimuli speed of 7.5 deg/s (Umino et al., 2008) at which the contrast sensitivity is close to maximal (Figure 9A). Before tamoxifen administration, both genotypes were indistinguishable in their ability to visualize rotating gratings at their contrast threshold. Tamoxifen treatment severely diminished the performance of cDKO:Cre+ mice without affecting their cDKO:Cre- littermates (Figure 9A). Interestingly, the deterioration of scotopic vision was somewhat resistant to tamoxifen treatment until day 22. These observations are in a good agreement with the data from the water maze task (Figure 9B). Thus, the loss of RGS affects mouse scotopic vision in a stereotyped manner that does not depend on the paradigm chosen to evaluate behavioral performance.

Evaluation of scotopic and photopic mouse vision by optomotor task.

(A) Scotopic (∼3 × 10−4 cd/m2) contrast sensitivity deficits in mice lacking RGS proteins (cDKO:Cre+) vs cDKO:Cre- mice as a function of time of post-tamoxifen administration (*p < 0.05, t-test, n = 4). (B) Comparison of visual performance of mice in water maze task and optomotor response tests. Mouse vision declined by half after 25 ± 1 days of tamoxifen administration in both experiments. (C) The reliance of high speed contrast sensitivity on intact ON-BC function. Deficits in photopic (70 cd/m2) contrast sensitivity of DKO mice are apparent at stimuli speeds of >12 deg/s (error bars are SEM; t-test: **p < 0.01, ***p < 0.001, n = 4–5). (D) Visual acuity is mainly unaffected by the loss of RGS11 together with RGS7 (DKO) as compared to Rgs11−/− controls. (error bars are SEM. t-test: *p < 0.05, n = 4–5 mice). (E) Photopic (70 cd/m2) contrast sensitivity deficits in mice lacking RGS proteins (cDKO:Cre+) vs cDKO:Cre- controls as a function of time of post-tamoxifen administration (error bars are SEM; t-test: ***p < 0.001, n = 7). (F) Comparison of progressive RGS elimination effects on scotopic vs photopic contrast sensitivities of mice. The half-time of contrast sensitivity decline is 25 ± 1 days for scotopic vs 9.0 ± 2 days for photopic conditions, respectively. The data for each group are normalized to their respective values upon beginning of the tamoxifen treatment (day 0) and expressed as percentage.

https://doi.org/10.7554/eLife.06358.019

Finally, we used the OKR approach to evaluate the effect of gradual RGS loss on photopic vision. The parallel processing of cone-generated signals by different classes of BCs makes signaling through ON-BC not entirely necessary for photopic vision (e.g., Figure 8). Therefore, we first determined the contribution of cone ON-BC to photopic (70 cd/m2) OKR performance using DKO mice lacking ON-BC responses and compared them to control Rgs11−/− littermates that show normal ON-BC function. We found that Rgs11−/− and DKO mice had virtually identical contrast sensitivity for slow stimuli (up to 12 deg/s), whereas at higher stimuli speeds (Sp) the DKOs displayed impaired discrimination of low-contrast gratings pattern compared to their Rgs11−/− counterparts (Figure 9C). In contrast, photopic visual acuity of DKO mice remained unaffected across most Sp and showed only a minor decline at the fastest speed of 50 deg/s (Figure 9D).

Having established the contribution of the cone ON-BC pathway to overall contrast discrimination, we evaluated the impact of declining RGS concentration on photopic vision. Tamoxifen administration in cDKO:Cre+ mice caused a 2.4-fold reduction in photopic contrast sensitivity observed at maximal stimuli speed of 50 deg/s in as early as 12 days (Figure 9E). The average photopic contrast sensitivity reached its lowest level by day 22, with a nearly a fivefold decline compared with either untreated littermates or a group of cDKO:Cre- controls that also received tamoxifen. This is in stark contrast to scotopic conditions (∼3 × 10−4 cd/m2) where contrast sensitivity did not start deteriorating until after day 22 following tamoxifen administration, and by day 35 it had reached a sevenfold decline as compared to cDKO:Cre- control mice (Figure 9F).

Sensitivity and kinetics of rod and cone ON-BC responses and behavioral performance of mice correlate with different RGS levels

The availability of quantitative information regarding RGS levels at any time point in the analysis allowed us to determine how signal transmission to ON-BCs influences visually-guided behavior with the RGS concentration as a common denominator (Figures 10, 11). Thus, we compared how both ON-BC ERG response parameters (i.e., b-wave onset, amplitude, and sensitivity) and behavioral sensitivity varied as a function of the RGS7 concentration. These results reveal that the effect of RGS7 concentration on b-wave onset kinetics occurred with a half-saturating concentration of 37 ± 2% of available protein (Figure 10A). Similarly, a halving of the response amplitude in rod ERG b-wave occurred at RGS7 levels of 39 ± 3% of available protein (Figure 10B), and a twofold decline of scotopic response sensitivity occurred at RGS7 levels of 26 ± 2% (Figure 10C). A similarly sharp dependence on RGS7 levels was observed for the scotopic vision of mice (Figure 10D). Animals showed a precipitous drop in their visual contrast sensitivity once the levels of RGS7 declined below ∼35%, with 26 ± 2% reflecting the half-saturating level. To compare directly changes in rod ON-BC response characteristics to behavioral performance, we calculated RGS protein concentration that produced half-saturating responses across these different measures (Figure 10E). The data reveals that rod-driven ERG response amplitudes, onset and sensitivity are disproportionately affected by changes in RGS concentration. Although measured response parameters are interrelated by their underlying biochemical mechanism, they correlate differentially with the mouse behavioral performance. For example, a threefold reduction in response amplitude and a twofold deceleration of ERG b-wave onset did not appear to have an appreciable effect on scotopic visual performance. Instead, mouse scotopic behavior tightly correlated with the rod ON-BC response sensitivity, albeit compounded by pronounced changes in other response parameters.

Correlation of RGS levels with key parameters of rod-rod ON-BC synaptic transmission and behavioral performance under scotopic conditions.

Dose response relationship between half-maximal values of rod ERG b-wave onset timing (A), maximal response amplitude (B), half-saturating light intensity (I0.5) (C) and performance in the behavioral task (D) were plotted as a function of RGS7 concentration quantitatively determined at each time point following tamoxifen administration. In panels AD data points are fitted with sigmoidal dose-response (variable slope) equation with the least squares method. (E) Comparison of changes from different response parameters. RGS7% was converted to RGS7 protein levels using quantitative values determined in Figure 4. Scotopic behavioral performance is resistant to RGS reduction and tracks most closely with rod ON-BC sensitivity. One-Way ANOVA with Bonferroni's post hoc test reveals that half-maximal values for sensitivity and behavior do not significantly differ from each other (p > 0.05, n = 5), but differ from those for response kinetics and amplitude (p < 0.05, n = 5).

https://doi.org/10.7554/eLife.06358.020
Correlation of RGS levels with key parameters of cone-cone ON-BC synaptic transmission and behavioral performance under photopic conditions.

Dose response relationship between half-maximal values of cone ERG b-wave onset timing (A), maximal response amplitude (B), half-saturating light intensity (I0.5) (C) and performance in the behavioral task (D) were plotted as a function of RGS7 concentration quantitatively determined at each time point following tamoxifen administration. In panels AD data points are fitted with sigmoidal dose-response (variable slope) equation with the least squares method. (E) Comparison of changes from different response parameters. RGS7% was converted to RGS7 protein levels using quantitative values determined in Figure 4. Photopic behavioral performance is highly affected by RGS reduction and tracks closely with response amplitude and kinetics. One-Way ANOVA with Bonferroni's post hoc test reveals that half-maximal value for sensitivity differs from all the other parameters (p < 0.05, n = 5).

https://doi.org/10.7554/eLife.06358.021

A similar relationship was derived for the cone-driven photopic ERG responses and behavioral sensitivity (Figure 11). We found that a halving of the photopic ERG b-wave amplitude occurred at an RGS7 concentration of 43 ± 2% of available protein, a twofold effect on onset kinetics occurred at an RGS7 concentration of 38 ± 2% of available protein, and a twofold decline in sensitivity occurred at an RGS7 concentration of 25 ± 2% of available protein. Thus the photopic ERG b-wave amplitude was the response parameter most sensitive to the reduction in RGS7. However, in contrast to rod-mediated vision, we found that behavioral performance relying on the cone ON pathway was very sensitive to even small decline in RGS7, losing half of their photopic visual performance with 43 ± 2% of the RGS remaining (Figure 11D). To compare changes in cone ON-BC response characteristics to behavioral performance we plotted the absolute RGS protein concentration that produced half-saturating responses (Figure 11E). We found that photopic vision is more sensitive to changes in the amplitude and kinetics of the cone ON-BC light-evoked responses rather than their light sensitivity.

Discussion

One of the biggest challenges in neuroscience is to understand how the regulation of cellular activity influences the properties of signals in neuronal circuits to control behavioral responses. These signals may vary in their temporal characteristics, sensitivity thresholds, and throughput bandwidth to ultimately match physiological demands (Marder, 2012). GPCR signaling pathways play a key role in setting many circuit parameters as they mediate effects of a vast number of neurotransmitters and neuromodulators (Nusbaum and Blitz, 2012). However, our understanding of the molecular mechanisms that shape signaling and their relationship to behavioral outcomes is rather limited.

Differential modulation of signal transmission in rod and cone ON-BCs

In this study, we used the well-defined rod and cone ON circuits of the retina to examine how changing key parameters of GPCR-mediated responses affected visually-guided behavior. Through genetic manipulations we observed graded alterations in the maximal response amplitudes, timing of response onset, and sensitivity of rod and cone ON-BCs, while testing the effect of these changes on mouse visual performance. We found stark differences between rod- and cone-driven ON circuits with respect to the influence of signal transfer parameters on visually-guided behavior. In the rod circuitry, changes in the timing or amplitude of signal transmission between photoreceptors and ON-BC neurons did not affect behavioral performance in visually-guided tasks that required intact dim vision. However, a change in the sensitivity of transmission at this synapse, albeit aggravated by kinetic deficits, was not tolerated behaviorally and loss of the response sensitivity strictly correlated with the decrease in behavioral performance of the animals. In contrast, cone-mediated vision was sensitive to minor changes in kinetic parameters of signal transmission, causing a pronounced decrease in behavioral performance even when light sensitivity was unaffected. The rod circuit is biased for high sensitivity to detect single photon absorptions close to absolute visual threshold (Pahlberg and Sampath, 2011), whereas cone pathways operate at higher light levels and require the capacity to extract temporal features from light stimuli (Rodieck, 1998). Thus, our observations suggest that differential specialization of rod and cone pathways involves adaptation at the first synapse in these respective circuits.

Here we find that the parameters of the GPCR-mediated response can be dissociated to provide a hierarchical influence on function. In ON-BCs, decreasing concentrations of two RGS proteins expressed in dendritic tips, RGS7 and RGS11, sequentially influence the onset timing, amplitude, and sensitivity of mGluR6-mediated responses (Figure 12). In this pathway, response kinetics and amplitude appeared to be the parameters most sensitive to changes in RGS concentration, whereas sensitivity required a much larger reduction in RGS concentration to produce an impact. There is a window in the RGS concentration range where profound effects on kinetics were not accompanied by appreciable changes in response sensitivity. These observations suggest a mechanism by which GPCRs can generate unique patterns of activity and differentially tune circuit properties based on varying levels of RGS proteins. Thus, regulatory nodes where information processing is mediated by GPCRs may be tuned to extract temporal features at one end of the RGS concentration range, whereas others may serve as sensitivity filters at the other end of the range. This possibility is supported by observations in photoreceptors where manipulation of RGS9 concentration affects temporal characteristics of the response without changes in sensitivity (Chen et al., 2000; Krispel et al., 2006) and in hippocampal pyramidal neurons where knockout of RGS proteins changes both response kinetics and sensitivity (Xie et al., 2010). Interestingly, levels of multiple RGS proteins have also been shown to fluctuate in response to extracellular cues (Hurst and Hooks, 2009; Traynor et al., 2009) suggesting that changes in RGS concentration may dynamically control GPCR response properties.

Model for RGS function in setting ON-BC responses to light.

In the dark, mGluR6 activated by high synaptic glutamate concentration produces ample amounts of GTP-bound Go, which closes TRPM1 channels. RGS proteins (RGS7 and RGS11) deactivate Go converting it back to inactive GDP-bound state but the equilibrium is dominated by excess of Go-GTP ensuring no channel activity. Upon light exposure, the activity of mGluR6 is reduced, and reduction in Go-GTP catalyzed by RGS proteins becomes dominant, allowing channels to open. When concentration of RGS proteins decline below a threshold point the speed of Go deactivation becomes rate-limiting to reduce the speed and magnitude of TRPM1 channel opening as well as its sensitivity to the reduction in glutamate concentration.

https://doi.org/10.7554/eLife.06358.022

Saturation of RGS concentration in ON-BCs and reproducibility of signal transfer

Under native conditions, RGS proteins in ON-BCs appear to be present in substantial excess over the minimal amount needed for sustaining wild-type response properties. Rod ON-BCs tolerate an almost 75% reduction in RGS levels before any measurable effects on response properties could be observed. To understand the basis for this overexpression we compared levels of RGS7 and RGS11 to that of the mGluR6 receptors. We found that RGS proteins are present at approximately threefold stoichiometric excess over mGluR6 dimers. Several recent studies suggest that the signal transduction components of the ON-BC are organized in a macromolecular complex (Morgans et al., 2010; Dhingra and Vardi, 2012). Indeed, both RGS7 and RGS11 form physical complexes with mGluR6 (Cao et al., 2009) and a recently discovered orphan receptor GPR179 (Orlandi et al., 2012; Ray et al., 2014). These interactions are required for proper postsynaptic targeting of RGS7/11 (Cao et al., 2009; Orlandi et al., 2012). The expression of RGS proteins also depends on mGluR6, and its elimination in mice significantly decreases RGS7 and RGS11 levels (Cao et al., 2009). Additionally, mGluR6 and its effector channel TRPM1 are further integrated by a scaffolding protein, nyctalopin (Cao et al., 2011; Pearring et al., 2011). The relatively rapid (∼50–100 ms) response of ON-BCs to flashes of light is unlikely to be supported by the stochastic action of RGS proteins if they were deactivating Gαo in the vicinity of TRPM1 simply by diffusion. Thus, these considerations suggest a model where functionally relevant RGS molecules must be physically integrated into the macromolecular signaling complex. Since RGS proteins are present in the molar excess over the mGluR6 complex, their depletion does not alter the signaling until the level reaches the stoichiometric threshold and RGS is reduced within the macromolecular assembly. Therefore, it seems reasonable to speculate that overexpression of RGS proteins over the stoichiometry with the mGluR6 signaling complex would minimize fluctuation in the kinetics and sensitivity of ON-BC responses, ensuring their reproducibility. In support of this hypothesis we find that titration of RGS proteins below a certain threshold point (∼1:1 stoichiometry with mGluR6 complex) detrimentally affects the signal-to-noise ratio of ON-BC responses and increases their variability.

These results additionally suggest that G protein deactivation normally is not a rate-limiting step in the cascade of events that lead to generation of a depolarizing response in ON-BCs. Previous studies in rod photoreceptors that utilize similar G protein cascade for light reception demonstrated that G protein deactivation limits the recovery of a photoresponse. In these neurons, increase in the concentration of RGS protein (RGS9) resulted in acceleration of the response termination kinetics (Krispel et al., 2006; Burns and Pugh, 2009). Since a reduction in mGluR6 activity underlies the light-evoked responses in ON-BCs (compared to an increase in rhodopsin's activity for phototransduction), G protein deactivation drives the activation phase of the response and RGS proteins influence its slope. By titrating the RGS concentration in the ON-BCs we found a critical threshold point, below which the onset of the ON-BC depolarization is sensitive to changes in G protein deactivation. The critical threshold point roughly corresponds to ∼25% of normal RGS levels (contributed by both RGS7 and RGS11) in wild-type synapses. Notably, a substantial excess in RGS concentration (threefold to fourfold), and hence G protein deactivation rates above this threshold, has no influence on ON-BC response properties. Thus, by setting the levels of RGS expression above this threshold, mGluR6 signaling in ON-BCs ensures that G protein deactivation is not a rate-limiting step in generation of a depolarizing response. Instead, this rate-limiting reaction could be associated with either TRPM1 channel opening or intrinsic deactivation of the mGluR6 receptor. Determining the identity of this rate limiting process that drives the onset of ON-BC depolarizing response will be an important area for future investigation.

Differential tuning of rod and cone circuits at the first visual synapse

Interestingly, we found that RGS proteins have similar impact on response properties of ON-BCs that form synapses with either rod or cone photoreceptors. Given monosynaptic connectivity between rods and rod ON-BCs, and the progressive night blindness associated with RGS loss, the implications of these changes for rod-mediated behavior are rather unequivocal. However, signaling of cones via parallel ON- and OFF- circuits, together with the heterogeneous nature of cone ON-BCs, complicates the interpretation of the dependence of photopic (cone) vision on cone-to-cone ON-BC signal transmission. The ability to discriminate changes in luminance across a broad range of stimulation frequencies is characteristic feature of photopic vision in many species, including mice (Umino et al., 2008). Previous studies have indicated that pharmacological or genetic blockade of the ON-pathway affects contrast sensitivity in mice and monkeys under photopic conditions (Schiller et al., 1986; Iwakabe et al., 1997). In line with these observations, we find that changes in amplitude and kinetics of the photopic ERG b-wave caused by the loss of RGS expression result in pronounced deficits in the temporal properties of contrast detection for cone vision. Ablation of RGS proteins severely impacted photopic contrast sensitivity for high-speed visual stimuli while leaving visual acuity relatively unaffected. This highlights an essential role that temporal regulation of mGluR6-TRPM1 signaling at the cone-to-cone ON-BC synapse plays for normal photopic vision. Although the details on how RGS proteins may contribute to the responses of individual cone ON-BCs remain to be established, it is possible that differential expression of RGS7 vs RGS11 at cone ON-BC synapses (Mojumder et al., 2009) may contribute to setting their unique profiles, thus increasing the temporal resolution of cone vision.

Materials and methods

Mouse strains, and tamoxifen gavage

Generation of mice with the constitutive deletion in Rgs7 (Rgs7−/−) (Cao et al., 2012), Rgs11 (Rgs11−/−) (Cao et al., 2008), or both (DKO) (Cao et al., 2012) was previously described. Conditional targeting of RGS7 was achieved by flanking exon 4 with LoxP sites (Rgs7flx/flx) (Cao et al., 2012). The resulting mice were crossed with Rgs11−/− line to generate cDKO:Cre- strain (Rgs11−/−: Rgs7flx/flx), and further with a Cre driver line ubiquitously expressing tamoxifen-inducible Cre-ERT2 recombinase B6.Cg-Tg(CAG-cre/Esr1*)5Amc/J to produce cDKO:Cre+ mice (Rgs11−/−: Rgs7flx/flx: CAG-CreERT2). To induce Cre expression 20 mg/kg tamoxifen (Sigma, St. Louis, MO), dissolved in corn oil and 10% ethanol, was administered by oral gavage with an 18–24 gauge smooth tip needle. Tamoxifen was administered once daily for 5 consecutive days. All procedures were carried out in accordance with the National Institute of Health guidelines and were granted formal approval by the Institutional Animal Care and Use Committees of the Scripps Research Institute (IACUC protocol number 14-001), Washington University (IACUC protocol number 20140236), and the University of Southern California (IACUC protocol number 10890).

Antibodies, recombinant proteins and Western blotting

The generation of sheep anti-TRPM1 is described (Cao et al., 2011). Rabbit anti-RGS7 (7RC1), was a generous gifts from William Simonds (NINDDK/NIH) and the guinea pig anti-mGluR6 antibody was a gift from Dr Takahisa Furukawa (Osaka University). Mouse anti-PKCα (ab11723; Abcam, Cambridge, MA), rabbit anti-RGS7 antibodies (07-237; Upstate Biotechnology, Billerica, MA), mouse anti-CtBP2 (612044; BD Biosciences) and rabbit anti-cone arrestin (ab15282; Millipore, Billerica, MA) were purchased.

Recombinant His-tagged RGS7 and RGS11 were co-expressed in Sf9 insect cells together with Gβ5, via baculovirus-mediated delivery, then purified by Ni-NTA chromatography as described previously (Martemyanov et al., 2005). Glutathione S-transferase (GST)-tagged mouse mGluR6 C terminus protein (aa840-871) was expressed in Escherichia coli and affinity purified on GSTrap HP column (GE, Fairfield, CT). Protein concentration was determined by bicinchoninic acid (BCA) Protein Assay Kit (Pierce, Carlsbad, CA) and adjusted to reflect the protein purity determined by densitometry of Coomassie blue-stained gels.

Whole retinas were removed from mice and lysed by sonication in ice-cold PBS supplemented with 150 mM NaCl, 1% Triton X-100, and Complete protease inhibitor tablets (Roche, Basel, Switzerland). Lysates were cleared by centrifugation at 20,800×g for 15 min at 4°C. Total protein concentration in the supernatant was measured by using BCA Protein Assay Kit (Pierce, Carlsbad, CA). Supernatants were added with SDS sample buffer (pH 6.8) containing 8 M urea and were subjected to 12.5% SDS/PAGE. Protein bands were transferred onto PVDF membranes, subjected to Western blot analysis first with primary antibodies against RGS7, RGS11 or mGluR6 and then with HRP-conjugated secondary antibodies, and detected by using ECL West Pico system (Pierce, Carlsbad, CA). For the quantitative detection of RGS7 we used rabbit anti-RGS7 antibodies (07-237; Upstate Biotechnology, Billerica, MA). Signals were captured on film and scanned by densitometer, and band intensities were determined by using NIH ImageJ software.

Immunohistochemistry

Dissected eyecups were fixed for 15 min in 4% paraformaldehyde, cryoprotected with 30% sucrose in PBS for 2 hr at room temperature, and embedded in optimal cutting temperature medium. 12-micrometer frozen sections were obtained and blocked in PT1 (PBS with 0.1% Triton X-100 and 10% donkey serum) for 1 hr, then incubated with primary antibody in PT2 (PBS with 0.1% Triton X-100 and 2% donkey serum) for at least 1 hr. After four washes with PBS with 0.1% Triton, sections were incubated with fluorophore-conjugated secondary antibodies in PT2 for 1 hr. After four washes, sections were mounted in Fluoromount (Sigma, St. Louis, MO). Images were taken with a Leica SP800 confocal microscope. Quantitative analysis of immunofluorescence from confocal images was performed using Leica software. At each time point (day: 0, 12, 17, 22, 28, 35) immunohistochemical staining of cDKO:Cre+ was repeated twice and at least two sections obtained from two individual animals were used for analysis and averaging. Sections were double stained for RGS7 and marker protein mGluR6, which co-localizes with RGS7. Positive staining of mGluR6 was used as a reference for the normalization of the recorded fluorescence intensity in the channel containing RGS7 protein. Rod and cone synapses were differentiated by: (i) counter staining with PNA and/or b-arrestin that specifically label cone terminals, (ii) sublamina position in the outer plexiform layer, (iii) specific clustering pattern of synapses at cone terminals. The fluorescence intensity within synaptic puncta was analyzed using line-scan mode and a constant puncta-encircling area, which tightly surrounded the contours of each puncta. Mean intensity (measured in pixels) were averaged across ∼20 individual punctas per imaged section, taking three sections per retina, and two to three retinas used per time point. Imaging parameters were the same for all sections and retinas. Values for groups were normalized using pre-tamoxifen RGS7 values as a 100% percentage of protein remaining. RGS7 protein percentages were plotted vs the tamoxifen time course and fitted nicely with a simple decay function in GraphPad Prism 6.

ERG

Electroretinograms were recorded by using the UTAS system and a BigShot Ganzfeld (LKC Technologies, Gaithersburg, MD). Mice (4–8 wk old) were dark-adapted (≥6 hr) or light adapted (50 cd/m2, 5 min) and prepared for recordings by using dim red light. Mice were anesthetized with an i.p. injection of ketamine and xylazine mixture containing 100 and 10 mg/kg, respectively. Recordings were obtained from the right eye only, and the pupil was dilated with 2.5% phenylephrine hydrochloride (Bausch & Lomb, Bridgewater, NJ), followed by the application of 0.5% methylcellulose. Recordings were performed with a gold loop electrode supplemented with contact lenses to keep the eyes immersed in solution. The reference electrode was a stainless steel needle electrode placed subcutaneously in the neck area. The mouse body temperature was maintained at 37°C by using a heating pad controlled by ATC 1000 temperature controller (World Precision Instruments, Sarasota, FL). ERG signals were sampled at 1 kHz and recorded with 0.3 Hz low-frequency and 300 Hz high-frequency cut-offs.

Full field white flashes were produced by a set of LEDs (duration < 5 ms) for flash strengths ≤2.5 cd·s/m2 or by a Xenon light source for flashes > 2.5 cd·s/m2 (flash duration < 5 ms). ERG responses were elicited by a series of flashes ranging from 2.5 × 10−4 to 2.5 × 101 cd·s/m2 in increments of 10-fold. Ten trials were averaged for responses evoked by flashes up to 2.5 × 10−1 cd·s/m2, and three trials were averaged for responses evoked by 2.5 × 100 cd·s/m2 flashes. Single flash responses were recorded for brighter stimuli. To allow for recovery, interval times between single flashes were as follows: 5 s for 2.5 × 10−4 to 2.5 × 10−1 cd·s/m2 flashes, 30 s for 2.5 × 100 cd·s/m2 flashes, and 180 s for 2.5 × 101 cd·s/m2 flashes.

ERG traces were analyzed using the EM LKC Technologies software, Sigma Plot and Microsoft Excel. The b-wave amplitude was calculated from the bottom of the a-wave response to the peak of the b-wave. To isolate the b-wave peak from sharp varying oscillatory potentials, a local regression smoothing algorithm using a bi-square weighting function was used in SigmaPlot (see Figure 7—figure supplement 1). Traces obtained from DKO, that completely lack the ON-BC triggered b-wave were subtracted from ERG traces recorded in cDKO:Cre+ mice following tamoxifen administration, each at corresponding light intensities. This was done to remove ambiquity of assigning diminishing b-wave due to interference with P1 component. The data points from the b-wave stimulus–response curves were fitted by Equation 1 using the least-square fitting method in GraphPad Prism6.

(1) R=Rmax,r×I/(I+I0.5,r)+Rmax,c×I/(I+I0.5,c).

The first term of this equation describes rod-mediated responses (r), and the second term accounts primarily for responses that were cone mediated (usually at flash intensities ≥1 cd·s/m2 for dark-adapted mice; index c). Rmax,r and Rmax,c are maximal response amplitudes, and I0.5,r and I0.5,c are the half-maximal flash intensities. Stimulus responses of retina cells increase in proportion to stimulus strength and then saturate, this is appropriately described by the hyperbolic curves of this function. The maximal amplitudes of rod and cone-driven b-wave stimulus–response curves in Tables 1, 2 are plotted against RGS7 protein concentration and fitted with a simple dose–response curve (variable slope) in GraphPad Prism6. Time to peak of the b-wave was measured as the time from the peak of a-wave to the peak of the b-wave.

Single cell recordings

Recordings of light-evoked currents in rod ON-BCs were performed as described previously (Okawa et al., 2010; Cao et al., 2012). Briefly, mice were dark-adapted overnight and euthanized according to protocols approved by the Institutional Animal Care and Use Committee of the University of Southern California (Protocol 10890). Retinas were isolated, embedded in a low-gelling temperature agar, and 200 µm slices were made on a vibrating microtome (Leica VT-1000S). Retinal slices were placed on the stage of an upright microscope and superfused with heated Ames Media (A-1420; 35-37°C) equilibrated with 5% CO2/95% O2. Whole-cell voltage clamp recordings (Vm = −60 mV) were made while delivering flashes of light that varied in strength from evoking a just-measurable response to those generating a maximal response. Noise variance was computed from the 0.2 s preceding the flash. Rod ON-BCs were identified in retinal slices based on the distinctive shape of the cell body and its position in the inner nuclear layer jammed up against the outer plexiform layer. The internal solution for whole-cell recordings consisted of (in mM): 125 K-Aspartate, 10 KCl, 10 HEPES, 5 NMG-HEDTA, 0.5 CaCl2, 1 ATP-Mg, 0.2 GTP-Mg; pH was adjusted to 7.2 with NMG-OH. Membrane currents were filtered at 300 Hz by an 8-pole Bessel filter, and digitized at 1 KHz.

Evaluation of scotopic vision by behavioral water maze task

Mouse visual behavior was assessed using a water maze task with a visible escape platform. The method for assessing visual function in this experiment is principled on a Morris water maze (Morris, 1984) and previous reports describing evaluation of mouse vision by a swimming-based task (Prusky et al., 2000). Mice are natural swimmers and this task exploits their innate inclination to escape from water to a solid substrate. This task uses an ability of a mouse to see a visible platform with a timed escape from water, as an index of its visual ability. Before testing, mice readily learned to swim to the visible escape-platform and performance usually plateaued at around 10 s within 15 trials for all treated groups. Mice which did not learn the task, for example, performance did not improve or plateau for at least the last three or more consecutive trials or had any visible motor deficits were discarded from the experiment. Visual-guided behavior was tested at 100, 1, 0.1, 0.01, and 0.001 cd/m2 and timed performances from 20 trials (four sessions of five trials each) for each mouse at each light-intensity were averaged. Uniform room luminance settings were stably achieved by an engineered adjustable light-source and constantly monitored with a luminance meter LS-100 (Konica Minolta, Tokyo, Japan). To be certain that we were measuring the mice's visual ability only and not memory, the platform was placed pseudo-randomly in the water tank and all external visual cues were eliminated.

Optokinetic evaluation of scotopic and photopic vision

Visual acuity and contrast sensitivity of mice was evaluated from optomotor responses using a two-alternative forced-choice protocol (Umino et al., 2008; Kolesnikov et al., 2011). Briefly, the animal was placed on a pedestal surrounded by four computer monitors and observed using infrared-sensitive television camera. For scotopic measurements, a round array of six infrared LEDs mounted above the platform was used to visualize the animal. Mice responded to visual stimuli (sine-wave vertical gratings presented by computer using staircase paradigm and invisible to the experimenter), by reflexively rotating their head in either clockwise or counterclockwise direction. The observer's task was to track the direction of optomotor responses and report it to the computer which determined the correctness of the choice (Prusky et al., 2004). Photopic visual acuity was determined as the threshold for spatial frequency (Fs) of the stimuli at 100% contrast. Contrast sensitivity was defined as the inverse of obtained contrast threshold values. Photopic responses of constitutive RGS7/11 DKO mice were measured at un-attenuated luminance of monitors (70 cd/m2 at the mouse eye level), over a range of various Sp, from 3 to 50 deg/s. Fs of stimuli was kept constant at its optimal value of 0.128 cyc/deg for all speed regimes thus providing a range of corresponding temporal frequencies (Ft) from 0.38 to 6.4 Hz, according to following equation: Ft = Sp × Fs (Umino et al., 2008). Photopic contrast sensitivities of cDKO mice were determined at specified low (6 deg/s) or high (50 deg/s) Sp at 70 cd/m2 luminance level. Scotopic contrast sensitivity of cDKO mice was evaluated at optimal values of Ft (0.75 Hz) and Sp (0.1 cyc/deg), and a corresponding stimuli speed of 7.5 deg/s. In this mode, the monitor luminance was attenuated to ∼3 × 10−4 cd/m2 by several neutral density film filters that formed a cylinder around the animal. All data were analyzed using independent two-tailed Student t-test, with accepted significance level of p < 0.05.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
    The light response of ON bipolar neurons requires G[alpha]o
    1. A Dhingra
    2. A Lyubarsky
    3. MS Jiang
    4. EN Pugh Jr
    5. L Birnbaumer
    6. P Sterling
    7. N Vardi
    (2000)
    Journal of Neuroscience 20:9053–9058.
  11. 11
    ‘mGlu receptors in the retina’–WIREs membrane transport and signaling
    1. A Dhingra
    2. N Vardi
    (2012)
    Wiley Interdisciplinary Reviews Membrane Transport and Signaling 1:641–653.
    https://doi.org/10.1002/wmts.43
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
    The first steps in seeing
    1. RW Rodieck
    (1998)
    Sunderland: Sinauer.
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56

Decision letter

  1. Jeremy Nathans
    Reviewing Editor; Howard Hughes Medical Institute, Johns Hopkins University School of Medicine, United States

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for choosing to send your work entitled “Sensitivity and kinetics of transmission at the first visual synapse differentially impact visually-guided behavior” for consideration at eLife. Your full submission has been evaluated by three outside experts and a member of the board of reviewing editors. The four reviewers are impressed with the high quality of the data, the breadth of the analysis, and the importance of the topic. However, there are methodological issues that would need substantial clarification, re-analysis, and/or additional experimentation for the manuscript to be considered at eLife. The eLife approach to this situation is somewhat different from the approach used by other journals. At eLife, we do not want to give the authors a list of major changes that must be made in order to continue with the review process. Instead, a submitted manuscript must be reasonably close to its final form at the time of submission for the review process to proceed to acceptance. We have therefore decided to reject the manuscript, but we want to provide you with the full feedback from the reviewers. We hope that this feedback will be helpful to you going forward. We also note that if you decide to substantially revise the manuscript in light of the comments and resubmit it as a new submission, we would be happy to consider it.

Reviewer #1:

Although the experimental data presented in this paper appears thorough, there are several problems with the analysis of the data that currently cast doubt on the interpretations.

1) Separation of rod and cone components of ERG b-wave signals. Figure 5C purports to plot the response-intensity dependence of the rod-driven component of the ERG b-wave, under control conditions and after tamoxifen application. However, it is crucial to demonstrate that the amplitudes plotted here are genuinely rod-driven, and unfortunately the raw data are not illustrated. What needs to be shown are sets of actual b-wave responses, across the full range of flash intensities, together with the extraction of the rod-driven and cone-driven components; these responses are not available, and so it is not possible to evaluate the measurements. Likewise, the fits of Equation 1 (in the subsection headed “Electroretinography”) to the measured amplitudes are not illustrated; these fits need to be illustrated for two reasons. First, it seems unusual that an exponent of unity (rather than a power law) is being applied. But secondly, the magnitude of the cone-driven component (listed in Table 1) is so large (>600 uV, even larger than the rod-driven component of ∼500 uV) that the extraction of the rod-driven component appears susceptible to the precise fitting procedures; however, the significance of this potential problem cannot be evaluated with seeing the results.

2) Figure 6: Sensitivity (6B), and half-saturation normalized response (6C). In Figure 6B, the measurements of sensitivity appear to have been taken from Table 3, where only the responsive cells have been included in the averages. The authors need to deal with the issue of the unresponsive cells, because response amplitude and sensitivity appear to be correlated. Thus, the omission of sensitivity measurements from the unresponsive (and presumably very insensitive) cells would seem to bias the plotted measurements.

The text (in the subsection headed “Progressive loss of Rgs7 changes the kinetics and sensitivity of light-evoked responses in rod ON-BCs”), and the axis label and caption to Figure 6C, give very confusing information about what is plotted. The caption and axis label refer to “Half-saturation (normalized) response” as a percentage of maximal. But my understanding of a half-saturating response is that it is half-maximal (i.e. 50%), so I cannot understand what is being plotted. The text refers to “sensitivity” but I cannot see any measure of sensitivity plotted. There seems to be something important here that the authors are not explaining to the reader.

3) Plots in Figure 9. In each of panels A, C and D in Figure 9, the authors show the extrapolated behavior at very low Rgs7 concentrations asymptoting to a level similar to that obtained at around 20% concentration. In the absence of measurements at lower concentrations, this appears somewhat implausible. For example, in panel C, why would the half-saturation intensity suddenly stop increasing as the Rgs7 concentration dropped below 20%? And hence, how reliable is this asymptotic level (which is used for curve fitting)? For the data in panel C, it would seem much more informative to plot sensitivity (rather than insensitivity), and to determine the level of Rgs7 that halves the sensitivity.

Reviewer #2:

This paper used a conditional knockout approach to study how altering the concentration of RGS proteins, which regulate G-protein inactivation, influences visual processing. The approach is a powerful one and the paper contains several interesting observations. Several issues, however, make interpretation of the data difficult.

Major issues:

Comparison of dependence of photopic and scotopic behavioral sensitivity on RGS concentration. The apparent difference in the dependence of photopic and scotopic behavioral sensitivity on RGS concentration is a particularly interesting result of the paper, and it is highlighted as such in the Discussion. Interpretation of this result is difficult, however, since the behavioral task itself differed. How can you rule out that the difference reflects the different task rather than the difference in light level and which retinal circuits are active? Specifically, the concern would be that the sensitivity to changes in the time course of the bipolar responses under photopic conditions reflects the use of a task requiring detection of flickering light inputs, while the task used under scotopic conditions may impose less stringent constraints on temporal sensitivity. This is a critical issue as it undercuts what is probably the most interesting result in the paper. I think it is essential that the same behavioral task be used to evaluate signaling via rod and cone circuits.

Time course of changes in behavior and ERG. The lack of sensitivity of the scotopic behavioral measurements on the ERG amplitude (in the subsection headed “Kinetics and sensitivity of rod ON-BC responses and behavioral performance correlate with different RGS levels”, and Figure 9) is somewhat unexpected. Why would this be the case? Sensitivity is a somewhat artificial measurement from a biological perspective because of the normalization, so it is particular unexpected that behavior tracks it rather than response amplitude. Could initially both signal and noise decline with declining RGS concentration such that behavioral threshold remains fixed? If this is the case, it should be apparent from noise recorded in the rod bipolar cells. An expanded discussion of this issue is needed.

The paper could be written much more clearly. There are numerous sentences that I struggled to interpret. A few examples are:

Abstract: “We demonstrate that key parameters of…”;

Introduction: “Combined with observed modulation…”;

Subsection “Progressive loss of Rgs7 changes the kinetics and sensitivity of light-evoked responses in rod ON-BCs”: “we first examined retinal responses…”;

Subsection “Saturation of RGS concentration in ON-BC ensures reproducibility of signal transfer”: “Notably, a substantial…”.

There are other examples, and generally the paper needs a careful editing job. There are also some very long paragraphs that could get broken up to help a reader, for example the first paragraph of Results.

Other issues:

The Abstract does not give a good idea of the results in the paper. For example, it does not state that the key manipulation in the paper is to alter the concentration of RGS proteins.

Introduction: Sampath and Rieke is probably not the best reference for the role of Go in On-BC response recovery.

Results, first paragraph, and Figure 2: the similarity of the synaptic connectivity shown in Figure 2E is important as the authors point out. But nothing is done to quantify such similarity. A similar issue arises in the discussion of Figure 3.

The evidence for 1:1:1 stoichiometry (Figure 4) is not particularly strong and not uniquely supported by the data. That part of the text needs to be revised to more accurately reflect the actual data.

Heterogeneity of rod bipolar responses: The ERG data and plots based on it suggest a smooth change in response properties as RGS concentration declines. But the single cell recordings seem to provide a much different picture, with large variability among cells. It would help to document that variability in more detail and explain to a reader why the heterogeneity is not a concern in the comparison of the ERG and behavioral data.

Why do rod bipolar responses change in shape so much? It is not immediately obvious why the dependence of the slope of the rising phase of the response on flash strength should change.

In the subsection headed “Kinetics and sensitivity of rod ON-BC responses and behavioral performance correlate with different RGS levels”: the error bars for the dependence of response properties on RGS concentration seem very small given that they should incorporate error in both determination of RGS concentration and in the response parameter of interest.

Inclusion of recordings from ON cone bipolar cells would enhance the paper. It is not clear why these are not part of the paper at present.

Reviewer #3:

The paper by Sarria et al presents a very nice set of data showing that RGS proteins in the ON bipolar cells are likely present at higher concentration than mGluR6, presumably to ensure that the critical step of deactivating Go is reliable. Interestingly, the effect of reduced RGS on visually driven behavior is not identical in the rod and cone pathways. While in scotopic vision, change in behavior correlates with response sensitivity; in photopic vision, change in behavior correlates with change in kinetics. The paper is well written, but the Abstract is not sufficiently informative.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for sending your work entitled “Sensitivity and kinetics of transmission at the first visual synapse differentially impact visually-guided behavior” for consideration at eLife. Your article has been favorably evaluated by K VijayRaghavan (Senior editor), Jeremy Nathans (Reviewing editor) and four reviewers, including the three who reviewed the first version of the manuscript. One of the reviewers, Noga Vardi, has agreed to share her identity.

I am including the four reviews at the end of this letter, as there are many specific comments in them that will not be repeated in the summary here.

In general, the reviewers were impressed with the importance and novelty of your work. Overall, the experiments appear to be carefully executed. However, all four reviewers find that the data is difficult to interpret in some places, and that the writing needs to be sharpened. This is a complex body of work and clear writing is critical.

We would like to encourage you to resubmit a revised manuscript that addresses the specific issues raised in the reviews below. With respect to the writing, eLife is trying to restore some balance to the world of scientific writing. We welcome self-critical comments and we believe that such comments actually enhance the reader's ability to judge the science fairly.

I will offer two examples of scientific writing that I think illustrates the preceding point. They are from two papers on ubiquitin that form the core of the discoveries for which the authors shared the Nobel Prize:

1) Proposed role of ATP in protein breakdown: conjugation of protein with multiple chains of the polypeptide of ATP-dependent proteolysis. Hershko A, Ciechanover A, Heller H, Haas AL, Rose IA. Proc Natl Acad Sci U S A. 1980 Apr;77(4):1783-6.

In the second paragraph of the Discussion section, after summarizing the evidence that leads the authors to propose the conjugation of APF-1 (=ubiquitin) to proteins as a way of marking them for degradation, the authors write: “Evidence that APF-1-proteins are intermediates in the breakdown of denatured protein as proposed in Figure 6 is indirect and inconclusive at this time.”

2) Activation of the heat-stable polypeptide of the ATP-dependent proteolytic system. Ciechanover A, Heller H, Katz-Etzion R, Hershko A. Proc Natl Acad Sci U S A. 1981 Feb;78(2):761-5.

In the last paragraph of the Discussion section, after describing their discovery and characterization of ubiquitin ligase, the authors write: “Evidence suggesting the role of the APF-1-activating enzyme in conjugation and in ATP-dependent protein breakdown is not conclusive at present.”

For beautiful examples of clear writing, let me suggest the following:

a) Photolyzed rhodopsin catalyzes the exchange of GTP for bound GDP in retinal rod outer segments. Fung B, Stryer L. Proc Natl Acad Sci U S A. 1980 May;77(5):2500-4.

b) Flow of information in the light-triggered cyclic nucleotide cascade of vision. Fung BK, Hurley JB, Stryer L. Proc Natl Acad Sci U S A. 1981 Jan;78(1):152-6.

Reviewer #1:

A general issue of very broad interest in biomedical science is how RGS proteins regulate the kinetics of GPCR signaling. A related specific issue of great interest to neuroscience is whether (and if so, how) the properties of synapses where GPCR signaling is present are tuned by RGS proteins.

The first issue was addressed quantitatively in the field of phototransduction, where it was shown that the expression level of RGS9 controls the limiting rate of deactivation of the GPCR cascade in rod photoreceptors activated by photon capture (Krispel et al., 2006, Neuron). A useful preparation for investigating the second issue is the synapse between rod and cone photoreceptors with “ON” bipolar cells, as the post-synaptic bipolar cell employs a metabotropic (mGluR6) cascade to control the activation of TRPM1 channels. While the full mGluR6 signaling path in the ON bipolar cells has yet to be determined, there is good published evidence that Rgs7 and Rgs11 are both involved (and that the presence of at least one is essential) in the responses of ON-bipolar cells.

This investigation elegantly uses an inducible expression system to conditionally knock down the expression levels of Rgs7 in Rgs11-/- in adult mice, and reports highly convincing dependence of the kinetics of the in bipolar cells response on the expression level. The quantification of the knockdown, both with Western analysis and with IHC, is impeccable. The electrophysiological data are beautifully presented and convincing in showing that cone- and rod-signaling through ON-bipolars are differentially affected by the knockdowns. Behavioral data show trends consistent with the physiological results, and the general understanding of conditions that favor rod vs. cone signaling. Overall, the paper beautifully characterizes the dependence of the light responses of rod- and cone- ON-bipolars on the expression level of RGS proteins.

1) In many places paper is not very well written for a general audience, but rather seems addressed to vision experts. Also, in too many places the authors slip into vague broad claims, instead of telling the reader precisely what can be concluded from the results.

2) Figure 1 presents a nice set of control data, but it's very tedious and wasteful of the reader's time, and should be relegated to the supplement with one or two brief sentences in the main text. The figure presentation should start with the current Figure 2, which presents the knockdown strategy and highly convincing evidence for its success at the molecular and cellular level.

3) Figure 4 is very powerful. To help the reader two more panels should be added: (1) a plot of Rmax, rods vs. Rgs7 level, and of time to peak (scotopic b-wave) vs. Rgs7 level.

4) Figure 5-7. It's very puzzling that the saturating ERG b-wave amplitude is systematically reduced as Rgs7 expression level declines, while the saturating amplitudes of the individual bipolars is not reduced. Could it be that the loss of Rgs7 and 11 causes the membrane potential of the bipolars to collapse, so that the driving force on the Trpm1 channels declines in vivo (but provided by the experimenter in single-cell recording)? (What were the zero-current holding potentials?) The authors need to comment on this apparent discrepancy. I think this is a clue to normal RGS function in the cells (see point #7).

5) Figures 8, 9. The behavioral data provide support for the role of Rgs7 and 11 in ON-bipolars in regulating vision, but they are not really explained in a manner that a general reader could understand. There from the physiological data, it seems that there are two general effect of the cKO: (1) the loss of ON-bipolar signaling amplitude (as manifest in the b-waves); (2) the slowing of ON-bipolar response kinetics. The authors should attempt to explain to the reader how the cellular losses explain the specific behavioral deficits. To me, most of the defects (except for that in panel F) seem modest, and thus the behavioral data mainly show that normal ON-bipolar signaling is needed for completely normal visual function. However, the results also show that the loss of these signals does not lead to blindness (complete loss of visual function).

6) The paper seems to reach for the stars and miss the moon at times. For example, to me the conclusion of the Abstract—“we demonstrated that key parameters of synaptic transmission: sensitivity, kinetics, and maximal response amplitude have unique ranges of dependence on RGS concentration. We further show that these parameters differentially impact visually guided- behavior mediated by rod and cone ON pathways. These findings illustrate that neuronal circuit properties can account for varying needs by adjusting parameters of synaptic transmission at individually-defined synapses”—is amazingly obtuse. The phrase “a key parameter of synaptic transmission” isn't conventional or even clear. The parameters in question are values of extracted from ERGs or electrical responses of ON-bipolar cells: presumably the actual synaptic signal transmission kinetics—the decrease in glutamate release and consequent mGluR6 and Go de-activation—are not changed. The “unique ranges of dependence” on RGS concentration is never shown (see comment #3 above about how it might be introduced.) The phrase “these parameters differentially impact…” is gobbledy gook. Rather, the loss of Rgs7 and 11 in cone ON-bipolars apparently results in a more severe phenotype.

7) The paper needs a schematic (say, in the Discussion) that would illustrate how the loss of RGS proteins produces the main ON-bipolar response defects. Even though this would necessarily remain speculative, the general reader needs to be given some understanding of how RGS depletion less to loss of ON-bipolar response amplitude and slowing of the responses. The authors don't ever return to the issue of how their results might impact the understanding of the post-synaptic responses of other neurons known to express RGS proteins.

8) The Introduction could be shortened by at least 30-50% without loss of clarity. The bottom line should be: “… and thus we did the following experiments.”

Reviewer #2:

This “new” submission is essentially a resubmission of the previous manuscript, and I'm afraid I have found it very difficult to evaluate. In some ways the authors have addressed previous criticisms, but on examining the paper more deeply many of the flaws raised previously still seem to be present. For example, it is still problematic to compare the ERG recordings (done with flashes ranging from dim to bright, in darkness or on a single rod-saturating background) with either the behavioral experiments (done at a range of room luminances) or with the OKR measurements (done at visual threshold). The new data in Figure 7 are likewise difficult to interpret, because they seem to be dominated by oscillatory potentials which are not even mentioned; instead the authors appear to take some unspecified measurement within those oscillations as representing the photopic b-wave, because the points plotted in D and E seem very different from what I get looking at the traces in C. Also, it had previously escaped my notice that, while the authors make a big point of the slowing of the ERG b-wave time course in Figure 5D, they do not comment on the corresponding results for single-cell recordings; inspection of Figure 6A suggests that the phenomenon may be very different in the rod ON-BCs themselves.

The authors have clearly done a lot of interesting experiments and they have some very valuable data. But what I can't really see is that as a paper it all hangs together in a sufficiently sound and coherent way. As a bottom line, the authors have not actually convinced me of the interpretation in their Abstract that “We demonstrate that key parameters of synaptic transmission… have unique ranges of dependence on RGS concentration”. Nor have they convinced me “that these parameters differentially impact visually guided-behavior mediated by rod and cone ON pathways”.

Reviewer #3:

The paper by Sarria et al. is a nice comprehensive attempt to understand how the quantity of RGS proteins affects parameters of the light response in ON bipolar cells and further influence different behavioral tasks. An emphasis is placed on comparing scotpic to photopic vision. My previous concern about the photopic ERG has been addressed; it seems that the concern about the behavioral assays (that should allow a direct comparison between rod and cone pathways, and between physiology and behavior) has also been addressed.

My major concern is the quantification of mGluR6:Rgs11:Rgs7. The authors used whole retinas for their quantitative Western blottings; however, mGluR6 and Rgs11 are present only in ON bipolar cells while Rgs7 is present also in the IPL. I saw no attempt to correct for that. If Rgs7 is distributed equally in OPL and IPL (a guess from immunolocalization), then the amount of Rgs7 will be reduced to about 10 fM and Rgs7+Rgs11 will equal mGluR6 amount. This quantification is important for understanding the composition of the mGluR6-related macromolecule and the mechanism for the light response. As analyzed here, it appears that RGSs are at excess, but if the IPL is factored in, then each mGluR6 molecule may have only one RGS molecule to modify its cascade. In fact, if the mGluR6 receptor functions as a dimer, then each receptor will have two RGS molecules, a more realistic outcome. This concern however, does not affect the conclusions of the paper regarding the difference between rods and cones (it may affect the Discussion).

Another concern is the quantification of immunostaining in rod bipolar vs ON cone bipolar cells. The authors do not mention how they determined the puncta identity. A common way is to counter-stain with PNA, but this is not demonstrated or mentioned in Methods. I therefore have to conclude that the authors simply used the pattern for this distinction. While the pattern gives some indication about the identity of the puncta, it is far from being adequate for quantification.

Figure 7. It seems to me that the effect of reduced RGS is much greater on Rmax and I0.5 than on time to peak. The authors should add a graph showing this parameter vs. days after tamoxifen (as they did for scotopic conditions). In particular, we learn later that in photopic conditions, the behavior was mostly affected by amplitude, so it will be nice to see these graphs next to each other.

Reviewer #4:

This paper uses ON bipolar signaling pathways in the retina to correlate the decrease in RGS expression with synaptic transmission and behavior. This is potentially a very powerful approach, and many of the results in the paper are clear and compelling. However, several substantial issues keep the paper from realizing its potential.

Comparison of scotopic and photopic behavior:

One of the central conclusions of the paper is that scotopic and photopic behavior have different sensitivities to properties of bipolar signaling (in the subsection headed “Differential modulation of synaptic transmission parameters in rod and cone ON-BCs”). This is based on Figures 10 and 11, which nicely plot physiological signaling properties and behavior against RGS expression levels. These plots show that scotopic behavior correlates with bipolar sensitivity, while photopic behavior correlates with bipolar kinetics and amplitude. My concern about this analysis is that the behavioral tasks differ. Scotopic behavior was measured using the OKR at speeds corresponding to peak sensitivity. Photopic behavior was measured at higher speeds, since only at those speeds did ON bipolar signaling appear to influence the OKR under photopic conditions. To be convincing, the authors need to show that the particular bipolar feature that behavior correlates with does not depend on task parameters.

Morphology:

A key point in interpreting the results is whether the synapses are altered morphologically or only functionally. The data on this point is not convincing. First, there is little quantification. Second, the resolution of the images provided is not sufficient to evaluate the possibility that synaptic components are rearranged. Third, some key procedures – such as how components at rod and cone synapses were isolated – are not described (or at least I did not find them easily).

Cone bipolar recordings:

Recordings from cone bipolar cells would help interpret the photopic behavioral results and substantially enhance the paper.

Writing:

The writing in many places is sufficiently confusing so as to alter or obscure the conclusions of a sentence. The entire paper needs to be gone through carefully with this issue in mind.

https://doi.org/10.7554/eLife.06358.023

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

The earlier version of this work has been evaluated by eLife, and while the paper was formally rejected at the time we were encouraged to resubmit it as new version provided that we address points that required additional work. We are happy to report that we performed extensive additional studies which we hope should clear the concerns raised by the Reviewers. From the summary of discussion among the reviewers there appears to be two major concerns about our study: 1) separation of rod vs cone components measured by ERG (points A and B) and 2) different behavioral paradigms used to evaluate scotopic (rod-dominated) and photopic (cone-dominated) vision (points C and D).

To address the first concern, we performed additional set of experiments where we measured ERG responses on a rod-suppressing light background across all time points following tamoxifen administration. This intensity of this background was very similar to what was used for the behavioral study to attain proper correspondence of the results. We further ensured that truly scotopic light intensities, which do not activate cones, were used for the analysis of rod pathway by ERG. We added this new information, and re-analyzed all the data. In sum, we are convinced that rod and cone driven ERG responses are cleanly separated.

To address the second concern, we established another behavioral paradigm where we evaluated temporal contrast sensitivity under scotopic conditions using optomotor test. Thus, we are now using the same behavioral test to evaluate vision under both scotopic and photopic conditions. Importantly, the results of the optomotor experiments closely matched the results observed in the water maze task, lending further credence to the interpretation that tuning of rod vs cone ON-BC pathways in respect to their reliance on RGS concentration ranges differentially contributes to behavior. Below are our responses to specific reviewers’ comments.

Reviewer #1:

1) Separation of rod and cone components of ERG b-wave signals. Figure 5C purports to plot the response-intensity dependence of the rod-driven component of the ERG b-wave, under control conditions and after tamoxifen application. However, it is crucial to demonstrate that the amplitudes plotted here are genuinely rod-driven, and unfortunately the raw data are not illustrated. What needs to be shown are sets of actual b-wave responses, across the full range of flash intensities, together with the extraction of the rod-driven and cone-driven components; these responses are not available, and so it is not possible to evaluate the measurements. Likewise, the fits of Equation 1 (in the subsection headed “Electroretinography”) to the measured amplitudes are not illustrated; these fits need to be illustrated for two reasons. First, it seems unusual that an exponent of unity (rather than a power law) is being applied. But secondly, the magnitude of the cone-driven component (listed in Table 1) is so large (>600 uV, even larger than the rod-driven component of ∼500 uV) that the extraction of the rod-driven component appears susceptible to the precise fitting procedures; however, the significance of this potential problem cannot be evaluated with seeing the results.

We now provide the raw plot of full range of b-wave amplitudes elicited across all light intensities (scotopic to photopic) before extraction of rod vs cone components (Figure 5B). We plot the fits according to Equation #1 right on the same graph, which matches experimental data very well. The biphasic curve fit reveals their saturation around predicted intensities for rods and cones respectively. Separately, we show the rod component phase of the response (Figure 5C) with respective fits. Furthermore, we provide raw data showing individual ERG traces elicited by scotopic flashes (Figure 5A). We would like to note that we are not fitting individual traces (Figure 5A) for the analysis but rather extract the peak amplitude values from these traces to be plotted as a function of the flash strength used to elicit them (Figure 5B). Values extracted from plots in Figure 5 correspond to Table 2 (conditional KO), not Table 1 (haploinsufficient strains). From the current Figure 5C it can be seen that rod-driven b-wave components are larger than cone-driven components. Furthermore, the rod component clearly reaches saturation, and thus can be fitted precisely. We also caught a mistake that was driving overestimation of cone-driven b-wave amplitudes in Figure 1 and Table 1. This error is now corrected.

2) Figure 6: Sensitivity (6B), and half-saturation normalized response (6C). In Figure 6B, the measurements of sensitivity appear to have been taken from Table 3, where only the responsive cells have been included in the averages. The authors need to deal with the issue of the unresponsive cells, because response amplitude and sensitivity appear to be correlated. Thus, the omission of sensitivity measurements from the unresponsive (and presumably very insensitive) cells would seem to bias the plotted measurements.

We looked into the issue of unresponsive cells in more detail. After analyzing the data more carefully it became apparent that the fraction of unresponsive cells across the time-course of tamoxifen administration did not substantially change and stayed in the 31% to 46% range.

We thus retracted our earlier claim regarding correlation between tamoxifen administration and increase in the number of unresponsive cells. The appearance of these cells is likely related to stochastic hyperactivation of Cre-recombinase and resultant complete loss of RGS proteins that renders cells completely unresponsive. Given that no parameters are extracted from these cells and that their fraction remains relatively constant throughout the experiment, we think that they are not significantly biasing our sensitivity measurements. Consistent with this, we found almost perfect correspondence in sensitivity measured by both patch clamp and ERG. We clarified these points in the text.

The text (in the subsection headed “Progressive loss of Rgs7 changes the kinetics and sensitivity of light-evoked responses in rod ON-BCs”), and the axis label and caption to Figure 6C, give very confusing information about what is plotted. The caption and axis label refer toHalf-saturation (normalized) responseas a percentage of maximal. But my understanding of a half-saturating response is that it is half-maximal (i.e. 50%), so I cannot understand what is being plotted. The text refers tosensitivitybut I cannot see any measure of sensitivity plotted. There seems to be something important here that the authors are not explaining to the reader.

We corrected Figure 6C to show half-saturating responses plotted in individual graphs and in their respective units for single cell and ERG responses vs. tamoxifen days.

3) Plots in Figure 9. In each of panels A, C and D in Figure 9, the authors show the extrapolated behavior at very low Rgs7 concentrations asymptoting to a level similar to that obtained at around 20% concentration. In the absence of measurements at lower concentrations, this appears somewhat implausible. For example, in panel C, why would the half-saturation intensity suddenly stop increasing as the Rgs7 concentration dropped below 20%? And hence, how reliable is this asymptotic level (which is used for curve fitting)? For the data in panel C, it would seem much more informative to plot sensitivity (rather than insensitivity), and to determine the level of Rgs7 that halves the sensitivity.

We have changed data presentation in these figures (now Figure 10 and 11), which now show sensitivity. We further added more experimental data points (28d and 35d) and the plots now more clearly show changes around asymptotes. The only graph where the last point (18%) shows saturation less clearly is half-saturation plot (Figure 10C and 11C). However, even if the exact value of half-saturation intensity for the 0-18% range of RGS is far off where the fit predicts it to be, it would not significantly affect the [RGS] half value which we estimate from this analysis. Nevertheless, formal analysis of curve fitting by F test for shows that the sigmoidal dose response curve fit is better than using exponential decay with R2=0.0991. Hence, we believe that our estimates of RGS concentration required to change response parameters and behavior are accurate. We further compared all measured parameters according to level of Rgs7 that halves the response as the reviewer suggested and present this data in panels E of Figure 10 and Figure 11.

Reviewer #2:

Comparison of dependence of photopic and scotopic behavioral sensitivity on RGS concentration. The apparent difference in the dependence of photopic and scotopic behavioral sensitivity on RGS concentration is a particularly interesting result of the paper, and it is highlighted as such in the Discussion. Interpretation of this result is difficult, however, since the behavioral task itself differed. How can you rule out that the difference reflects the different task rather than the difference in light level and which retinal circuits are active? Specifically, the concern would be that the sensitivity to changes in the time course of the bipolar responses under photopic conditions reflects the use of a task requiring detection of flickering light inputs, while the task used under scotopic conditions may impose less stringent constraints on temporal sensitivity. This is a critical issue as it undercuts what is probably the most interesting result in the paper. I think it is essential that the same behavioral task be used to evaluate signaling via rod and cone circuits.

We completely agree with the reviewer’s comment. To deal with this issue we evaluated mouse vision using the scotopic contrast sensitivity test. This is exactly the same paradigm that we used to evaluate photopic vision, therefore we think that the data are now directly comparable. Strikingly, we still detect a very large difference in behavioral sensitivity of mice to the RGS loss under photopic vs. scotopic conditions. Importantly, the results obtained in the scotopic contrast sensitivity task also agree very well with measurements in object recognition task in the water maze. Together, these data further boost confidence in the conclusions from this work.

Time course of changes in behavior and ERG. The lack of sensitivity of the scotopic behavioral measurements on the ERG amplitude (in the subsection headed “Kinetics and sensitivity of rod ON-BC responses and behavioral performance correlate with different RGS levels”, and Figure 9) is somewhat unexpected. Why would this be the case? Sensitivity is a somewhat artificial measurement from a biological perspective because of the normalization, so it is particular unexpected that behavior tracks it rather than response amplitude. Could initially both signal and noise decline with declining RGS concentration such that behavioral threshold remains fixed? If this is the case, it should be apparent from noise recorded in the rod bipolar cells. An expanded discussion of this issue is needed.

We think that the reviewer’s intuition here is correct. The traditional definition of sensitivity needs to consider the ability to discriminate light-driven signals from the noise, or a signal-to-noise ratio. While most of the emphasis of the manuscript is on evaluating changes in the light induced changes in signal, we had not examined how elimination of RGS proteins affects noise. In the revised manuscript we separately evaluated changes in noise and this information can now be used to approximate changes in signal-to -noise ratio. While the signal per photon absorbed changes in magnitude (as indicated by their half-maximal flash strength), we detected no change in rod ON-BC noise (Table 3). This indicates that the signal-to-noise ratio is declining with time following tamoxifen administration. The lack of effect on behavioral threshold with declining signal-to-noise ratio may then be a reflection of a limiting signal-to-noise ratio that is superceded by the rod ON-BC signal-to-noise ratio as it declines. A treatment of this point is now included in the Discussion.

The paper could be written much more clearly. There are numerous sentences that I struggled to interpret. A few examples are:

Abstract:We demonstrate that key parameters of…;

Introduction: “Combined with observed modulation…;

Subsection “Progressive loss of Rgs7 changes the kinetics and sensitivity of light-evoked responses in rod ON-BCs”:we first examined retinal responses…;

Subsection “Saturation of RGS concentration in ON-BC ensures reproducibility of signal transfer”:Notably, a substantial….

There are other examples, and generally the paper needs a careful editing job. There are also some very long paragraphs that could get broken up to help a reader, for example the first paragraph of Results.

We have edited the manuscript, hopefully improving its readability.

Other issues:

The Abstract does not give a good idea of the results in the paper. For example, it does not state that the key manipulation in the paper is to alter the concentration of RGS proteins.

We modified the Abstract and it now contains reference to RGS proteins.

Introduction: Sampath and Rieke is probably not the best reference for the role of Go in On-BC response recovery.

Sampath and Rieke were the first to show directly the specific role of G protein deactivation in limiting the onset of ON-BC responses. This reference is cited in the context of Go deactivation. We have additionally included reference to Dhingra and Vardi (Dhingra et al., 2000) who identified Go to be the main G protein species driving depolarizing response of ON-BC.

Results, first paragraph, and Figure 2: the similarity of the synaptic connectivity shown in Figure 2E is important as the authors point out. But nothing is done to quantify such similarity. A similar issue arises in the discussion of Figure 3.

We are unsure about the type of the quantification that might be required here. We see perfect apposition of pre-synaptic (e.g. CtBP2) and post-synaptic (e.g. mGluR6) markers with no difference between genotypes. In the event this alignment is disrupted (as seen in several nob mutants) the synaptic morphology is usually drastically affected and/or either pre- or post-synaptic markers do not reveal staining in the synaptic layer. We have previously examined the morphology of photoreceptor to ON-BC synapses (Cao et al., 2012 PNAS) in mice completely lacking both RGS proteins by EM while quantifying structural features and found that it was not affected by RGS loss.

The evidence for 1:1:1 stoichiometry (Figure 4) is not particularly strong and not uniquely supported by the data. That part of the text needs to be revised to more accurately reflect the actual data.

We revised the manuscript and are now providing exact values for RGS and mGluR6 content measured in our experiments. We would like to note that these experiments were carefully designed and included purified recombinant protein standards spiked into respective mouse knockout retinas. We conducted three independent experiments and generally believe that the data are reliable. We also modified the text to provide more accurate approximation of RGS:mGluR6 stoichiometry.

Heterogeneity of rod bipolar responses: The ERG data and plots based on it suggest a smooth change in response properties as RGS concentration declines. But the single cell recordings seem to provide a much different picture, with large variability among cells. It would help to document that variability in more detail and explain to a reader why the heterogeneity is not a concern in the comparison of the ERG and behavioral data. Why do rod bipolar responses change in shape so much? It is not immediately obvious why the dependence of the slope of the rising phase of the response on flash strength should change.

By its nature, ERG recordings reflect average behavior of thousands of cells. It is common to see some variability in a single cell electrophysiological recordings directly from individual bipolar cells. One of the major conclusions of the paper is that in ON-BC RGS are present in substantial excess and this in part ensures reproducibility of their responses and high sensitivity. Thus, it is quite expected that manipulation with concentration beyond the point where it starts affecting the response properties, produces greater variability. This situation is further compounded by the variability associated with the activation of Cre-recombinase which also varies on cell-by-cell basis. Nevertheless, despite the higher than normal variability we were able to perform quantitative analysis of changes in response sensitivity measured in the single cell patch clamp experiments with a reasonable precision (Figure 6B). Importantly, we directly compare single cell data and ERG and find them to be in close agreement (Figure 6C). We thus believe that heterogeneity of ON-BC responses in our experiments does not detrimentally affect our ability to interpret the data. Furthermore, we mostly base our conclusions on parameters measured by ERG, which show much less variability.

Change in the slope of the light response onset upon RGS loss is expected from the deceleration of the G protein inactivation and resultant slowing of the TRPM1 channel opening and documented previously (Cao et al., 2012 PNAS). Again, we find changes in the response onset timing to be in agreement between ERG and single cell recordings.

In the subsection headed “Kinetics and sensitivity of rod ON-BC responses and behavioral performance correlate with different RGS levels”: the error bars for the dependence of response properties on RGS concentration seem very small given that they should incorporate error in both determination of RGS concentration and in the response parameter of interest.

In the revised figures (Figure 10 and Figure 11) we have included error bars that reflect variability in the determination of RGS concentration (X axis) in addition to errors in calculated parameter of interest (Y axis).

Inclusion of recordings from ON cone bipolar cells would enhance the paper. It is not clear why these are not part of the paper at present.

Cone ON-BCs are rare when sampling randomly from a slice for patch recordings. Although we agree that having single cell data would have been ideal, obtaining a sufficient data set for the analysis of tamoxifen inducible changes in this rare neuronal population was not feasible in the set of experiments presented. However, we do evaluate the average cone-driven responses of cone ON-BC by ERGs and compare their properties to rod-driven ON-BC responses, with little difference in respect to how parameters of interest in these two neuronal populations respond to change in RGS concentration.

Reviewer #3:

The paper is well written, but the Abstract is not sufficiently informative.

We have revised the Abstract to make it more informative.

[Editors' note: the author responses to the re-review follow.]

Reviewer #1:

1) In many places paper is not very well written for a general audience, but rather seems addressed to vision experts. Also, in too many places the authors slip into vague broad claims, instead of telling the reader precisely what can be concluded from the results.

We have edited the paper substantially to increase its readability for a general audience. This includes elaborating and clarifying how we reach conclusions and substantiate claims.

2) Figure 1 presents a nice set of control data, but it's very tedious and wasteful of the reader's time, and should be relegated to the supplement with one or two brief sentences in the main text. The figure presentation should start with the current Figure 2, which presents the knockdown strategy and highly convincing evidence for its success at the molecular and cellular level.

We understand the reviewer’s point and agree that moving current Figure 1 to the supplement would have probably been a better strategy. However, eLife policies state that every piece of supplemental material should be related to a particular main figure. In the case of our Figure 1, the material presented provides stand-alone body of data with no logical connection to other figures. In fact, our first submission to eLife had this data in the supplement but we were asked to make it a main figure. The figure represents an important pretext to the current work, the fact that knocking down RGS expression to 25% of normal has little effect on the light response properties.

3) Figure 4 is very powerful. To help the reader two more panels should be added: (1) a plot of Rmax, rods vs. Rgs7 level, and of time to peak (scotopic b-wave) vs. Rgs7 level.

These analyses are provided in Figure 10A and B. Our concern is that it might be premature to discuss the relationship between RGS levels and their influence on ERG parameters before we present ERG data and walk readers through our analysis of the response parameters.

4) Figure 5-7. It's very puzzling that the saturating ERG b-wave amplitude is systematically reduced as Rgs7 expression level declines, while the saturating amplitudes of the individual bipolars is not reduced. Could it be that the loss of Rgs7 and 11 causes the membrane potential of the bipolars to collapse, so that the driving force on the Trpm1 channels declines in vivo (but provided by the experimenter in single-cell recording)? (What were the zero-current holding potentials?) The authors need to comment on this apparent discrepancy. I think this is a clue to normal RGS function in the cells (see point #7).

We found that response amplitudes of rod ON-BC are reduced when measured by either ERG or single cells recordings with very good correspondence between these very different methods. The Y scale in Figure 6A now has been set to the same range for every family to show more effectively the reduction in the amplitude of single cell responses and this decline is further documented by the statistical analysis of responses from multiple cells presented in Table 3.

5) Figures 8, 9. The behavioral data provide support for the role of Rgs7 and 11 in ON-bipolars in regulating vision, but they are not really explained in a manner that a general reader could understand. There from the physiological data, it seems that there are two general effect of the cKO: (1) the loss of ON-bipolar signaling amplitude (as manifest in the b-waves); (2) the slowing of ON-bipolar response kinetics. The authors should attempt to explain to the reader how the cellular losses explain the specific behavioral deficits. To me, most of the defects (except for that in panel F) seem modest, and thus the behavioral data mainly show that normal ON-bipolar signaling is needed for completely normal visual function. However, the results also show that the loss of these signals does not lead to blindness (complete loss of visual function).

We revised the manuscript to provide a more detailed explanation of the behavioral data and their interpretation. The behavioral deficits that we measure for the rod pathway is straightforward: complete loss of scotopic vision. The phenotype is not modest by any means and relationship between the lack of b-wave and night blindness is well established in humans and animal models. “How the cellular losses explain the specific behavioral deficits” is the main investigative line of the paper and we present our thinking about this point in the Discussion, which we have carefully revised and edited for clarity.

6) The paper seems to reach for the stars and miss the moon at times. For example, to me the conclusion of the Abstract—“we demonstrated that key parameters of synaptic transmission: sensitivity, kinetics, and maximal response amplitude have unique ranges of dependence on RGS concentration. We further show that these parameters differentially impact visually guided- behavior mediated by rod and cone ON pathways. These findings illustrate that neuronal circuit properties can account for varying needs by adjusting parameters of synaptic transmission at individually-defined synapses.—is amazingly obtuse. The phrasea key parameter of synaptic transmissionisn't conventional or even clear. The parameters in question are values of extracted from ERGs or electrical responses of ON-bipolar cells: presumably the actual synaptic signal transmission kinetics—the decrease in glutamate release and consequent mGluR6 and Go de-activation—are not changed. Theunique ranges of dependenceon RGS concentration is never shown (see comment #3 above about how it might be introduced.) The phrasethese parameters differentially impact...is gobbledy gook. Rather, the loss of Rgs7 and 11 in cone ON-bipolars apparently results in a more severe phenotype.

We have done our best to revise the manuscript and improve clarity. Specifically, we rephrased the statements mentioned by the reviewer hopefully clarifying the points that we were trying to make. The data establishing the dependence of ERG parameters on RGS concentration are presented in Figures 10 and 11. As we mention in our response to comment #3, the analysis that the reviewer is asking for is already reported and serves as a basis for our conclusions. In the revised manuscript, we clarified how we analyzed these changes and used this analysis to inform the conclusions.

7) The paper needs a schematic (say, in the Discussion) that would illustrate how the loss of RGS proteins produces the main ON-bipolar response defects. Even though this would necessarily remain speculative, the general reader needs to be given some understanding of how RGS depletion less to loss of ON-bipolar response amplitude and slowing of the responses. The authors don't ever return to the issue of how their results might impact the understanding of the post-synaptic responses of other neurons known to express RGS proteins.

We have added a schematic to provide our interpretation of this process (Figure 12).

8) The Introduction could be shortened by at least 30-50% without loss of clarity. The bottom line should be:… and thus we did the following experiments.

We revised the manuscript and shortened the Introduction section where we could. However, we felt that dramatic reduction in length might not do the service to a general reader, as it would not allow us to introduce broadly relevant problems at hand.

Reviewer #2:

For example, it is still problematic to compare the ERG recordings (done with flashes ranging from dim to bright, in darkness or on a single rod-saturating background) with either the behavioral experiments (done at a range of room luminances) or with the OKR measurements (done at visual threshold).

All experiments were performed under 2 conditions: scotopic to measure rod pathway and photopic to assess cone pathway function.

For the photopic conditions, we used a rod-suppressing background of 50 cd/m2 in ERG while applying 5 to 1000 cd*s/m2 flashes with 10 to 260 cd*s/m2 half-saturating light intensities used to derive kinetic parameters and 70 cd/m2 in photopic OKR making the amount of light delivered to the retinas under different paradigms comparable and representative of the truly photopic conditions.

For the scotopic conditions the flash intensities in ERG ranged from 2.5x10-4 to 2.5x10-1 to cd*s/m2 and equivalent flash intensities were used for the single cell recordings. Half-saturating light intensities under different tamoxifen treatment time points ranged between 7x10-4 and 6x10-2 cd*s/m2 and were used for the analysis of kinetic parameters (time to peak). For the behavior, we used 3x10-4 (OKR) and 5x10-3 (water maze) cd/m2, which are significantly above the visual threshold. The reviewer is correct in pointing out the caveats of comparing electrical responses to behavior under scotopic conditions, mainly because behavioral assessment requires steady light illumination whereas electrophysiological recordings use flashes on a dark background. However, we matched the conditions, to the extent possible, and used alternative behavioral paradigms to evaluate scotopic vision. The light intensities used for the electrophysiological characterization and behavioral evaluation were in the same range corresponding to pure scotopic conditions. Using a mouse model of abrogated rod function (Gnat1-KO) we have previously determined that under those conditions cone-driven responses are negligible and do not impact OKR behavior (Kolesnikov et al., 2011). Furthermore, we found no discernable difference in visual function deterioration when comparing the results of OKR to water maze. Similarly, the results of single cell experiments closely matched data from ERG. Considering these points we believe that we can draw valid conclusions from comparing rod versus cone ON-BC responses and visually–guided behavior mediated by rod and cone circuits.

The new data in Figure 7 are likewise difficult to interpret, because they seem to be dominated by oscillatory potentials which are not even mentioned; instead the authors appear to take some unspecified measurement within those oscillations as representing the photopic b-wave, because the points plotted in D and E seem very different from what I get looking at the traces in C.

Oscillatory potentials are a very common feature of photopic ERG. Several approaches have been developed to deal with them when analyzing b-wave parameters. Common ways include either frequency filtering the data or applying local smoothing algorithm to the traces with the least square function to eliminate orderly waveform fluctuations (i.e. see Lyubarsky, 1999, JNeurosci., 442; Robson, 2003, JPhysiol, 509). In our analysis, we have used the smoothing approach for normalizing the data. However, we do show raw data before filtering. In the revised manuscript we are providing a more detailed description of this approach and show the results on the manipulation in Figure 7–figure supplement 1.

Also, it had previously escaped my notice that, while the authors make a big point of the slowing of the ERG b-wave time course in Figure 5D, they do not comment on the corresponding results for single-cell recordings; inspection of Figure 6A suggests that the phenomenon may be very different in the rod ON-BCs themselves.

In fact, we do see similar slowing in the onset kinetics of the currents recorded from the rod ON-BC by single cell as we do by ERG. Referencing trace peaks in Figure 6A to the “X” axis clearly shows that the time to peak increases about 2-fold over the course of 35 day tamoxifen treatment: from ∼150 to ∼ 300 ms. We did not formally analyze time-to-peak parameter from the single cell recordings due to more than usual variability of the responses that become apparent upon RGS loss. We comment on this in the manuscript.

Reviewer #3:

My major concern is the quantification of mGluR6:Rgs11:Rgs7. The authors used whole retinas for their quantitative Western blottings; however, mGluR6 and Rgs11 are present only in ON bipolar cells while Rgs7 is present also in the IPL. I saw no attempt to correct for that. If Rgs7 is distributed equally in OPL and IPL (a guess from immunolocalization), then the amount of Rgs7 will be reduced to about 10 fM and Rgs7+Rgs11 will equal mGluR6 amount. This quantification is important for understanding the composition of the mGluR6-related macromolecule and the mechanism for the light response. As analyzed here, it appears that RGSs are at excess, but if the IPL is factored in, then each mGluR6 molecule may have only one RGS molecule to modify its cascade. In fact, if the mGluR6 receptor functions as a dimer, then each receptor will have two RGS molecules, a more realistic outcome. This concern however, does not affect the conclusions of the paper regarding the difference between rods and cones (it may affect the Discussion).

Our investigations indicate that Rgs7 in the retina is specifically present in appreciable quantities only in the OPL. It is true that immunostaining picks up the signal in the IPL, however we found that signal to be completely non-specific as it was also present in DKO retinas lacking Rgs7. We now present this data in Figure 4–figure supplement 1. In this study we used KO samples to control for the specificity of the bands that we designate as Rgs7 by western blotting as well. Therefore, we think that our estimate of absolute Rgs7 content in the retinas is accurate and reflect primarily the amount present in the ON-BC synapses.

We thank the reviewer for reminding us that mGluR6 exists as a constitutive dimer and functions as a single unit. This consideration, indeed calls for an upward adjustment of our RGS:mGluR6 ratio from ∼ 2:1 to ∼ 3:1. We discuss this calculation and its implications in the revised version of the manuscript.

Another concern is the quantification of immunostaining in rod bipolar vs ON cone bipolar cells. The authors do not mention how they determined the puncta identity. A common way is to counter-stain with PNA, but this is not demonstrated or mentioned in Methods. I therefore have to conclude that the authors simply used the pattern for this distinction. While the pattern gives some indication about the identity of the puncta, it is far from being adequate for quantification.

We identify cone synapses by counter-staining with b-arrestin which specifically labels cone terminals. In addition, as reviewer mentioned cone synapses are located in a distinct sublamina of OPL and can be easily identified by a characteristic clustering pattern. We use both hallmarks to distinguish between rod and cone synapses. We clarified this point in the revised manuscript and provided a supplementary figure (Figure 1–figure supplement 1) to illustrate how we distinguish cone synapses.

Figure 7. It seems to me that the effect of reduced RGS is much greater on Rmax and I0.5 than on time to peak. The authors should add a graph showing this parameter vs. days after tamoxifen (as they did for scotopic conditions). In particular, we learn later that in photopic conditions, the behavior was mostly affected by amplitude, so it will be nice to see these graphs next to each other.

We provide the analysis requested by the reviewer in Figure 10 and Figure 11 where we show how Rmax, I0.5 and time-to-peak depend on RGS concentration (days after tamoxifen) side by side.

Reviewer #4:

Comparison of scotopic and photopic behavior:

One of the central conclusions of the paper is that scotopic and photopic behavior have different sensitivities to properties of bipolar signaling (in the subsection headed “Differential modulation of synaptic transmission parameters in rod and cone ON-BCs”). This is based on Figures 10 and 11, which nicely plot physiological signaling properties and behavior against RGS expression levels. These plots show that scotopic behavior correlates with bipolar sensitivity, while photopic behavior correlates with bipolar kinetics and amplitude. My concern about this analysis is that the behavioral tasks differ. Scotopic behavior was measured using the OKR at speeds corresponding to peak sensitivity. Photopic behavior was measured at higher speeds, since only at those speeds did ON bipolar signaling appear to influence the OKR under photopic conditions. To be convincing, the authors need to show that the particular bipolar feature that behavior correlates with does not depend on task parameters.

Two previous studies addressed the issue of mouse performance in OKR task under scotopic conditions. Umino et al. (2008) showed that scotopic contrast sensitivity shows an inverted U dependence on temporal frequency with maximum at ∼0.75 Hz (corresponding to 7.5 deg/s stimuli speed at optimal spatial frequency of ∼0.1 cyc/deg) that sharply declines at speeds exceeding 20 deg/s. In contrast, photopic contrast sensitivity of WT mice is far less reduced at high speeds (as also demonstrated in our Figure 9C) at similar optimal spatial frequency. Unfortunately, these characteristic differences in mouse OKR responses in the two light regimens did not allow us to use the same set of parameters in our scotopic and photopic OKR measurements. Considering that scotopic OKR responses become unreliable above 20 deg/s and that we observe deterioration of visual function, we chose to only use optimal parameters for scotopic OKR evaluation. On the other hand, since the constitutive Rgs7/11 DKO mice did not exhibit any deficits in their photopic OKR behavior at lower stimuli speeds (Figure 9C), we restricted our analysis to high speed stimulation where the difference between control and cCKO mice was clearly observed.

A second study by Kolesnikov et al. (2011) examined scotopic OKR when rod pathway function was severely reduced but not completely eliminated (Gngt1 knockout). In that paper, we found that diminishing rod signaling produces parallel down-ward shift in the behavioral performance across spatial frequencies shifting only light sensitivity. Because loss of synaptic transmission (induced by RGS proteins in the ON-BC) is expected to have similar effect on vision as decrease in rod photosensitivity, we do not expect our cDKO model to be any different in terms of the temporal frequency dependence.

Having said this, we would like to emphasize that we evaluated scotopic vision using 2 different behavioral paradigms: OKR and water-maze and found the results to be in almost perfect correspondence. Thus, we think that differences between the tasks that we use do not influence our interpretations.

Morphology:

A key point in interpreting the results is whether the synapses are altered morphologically or only functionally. The data on this point is not convincing. First, there is little quantification. Second, the resolution of the images provided is not sufficient to evaluate the possibility that synaptic components are rearranged. Third, some key procedures – such as how components at rod and cone synapses were isolated – are not described (or at least I did not find them easily).

We have previously shown that complete elimination of Rgs7 and Rgs11 even early in development does not affect synaptic morphology and the ability of photoreceptors to make synaptic contacts with ON-BC (Cao et al., 2012; PNAS). To address reviewers’ comment we added higher power image documenting intact TRPM1 synaptic clustering and apposition of pre and post synaptic markers (Figure 2–figure supplement 2). We also performed quantification of histochemical architecture in cDKO retinas post tamoxifen treatment. Finally, we expanded our description of how we identify and quantify rod and cone synapses, providing additional supporting evidence (Figure 2–figure supplement 1).

Cone bipolar recordings:

Recordings from cone bipolar cells would help interpret the photopic behavioral results and substantially enhance the paper.

Cone ON-BCs are rare when sampling randomly from a slice for patch recordings. Although we agree that having single cell data would have been ideal, obtaining a sufficient data set for the analysis of tamoxifen inducible changes in this rare neuronal population was not feasible in the set of experiments presented (note that the data on OFF cone bipolar cells in Arman and Sampath (2012) required nearly 1.5 years to collect, and this is for one time point). However, we do evaluate the average cone-driven responses of cone ON-BC by ERGs and compare their properties to rod-driven ON-BC responses, with little difference in respect to how parameters of interest in these two neuronal populations respond to change in RGS concentration.

Writing:

The writing in many places is sufficiently confusing so as to alter or obscure the conclusions of a sentence. The entire paper needs to be gone through carefully with this issue in mind.

We appreciate the reviewer’s attention to writing issues. We revised the manuscript editing it for grammar and usage, hopefully making the result more readable.

https://doi.org/10.7554/eLife.06358.024

Article and author information

Author details

  1. Ignacio Sarria

    Department of Neuroscience, The Scripps Research Institute, Jupiter, United States
    Contribution
    IS, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  2. Johan Pahlberg

    Jules Stein Eye Institute, Department of Ophthalmology, University of California, Los Angeles, Los Angeles, United States
    Contribution
    JP, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0002-4038-9693
  3. Yan Cao

    Department of Neuroscience, The Scripps Research Institute, Jupiter, United States
    Contribution
    YC, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  4. Alexander V Kolesnikov

    Department of Ophthalmology and Visual Sciences, Washington University in St.Louis, St. Louis, United States
    Contribution
    AVK, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  5. Vladimir J Kefalov

    Department of Ophthalmology and Visual Sciences, Washington University in St.Louis, St. Louis, United States
    Contribution
    VJK, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  6. Alapakkam P Sampath

    Jules Stein Eye Institute, Department of Ophthalmology, University of California, Los Angeles, Los Angeles, United States
    Contribution
    APS, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  7. Kirill A Martemyanov

    Department of Neuroscience, The Scripps Research Institute, Jupiter, United States
    Contribution
    KAM, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    kirill@scripps.edu
    Competing interests
    The authors declare that no competing interests exist.

Funding

National Institutes of Health (NIH) (EY018139)

  • Kirill A Martemyanov

National Institutes of Health (NIH) (DA026405)

  • Kirill A Martemyanov

National Institutes of Health (NIH) (EY017606)

  • Alapakkam P Sampath

National Institutes of Health (NIH) (EY019312)

  • Vladimir J Kefalov

National Institutes of Health (NIH) (EY021126)

  • Vladimir J Kefalov

National Institutes of Health (NIH) (EY002687)

  • Kirill A Martemyanov

National Institutes of Health (NIH) (EY000331)

  • Kirill A Martemyanov

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Mrs Martemyanova for help with animal breeding and genotyping and Dr Alicia Faruzzi Brantley for help with setting up behavioral experiments. This work was supported by NIH grants EY018139 (KAM), DA026405 (KAM), EY017606 (APS), EY019312 and EY021126 (VJK), EY002687 to the Department of Ophthalmology and Visual Sciences at Washington University, and EY000331 to the Jules Stein Eye Institute at UCLA.

Ethics

Animal experimentation: All procedures were carried out in accordance with the National Institute of Health guidelines and were granted formal approval by the Institutional Animal Care and Use Committees of the Scripps Research Institute (IACUC protocol number 14-001), Washington University (IACUC protocol number 20140236), and the University of Southern California (IACUC protocol number 10890).

Reviewing Editor

  1. Jeremy Nathans, Howard Hughes Medical Institute, Johns Hopkins University School of Medicine, United States

Publication history

  1. Received: January 6, 2015
  2. Accepted: April 11, 2015
  3. Accepted Manuscript published: April 16, 2015 (version 1)
  4. Version of Record published: April 29, 2015 (version 2)

Copyright

© 2015, Sarria et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,083
    Page views
  • 255
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Scopus, Crossref.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Computational and Systems Biology
    2. Neuroscience
    Christopher M Kim, Carson C Chow
    Research Article
    1. Human Biology and Medicine
    2. Neuroscience
    Ming Song et al.
    Research Article Updated