What makes rods effective in low light

The model predicted saturation lack of photocurrent modulations at the higher irradiances used in our experiments Fig. However, while the model predicted rod saturation at high light levels, modulations in the concentration of unbleached rhodopsin red and orange curves for moderate and strong contrast were retained at all irradiances see inset in Fig.

This indicates that if the model incorporated a reduced gain of the phototransduction cascade, this might prevent saturation and allow rods to respond to this stimulus. Computational model of rod light responses. Left y -axis: elicited photocurrents to moderate blue or high cyan contrast stimulus; right y -axis, logarithmic: number of unbleached rhodopsin molecules in outer segment to moderate red or high orange contrast stimulus. Inset: Detailed view of number of rhodopsin molecules.

Both insets are plotted on the same scale. Black: no translocation replicated blue curve from b. Dark gray: smooth translocation replicated blue curve from d. Light gray: step-like translocation. They thus did not account for any aspect of photoreceptor light adaption, such as well-established irradiance-dependent translocation of elements of the phototransduction cascade between inner and outer segments 21 , For example, at higher light levels, arrestin moves into the outer segment, increasing its effective concentration; while transducin and recoverin leave the outer segment, reducing their concentration.

These translocations have the net effect of reducing the gain of the phototransduction cascade, thereby contributing a mechanism of light adaptation. We modified the model to include these translocation events Fig. This modification resulted in modulated photocurrents at all light levels Fig. Second, saturation was less apparent for stronger stimulus contrast compare cyan and blue traces in Fig.

Third, the time course of response recovery was faster at higher light levels see also magnified views in Fig. Thus, translocation of transducin, arrestin, and recoverin in the model was crucial to enable photocurrents at high light levels. However, the specific time course of the translocation events was less important. We tested the effects of implementing the translocations as artificial, step-like events dashed lines in Fig. As expected, such immediate concentration changes resulted in more rapid recovery in photoresponse light gray vs.

In other words, the light-level-dependent time course of photocurrent recovery needs to be influenced by an additional component, beyond the translocation events. Interestingly, faster response re-emergence at higher irradiances coincided with the increased rate of rhodopsin bleaching red curve in Fig.

Rhodopsin bleaching reduces the rate of isomerization events and might thus be one of the mechanisms allowing for rod responses at high light levels, similar to the suggested role of bleaching adaptation in cones A technical challenge in achieving this was that the bright background light would itself bleach exogenous chromophore.

While application of chromophore is generally considered to support visual responses in the ex vivo preparation, one would expect the opposite in the case of bleaching adaptation, with responses to contrast steps at the bright background suppressed following epochs of chromophore application.

Bleaching adaptation supports recovery of contrast sensitivity at high irradiance. Test for statistical significance was performed with a permutation test Methods. In accordance with our earlier results Fig. Conversely, when adding 9- cis retinal during the dim period, responses of some ganglion cells recovered toward the end of the dim period, indicating that the lack of responses in the corresponding control trials was at least partly attributable to rhodopsin bleach.

This indicates that enhanced regeneration of rhodopsin had detrimental effects on ganglion cell responses at high light levels, consistent with the hypothesis that rhodopsin bleaching is indeed one of the mechanisms that allow rods to escape saturation in bright backgrounds.

Interestingly, responses to lower contrast stimuli were more strongly affected by rhodopsin regeneration than responses to higher contrast Fig. Here, we characterized rod-driven visual responses in the mouse over a large irradiance range, from scotopic to high photopic light levels. All our data—ERG and ganglion cell recordings from isolated retina ex vivo, and recordings of LGN responses in vivo—shows that rods can drive visual responses to moderate contrast stimuli, within the range encountered regularly during natural viewing 24 , at all physiologically relevant irradiances.

It further shows that, at photopic backgrounds, this capacity is associated with a recovery in contrast sensitivity over extended exposure, and that this recovery occurs faster at higher backgrounds. Such rod-driven responses were not only readily observed in cone-deficient mice, but also contributed to the visual responses in visually intact animals.

Together, our data show that the presence and re-appearance of rod responses depends not simply on the background intensity, but on the interaction between background intensity and the duration of exposure. Surprisingly and counter-intuitively, both higher background intensity and prolonged exposure result in reduced saturation, i.

We initially aimed to use transgenic cone-deficient mice to describe the transition to rod saturation. The concept of rod saturation at bright backgrounds is widely accepted 25 , 26 , 27 , 28 , 29 , and dates back at least to the classical study of Aguilar and Stiles 3 , who found incremental rod saturation with the psychophysical two-color incremental threshold test.

Green 4 measured ERG responses from rat retina and found rod saturation with a two-color incremental threshold test similar to the one used by Aguilar and Stiles. However, an important aspect of both studies, often overlooked, is that they report loss of contrast sensitivity rather than full rod saturation. Our own work highlights the importance of this distinction.

We too see loss of rod contrast sensitivity when stepping to bright backgrounds, but this does not correspond to full rod saturation. Rather, rod responses were always apparent if stimuli of sufficient contrast were applied. When interpreting the results of visual experiments carried out at high backgrounds, especially when considering whether rods contribute to visual processing, several factors should be considered.

First, in addition to stimulus contrast, is the sensitivity of the recording method: stimulus-evoked rod responses Fig.

Second, thanks to the recovery of rod contrast sensitivity over time at photopic backgrounds, rod responses to moderate contrasts may re-emerge even after lacking for many minutes. Third, even more importantly, the recovery rate in contrast sensitivity is itself positively correlated with irradiance. High backgrounds are commonly used as an experimental strategy to study cone vision.

Our experiments confirm that stepping to high backgrounds can indeed produce retinal responses that are strongly or even exclusively cone-driven. However, they further reveal that the duration of this exclusively cone-driven operational regime is shortest at higher light levels. Thus, care needs to be taken to limit exposure to photopic light levels if the aim is to exclude rod responses Indeed, it is not safe to assume that rods stay silent at any background light intensity, and further irradiance increase, once in the photopic range, can actually be detrimental to the objective of isolating cone vision.

Since wild-type mice have very similar spectral sensitivity of rhodopsin and M-opsin, unwanted rod intrusion at high light levels can easily go unnoticed. The results of the computational modeling are instructive with respect to the potential mechanisms that allow rods to signal at even the brightest light levels.

A previously published model of rod phototransduction 20 predicted that rhodopsin isomerization was triggered by visual stimuli at all backgrounds red curve, Fig. Inclusion of irradiance-dependent translocation of transduction cascade components between inner and outer segment transducin, arrestin, and recoverin, Fig.

The net effect of the translocations is a reduced gain of the cascade, such that the stimulus-induced variation of activated rhodopsin was translated into a modulation of photocurrent. According to the model, then, these translocations are a prerequisite for rod responses at high backgrounds. They do not explain all aspects of such activity, however, as even when we modeled these translocations as instantaneous events, the model recreated the slow build-up of rod responses under extended exposure to high irradiance and the positive correlation between irradiance and the rate of this recovery Fig.

In the model, the response recovery rate coincided with the rate of rhodopsin bleaching, which naturally occurs more rapidly at higher irradiance. Our experiments confirmed this hypothesis Fig. Note, however, that bleaching can also have the opposite effect and contribute to saturation, because bleached rhodopsin i.

The description of bleaching adaptation raises an interesting further question. Why did our ex vivo preparation not stop responding to light simply because all its visual pigment had become bleached? We were able to record responses from cone-deficient retinas isolated from the retinal pigment epithelium RPE for many hours at high light levels that should have fully bleached available rhodopsin in minutes or even seconds of exposure.

How can rods then retain responsiveness? Kaylor et al. This process would naturally be strongest at high light levels, resulting in the maintenance of a steady-state pool of regenerated rhodopsin. Alternatively, our ex vivo retina preparation may retain sufficient RPE we occasionally see small specks of pigment epithelium remaining on the isolated retina to recycle chromophore. The mathematical model allows exploration of rod behavior at high irradiance with variations in chromophore recycling.

Higher rates of chromophore recycling model parameter k Rrecyc abolish responses at high irradiance Supplementary Fig. In essence, only intermediate values of k Rrecyc predicted photopic rod responses Supplementary Fig. Could changes in other key model parameters support light responses at high backgrounds, despite a lower chromophore regeneration rate likely encountered in ex vivo experimental conditions?

The behavior at high irradiance was also dependent upon both the total number of rhodopsin molecules parameter Rhod total , columns in Supplementary Fig.

Lower values for either parameter supported photopic responses, allowing the model to match responses recorded ex vivo at lower values for k Rrecyc. Therefore, the sensitivity of rod responses at high irradiance could depend on naturally occurring variations in the properties of the phototransduction cascade and its regulation e. Such variations might exist between species, between individuals of the same species, between different rods in the same retina, or even within the same rod across the circadian cycle.

This observation might further explain discrepancies in the literature concerning the rod saturation threshold Introduction. The updated computational model presented here provides a hypothetical biochemical explanation of our experimental observations. All of our adjustments to the original Invergo et al. Thus, we did not fit the model to our data, but found that implementing already well-established adaptational processes resulted in a surprisingly good qualitative match between the model behavior and our experimental observations.

Explaining rod responses at high light intensities does therefore not require any additional mechanisms beyond what is already known about rod behavior.

Nevertheless, additional consequences of bright light exposure may well exist. It was recently shown 38 that reduced sensitivity of the phototransduction cascade after strong bleaching is partly compensated by properties of the rod inner segment to achieve more robust voltage responses.

Furthermore, the supramolecular organization of rhodopsin and transducin into clusters on the disc membrane has been proposed to be important for fast and reliable light responses 39 , 40 , In the context of bright backgrounds, with reduced concentration of both rhodopsin due to bleaching and transducin due to translocation , regulation of such scaffolding might contribute to controlling the gain of photoresponses.

Do these conclusions drawn for mice translate to other species, including humans? The most direct proof for rod-mediated vision under photopic conditions would be obtained by psychophysical tests with rod monochromats. This is difficult due to the common photophobia of such individuals when exposed to bright light However, several studies showed that rods impact aspects of color perception at scotopic and mesopic light levels in human trichromats 43 , 44 , 45 , dichromats 46 , and blue cone monochromats 47 reviewed by Zele and Cao This influence of rod signaling on color vision could be used to test for rod contribution to visual perception also under bright light conditions.

Further, insights into rod-mediated vision under bright light can also be gained form a subset of rod-monochromatic individuals, which are not blinded in such an environment It is interesting that the rod system in these individuals is apparently less prone to saturation. According to our model, this could be explained by differences in rod biochemistry in these particular patients resulting in different gain control.

Deeper insight into the underlying mechanisms of this effectively non-saturating phenotype might reveal new opportunities to treat rod monochromats by appropriately interfering with the properties of the rod cascade, potentially by reducing the gain.

While such a reduced gain would be counterproductive for low-light vision, our daily lives, with electrical lighting all around us, happen mostly beyond the scotopic range, so that such treatment could indeed have a net positive benefit.

We used several transgenic mouse lines in which cone responses are abolished due to mutations disrupting the cone phototransduction cascade. Biel for ex vivo experiments, the cone-specific alpha-subunit of the cyclic nucleotide gated channel is mutated, preventing voltage changes in cones upon light activation. All ex vivo experiments were performed with explanted retinas, with RPE removed.

In contrast to older animals 12—13 months , younger animals 5 months did not have complete response suppression after switching to photopic light levels, which indicates residual cone function. In other words, the complete response suppression to moderate contrast stimuli in experiments described here suggests that cones were indeed non-functional in those retinas and responses, when present, were rod-driven.

Experiments were performed during daylight circadian times experiment start in the morning or early afternoon. The eye cups were removed, put in Ringer solution in mM: NaCl, 2. Experiments were performed as described previously Briefly, the mounted retina was placed ganglion cell-side down in the recording chamber, and good electrode contact was achieved by negative pressure through the perforated MEA. Spike sorting and thereby assignment of spikes to individual units presumably ganglion cells was performed semi-manually with custom written software Matlab.

Data analysis was based on the spiking responses of individual units. We estimated the instantaneous firing rate of ganglion cells by convolving the spike train i.

Given the flow rate of our perfusion system, it took about 1. Ex vivo ERG recordings were performed as described previously utilizing the same electrode MEA system as described above Noisy electrodes were discarded and all remaining electrodes were averaged for the analysis of ex vivo ERG responses. Visual stimulation followed the same protocol described before Briefly, the retina was stimulated with full-field gray scale visual stimuli with a computer-controlled digital light processing DLP projector PG-FX-L, Sharp or K11, Acer and focused onto the photoreceptors through the condenser of the microscope Supplementary Fig.

We linearized the gamma function of the projector output. While changing the ND filters, we closed the shutter to prevent intermittent exposure to bright light. We usually started the experiments with ND8 i. Unless otherwise noted, we presented the same set of visual stimuli at each ND level during an experiment.

It truly reflects photoisomerizations only at low intensities; at high backgrounds, bleaching adaptation leads to a much lower effective rate of isomerizations. For a flash stimulus of intensity I stim , presented on a background of intensity I back , the definitions are as follows:.

The firing rate curves on the right in Figs. Whether or not a ganglion cell responded to a block five repetitions of contrast steps was determined manually. For each unit and each stimulus block, we manually inspected spike raster plots and firing rates. The amplitude of the response used in Fig.

These amplitudes were normalized for each ganglion cell separately to its maximal response across the experiment. Averaging across experiments. In the experiments depicted in Fig.

Supplementary Fig. Relative response strength. For the experiments with 9- cis retinal Fig. To this end, we took the average spike rate to the last 50 stimulus repetitions within a min segment as a template Fig.

Population analysis. For four retinas, we applied 9- cis retinal Ringer during the three last experimental segments, as shown in Fig. For the other two retinas, we only recorded under control conditions. For each of the four stimulus edges, data as shown in Fig.

Because this approach is based on template matching, we only included experimental segments in the population data during which ganglion cell spiking was robust, and in which there was a template to be matched e. Whether or not to include the responses of a ganglion cell during an experimental segment in the analysis was judged manually and independently for each of the four stimulus edges. Statistical analysis. We then split this sample in two, measuring the distance between their medians.

We repeated this 10 5 times. Coming from the same population, the distribution of these distances is symmetric around 0. The null hypothesis was that the experimentally observed distance between the medians of control and retinal conditions was from the same distribution, with the alternate being that median of retinal conditions is smaller than that of control.

We took as p -value to support the null hypothesis the fraction of samples out of 10 5 for which the distance between medians was at least as large as the experimentally observed value.

The p -value is color coded in Fig. Significance testing was performed by using flashes of three consecutive stimulus sets, i. Figure 1b shows the moving average for this analysis averaging three stimulus sets per data point, shifting by one stimulus set for the next data point; no averaging was done across light level transitions.

Response reliability. ERG responses at high light levels were usually very small, but nevertheless often clearly distinct from the voltage fluctuations of the background activity. A craniotomy was drilled above the coordinates for the dLGN B—2. Light stimuli were delivered to the eye contralateral to the recorded brain hemisphere.

For in vivo experiments presented in Fig. In all cases, spectral power densities for each LED were measured using a calibrated spectroradiometer Bentham Instruments Ltd. The background intensity at the brightest light levels used in vivo was 4.

For experiments in Fig. For contrast sensitivity experiments Fig. For experiments with fixed contrast Fig. Four multiunits were excluded because they stopped responding completely after the first light level switch. In one mouse, recordings could only be performed up to the highest light level of 10 6.

Firing rate has been calculated by convolving the spike train i. The background firing rates from 20 groups was then averaged and taken as the mean background firing rate for these 20 groups. We applied a Wilcoxon rank sum test 1-sided to test for significant differences between the 20 background and the 20 response values, i. These significance tests were performed on a running average with shifts of two groups for each data point. A rod cell has an elongated structure with the outer segment specialized for photoreception.

See diagram on right It is this segment that contains the many discs which are membrane enclosed sacks densely packed with photoreceptor molecules. The photoreceptive molecule is rhodopsin which consists of the protein opsin linked to cis retinal a prosthetic group. In the fovea, there are NO rods The cones are also packed closer together here in the fovea than in the rest of the retina. Also, blood vessels and nerve fibers go around the fovea so light has a direct path to the photoreceptors.

Here is an easy way to demonstrate the sensitivity of your foveal vision. Stare at the "g" in the word "light" in middle of the following sentence:. The "g" in "light" will be clear, but words and letters on either side of the "g" will not be clear. One part of the retina does NOT contain any photoreceptors.

This is our "blind spot. It is in this region that the optic nerves come together and exit the eye on their way to the brain. Hold the image or place your head from the computer monitor about 20 inches away. With your right eye, look at the dot. Slowly bring the image or move your head closer while looking at the dot.

Reverse the process. Move the image slowly closer to you and the dot should disappear. For this image, close your right eye. With your left eye, look at the red circle. Slowly move your head closer to the image.