The photovoltage of rods and cones in the dark-adapted mouse retina

Abstract

Research on photoreceptors has led to important insights into how light signals are detected and processed in the outer retina. Most information about photoreceptor function, however, comes from lower vertebrates. The large majority of mammalian studies are based on suction pipette recordings of outer segment currents, a technique that doesn't allow examination of phenomena occurring downstream of phototransduction. Only a small number of whole-cell recordings have been made, mainly in the macaque. Due to the growing importance of the mouse in vision research, we have optimized a retinal slice preparation that allows the reliable collection of perforated-patch recordings from light responding rods and cones. Unexpectedly, the frequency of cone recordings was much higher than their numeric proportion of ∼3%. This allowed us to obtain direct functional evidence suggestive of rod–cone coupling in the mouse. Moreover, rods had considerably larger single photon responses than previously published for mammals (3.44 mV, SD 1.37, n = 19 at 24°C; 2.46 mV, SD 1.08, n = 10 at 36°C), and a relatively high signal/noise ratio (6.4, SD 1.8 at 24°C; 6.8, SD 2.8 at 36°C). Both findings imply a more favourable transmission at the rod–rod bipolar cell synapse. Accordingly, relatively few photoisomerizations were sufficient to elicit a half-maximal response (6.7, SD 2.7, n = 5 at 24°C; 10.6, SD 1.7, n = 3 at 36°C), leading to a narrow linear response range. Our study demonstrates new features of mammalian photoreceptors and opens the way for further investigations into photoreceptor function using retinas from mutant mouse models.

Figure 1. Rods and cones in the mouse retina are accessible to patch-clamp recording

Aa, Lucifer Yellow (LY) stain of a rod made by attaining the whole cell configuration at the end of a perforated patch-clamp recording (spherule labelled with a star; patch pipette still in situ). Ab, rod membrane potential responses evoked by flashes of light of strength 0.79, 2.8, 8.2, 26, 79, 239 photons μm⁻². Responses are averages of 3–8 sweeps. Ac, another rod's response to a long step of light of 9060 photons μm⁻² s⁻¹. Inset shows an enlarged section around step onset. Ba, LY stain of a cone (pedicle labelled with a star; pipette withdrawn before image acquisition). Bb, cone membrane potential responses evoked by flashes of light of strength 7.3, 16, 32, 65, 124, 240, 431, 775, 1550, 3100, 7000, 13500 photons μm⁻². Inset shows an enlarged section around the peak. The star labels the late cone plateau. Responses are averages of 3–7 sweeps. Bc, membrane potential response of another cone to a long step of light of 4 × 10⁴ photons μm⁻² s⁻¹. Baselines in Ab, Bb, Ac and Bc were aligned to each other (max shift 3 mV), and all records were ‘box car’ filtered with a running window of 20 ms. Light stimuli were delivered in darkness using green light (514 nm). Data obtained at 24°C.

Figure 2. Dark membrane potentials and flash response kinetics are perturbed by recordings

A, graph showing recording duration (i.e. the interval between seal formation and its spontaneous loss) versus V_dark in rods (rectangles) and cones (triangles), at either 24°C (white) or 36°C (black). Rod V_dark values were significantly more positive than those of cones at the same recording temperature. At both temperatures, rod V_dark values were spread in a wide range. Recording duration and V_dark were inversely correlated among rods at 24°C (see Results), a striking trend that was also observed when comparing rods and cones at both temperatures (panel A). These data suggest that V_dark was depolarized from its unperturbed state, particularly for rods (see Discussion). Cone recordings lasted more than those of rods at 24°C (P = 0.003, MWW test), while rods at 24°C were more stable than rods at 36°C (P = 0.025, MWW test). In each cell the reported value of V_dark was the best observed during the recording session, which typically occurred within a few minutes after the seal was formed. B, traces above showing the progressive slowing in flash response kinetics expressed by a rod while it is being recorded (each trace is the average of 40 responses to 1.2 photons μm⁻² flashes). Note that the increase in time-to-peak (TTP) did not affect response amplitude. The graph below confirms this important conclusion by collecting observations from 26 rods and 5 cones (both at 24°C and 36°C). Each point represents a measure of response amplitude and TTP at a particular time from the beginning of recording, normalized to the values observed in the fastest flash response of that particular photoreceptor. A linear regression through the data shows that increases in TTP are only weakly correlated to changes in amplitude (log scale on the abscissa).

Figure 3. Rod and cone light sensitivities to green light (514 nm) provide evidence suggestive of rod–cone coupling

A, graph showing peak hyperpolarization as a function of flash strength, for three rods (white squares and triangles, grey circles) and four cones (white circles and squares, black circles and diamonds). Data are mean ± SEM. B, normalized peak response as a function of flash strength in several rods and cones. Hill functions are shown, which provide an excellent fit to the data points (omitted for clarity; thick lines indicate the range of flash strength covered by the underlying data). C, The late plateau of the bright flash response (b) saturates at the same strength in a cone as in rods while, in the same cone, the peak (a) saturates >2 log units above. This suggests that the late cone plateau originates in rods and is transmitted via gap junctions. Data are means ± SEM. Rods recorded in loose seal mode to avoid a rundown of response kinetics (n = 3). D, grey lines representing the ratio of the percentage response of each cone to that of a reference rod with a light sensitivity in the lower rod range (dashed line in panel B). The ratio minima (black dots) give an upper bound for any functional rod–cone electrical coupling at dim flash strengths (1–20 photons μm⁻²). All data obtained at 24°C.

Figure 4. Single photon responses (SPRs) in mouse rods are large

A, dim flashes of three strengths (0.10_b, 0.26_c, 0.44_d photons μm⁻²) were delivered at 5 s intervals on a dark background as shown by the top trace. Below is a continuous stretch of membrane potential recorded in a rod at 24°C (280 s, shown in 8 consecutive segments). Note the occurrence of ‘failures’ and the high variability in response amplitude. Increasing flash strength reduced the frequency of response failures: this is shown in the histograms below, one for each flash strength, which give the distribution of response sizes quantified as the area under the voltage record measured in a 300 ms segment centred on response peak. The graph at top-right plots the mean number of photoisomerizations per flash m (estimated from the response failure probability by assuming Poisson statistics) as a function of dim flash strength. A linear relation is precisely what one expects from photon absorption statistics. Data are from 301_a (zero strength, non-overlapping recording segments), 44_b, 87_c, 172_d flashes, respectively. Ba and Bb, dim flash responses of several rods chosen to represent the range of observed SPR amplitudes (24°C top row; 36°C bottom row). Individual records are shown with a thick line in light grey, and darken where they overlap with other records. For each rod flash strength (i), mean photoisomerizations (m), and SPR amplitude (a) are given. The two values of a report estimates obtained using the μ/m or σ²/μ methods (see Results). The duration of the shown records was, from left to right 3, 3, 2, 3, 2 s (Ba) and 3, 2, 3, 2, 2 (Bb). All records were ‘box car’ filtered with a running window of 20 ms.

Figure 5. Rod dim flash responses can be markedly sublinear

Aa, normalized flash response amplitude versus mean number of photoisomerizations (Rh*) in 5 rods at 24°C. Data points for three rods are shown (a: open squares, c: open diamonds, e: open circles) with Hill fits (thick lines). Two more rods (b, d) are represented only with their fits (thin lines) for clarity. A few Rh* are sufficient to recruit a significant fraction of the rods’ available voltage swing. The same 5 rods are reproduced in the inset graph focusing on small values of m, with amplitudes normalized to the slopes of their respective fits at m = 0 (Michaelis–Menten [M–M] fits are shown here (see Results); the units a₀ in the ordinate represent the single photon response amplitude that one would determine by delivering flashes of strength approaching zero. The fits predict that rods diverge from linearity already with two Rh*. Ab, normalized flash response amplitude versus mean number of Rh* in three rods at 36°C (fitting procedures are the same as in Aa). Near body temperature the rod dynamic range is somewhat shifted to the right (see Results), leading to a decrease in the expected degree of sub-linearity (inset). B, the prediction of a significant sub-linear summation of multiple Rh* was accurately tested in a rod at 24°C (b in panel Aa), by delivering dim flashes at two strengths differing by a factor of four. Mean responses (black traces) and one SEM range (grey areas) are shown for the weak (x) and stronger (y) flash. After normalizing over flash strength (expressed in terms of m) the weaker flash was found to give a significantly larger response. The graph on the right compares the same responses with their amplitudes normalized to those expected in the linear approximation (units of ma₀). The curve is the M–M fit to the full set of data points available for this rod, including bright flashes. Error bars are SEM.