Journal of Neuroscience Methods, 24 (1988) 65-72 65 Elsevier

NSM 00810

Measurement of DC and AC spectral sensitivities of retinal horizontal cells by "voltage clamp by light"

Masahiro Yamada and Syozo Yasui Applied Optics Section, Electrotechnical Laboratory, Tsukuba Science City, Ibaraki (Japan) and Department of Biological Regulation,

National Institute for Basic Biology, Okazaki (Japan)

(Received 1 July 1987) (Revised 23 October 1987)

(Accepted 7 December 1987)

Key words: Retina; Horizontal cell; Spectral sensitivity; Color vision; Voltage clamp

The method of "voltage clamp by light" was applied to measure spectral sensitivities of second-order visual neurons, namely L- and R/G-type horizontal cells in the carp retina. The present equipment employs (i) a ceramic photomodulator to facilitate a fast servomechanical control of retinal illuminance, (ii) an electronic circuit to compensate for the synaptic transmission delay, and (iii) a manual selection switch for the system to operate on negative feedback for either depolarizing or hyperpolarizing responses to light. These features allowed us to determine quickly and simultaneously both DC and AC spectral sensitivities, although the AC case was examined only at 1 Hz in this report. In L-type cells, the AC spectral sensitivity was similar in shape to the DC result. These sensitivity curves differed from microphotospectrometric absorption of red-sensitive cones: in L-type cells at both ends of the visible spectrum and in R / G units with deep-red light.

Introduction

Amongst the various kinds of interneurons in the teleost retina are horizontal cells that modify trichromatic signals from cone photoreceptor cells and which are thought to subserve an early stage of color information processing. Thus, while the luminosity-type (L-type) horizontal cells hyper- polarize in response to all visible light, the chro- maticity-type (C-type) horizontal cells, such as R / G units, depolarize in response to certain wave- lengths (e.g. Kaneko, 1970; Mitarai et al., 1974), in contrast to the well-known fact that under normal circumstances vertebrate photoreceptors of every kind can only hyperpolarize.

Here, spectral sensitivities of these cone-driven

Correspondence: S. Yasui, Department of Biological Regu- lation, National Institute for Basic Biology, Okazaki, 444, Japan.

second-order neurons were measured in the light- adapted carp retina by using a voltage-clamp scheme which makes use of the natural stimulus, namely light. This idea of "voltage clamp by light" for controlling the membrane potential of non-spiking visual cells has been described in the literature (Padmos and Norren, 1972; Fran- ceschini, 1979; Smakman and Pijpker, 1983). In the previous applications, DC-motor driven neu- tral-density wedges were employed for the illumi- nance control to measure sensitivities of in- vertebrate photoreceptor cells to the wavelength or the E-vector of incident light. The conventional protocol of spectral sensitivity measurement is time-consuming; one records a number of flash responses at each test wavelength and then in- terpolates the color-dependent threshold intensity of light that produces a criterion membrane volt- age usually set at the response peak. The voltage clamp by light automatizes all this by presetting the criterion voltage as a command signal. This

0165-0270/88/$03.50 © 1988 Elsevier Science Publishers B.V. (Biomedical Division)

66

greatly increases the speed of measurement, which is critical especially for vertebrate retinal neurons that deteriorate easily during intracellular record- ing. We have employed a ceramic photomodulator rather than a neutral density wedge to generate light of controlled intensity. This makes our equipment especially fast in its clamp action. In addition, an electronic circuit is provided to offset approximately the synaptic delay; without this compensation the clamp performance is severely hampered when second-order cells are to be ex- amined as in the present application. The voltage clamp system with these features enabled us to obtain a unique collection of data, including DC and AC spectral sensitivities simultaneously re- corded from L-type horizontal cells. From these results we have made certain observations not previously reported for horizontal cells. A part of this work has appeared in abstract form (Yasui and Yamada, 1985).

Materials and Methods

Physiological recording: conventional part The retina isolated from the decapitated carp,

Cyprinus carpio, was mounted in a chamber with the photoreceptor side up. The retina was kept light-adapted and was continuously superfused with a Ringer's solution composed of (in mM) NaC1 102, NaI-ICO 3 28, KC1 2.6, CaCI: 1, MgCI: 1 and glucose 5, maintained a pH 7.6 and aerated with a gas mixture of 95% O z and 5% CO a. Glass micropipettes filled with 3 M potassium chloride and having a tip resistance of about 100-150 M~2 were used to record the intracellular potential. Horizontal ceils were identified on the basis of location (90-150/~m deep from the receptor side), response waveform and area effect (spatial sum- marion property). Their subtypes were determined according to the spectral response profile.

Voltage clamp by light The apparatus we have developed is illustrated

in Fig. 1. The error signal that represents the difference between the command and actual mem- brane voltages drives a PLZT photomodulator

F--- ~ . . . . . . . $W ., V c Commend iController ~ ~- .---<"x -

" 0river

sL !

S H ~ PLZT! ! ! ~ - I F NO 0 - - ~ /F ~r'-:m~fi:':- ]

Record Fig. 1. A schematic drawing illustrating the apparatus for voltage clamp by light, which employs a PLZT photomodula- tor for fast clamp action and also includes a sign-inverting switch (SW) to deal with either depolarizing or hyperpolarizing membrane potentials V m in response to light. See Fig. 2 for the controller unit. The opdcal equipment is standard except for the use of PLZT, viz., SH, electromagnetic shutter; SL, Slit; IF, interference filter (half-bandwidth of about 10 nm); ND, neu-

tral density filter; D, diaphragm; PD, photodiode.

(BLA-20, Motorola), in such a way as to minimize the difference by supplying the suitable retinal irradiance. The PLZT device is a ceramic con- denser sandwiched by polarizer and analyzer. Its optical transmittance can be controlled over a range of 3 log unit by changing the driving DC voltage: Another notable feature of this device is its high-speed action; the modulation frequency can be up to 10 kHz.

An electronic controller operates upon the error signal and produces the signal to drive the PLZT unit. The controller is implemented by using sum- ming amplifiers and integrators as shown in Fig, 2, and it has a transfer function expressed as K(1 + Tts)(1 + T2s)/[s(1 + 0.01 s)] where s de- notes the Laplace transform variable. K, T~ and T2 are adjustable manually through 3 potentiome- ters; typically, K = 23, T~ = 0.18 and T 2 = 0.34 in the present experiment. The factors 1 + T 1 s and 1 + T 2 s are phase-advancing, and compensate for the relatively slow kinetics of the cellular response to light and also serve as an approximate way to offset the synaptic delay; see Yamada et al. (1985) for the response latency of carp horizontal cells. Both hyperpolarizing and depolarizing responses can be handled by manipulating a sign-inverting manual switch in such a way as to form a nega-

Polenliometer Amphlier

lmegrator

VolIo~ Error Signol q V e

~T~/q >~. To PLZT ~ ' ~ l ~ ' ~ b ~ . . . . . . ~ Driver

Fig. 2. A block diagram of the electronic controller which operates on the voltage error signal V~ to produce the signal for driving the PLZT photmodulator. Digits such as 1, 10, 100 attached to the amplifier and integrator symbols indicate the amplication factor. One amplifier with " -1" mark is a sign inverter. The 3 potentiometers are manually adjustable and

their typical attenuation factors are indicated in the text.

tive-feedback control mechanism. The membrane polarization of carp horizontal cells can be made to follow command signals with frequencies f rom 0 (DC) up to 3 Hz.

Optical system A photodiode (VDT-500, United Detector) was

used to record the intensity of the light needed to control the membrane potential in the way pre- scribed by the command signal (Fig. 1). Except for the presence of the PLZT photomodulator , the photostimulator with a quartz halogen lamp was conventional (Fig. 1). The residual infrared light escaping from the interference filter was eliminated by a suitable infrared-cut filter (HA-30, Hoya Glass Works). At its maximum transmittance, the PLZT unit attenuated the light intensity by a factor of 0.8 log unit. Under this condition of PLZT and with a 0 log-unit neutral density filter, the measured values of retinal irradiance were 1.5, 1.4, 1.3, 0.85, 1.1, 0.99, 1.1, 1.4, 1.4, 1.5, 1.1 and 0 . 7 6 / t W / c m 2, corresponding in that order to 419, 437, 462, 481, 502, 523, 533, 580, 617, 662, 701 and 739 nm of the test wavelength.

67

Results

The stimulating light encompassed the whole retina. The wavelength was varied from 419 nm up to either 701 or 739 nm with appropriate steps and the spectral scan was repeated as often as possible. Unless the result was adequately re- producible, the data were discarded. This excluded unstable recordings such as those which suffered from a drift of the membrane potential in the dark. At each wavelength tested, the membrane potential was clamped at the criterion value (several millivolts from the dark level of - 5 0 - - 2 0 mV) at least for 3 s to determine the DC spectral sensitivity. An AC spectral sensitivity was measured by adding a 1 Hz triangular signal to the DC command. The AC command was made triangular rather than sinusoidal because the fidel- ity of voltage control could be inspected more readily.

L-type horizontal cells A total of five L-type horizontal cells were

examined for the DC and 1 H z - A C spectral sensi- tivities. Fig. 3 shows a set of raw records. The voltage clamp worked satisfactorily throughout either direction of the spectral sweep, although a brief instability was often incurred when a new interference filter was manually set to change the wavelength (Fig. 3, inset). According to the photo- diode record, the mean intensity of light was lowest at about 615 nm, indicating that the cell was maximally sensitive at about this wavelength. While this refers to the DC spectral sensitivity, the peak-to-peak amplitude of AC modulated inten- sity as a function of wavelength is a measure of AC spectral sensitivity. At 1 Hz this also appears highest around 615 nm.

Although the original trace of light intensity provides a good idea of the spectral sensitivity, some correction was needed to improve the accu- racy of measurement. This was because the photo- diode used did not have a perfectly equal sensitiv- ity over the spectral range examined. Thus, correc- tive calculations were made on the photodiode records; the actual retinal irradiance measured in /~W/cm 2 was converted, at each wavelength, to the flux of photons. For each scan, the wave-

68

L-type horizontal cell

V photodiode

:t i

/1 [ / p'+"+~ ++ 'm[" .... oi i ~ ; i

4 1 9 4 6 2 5 0 2 5 3 3 617 701

'I? Jt+~l ii,j+~,+li~ tI'll %,, ttjttl

I i l

i i , 701 617 533 502 4 6 2 419 4 6 2 502

4 3 7 481 523 5 8 0 6 6 2 7 3 9 6 6 2 5 8 0 5 2 3 481 4 3 7 4 3 7 481

, I, i

photodiode

I IL/, ~'~%P'~' L "' ' ~ ~ ' ! i ', i

! V"~%%'V',/ i

580 533 523 n m

mV

-28 L Vrn dark level

mV

-2a~ ¥c : AC(1 Hz- t r i angu la r l -modu la ted

I t

60 s e e

Vm

Vc

I !

10 see

Fig. 3. A set of raw records obtained through the voltage clamp by light for the measurement of DC and AC spectral sensitivities of a carp L-type horizontal cell. The actual membrane potential (I'm; middle trace) was a b o u t - 32 mV in the dark. The command voltage sj~,n~! (Vc, bottom trace) was a triangular wave (frequency: 1 Hz, peak-to-peak amplitude: 2 mV) added to a constant 6 mV hyperpolarization from - 32 inV. The intensity of light needed to maintain the voltage d a m p was recorded as the output voltage of a photodiode while the wavelength was varied from 419 to 739 nm and also in the reversed direction (top trace). Each trace marked

by *-, is reproduced on the right with an expanded time scale.

length-dependent quantal flux was then nor- maliTed to its minimum that occurred at 615 nm. As usual, we express the sensitivity by reciprocal of this relative quantity. Fig. 4 shows one such final result compiled from an L-type horizontal cell. As is evident in this example, the DC and AC spectral sensitivity curves have turned out to be quite similar in shape, each yielding a monophasic profile that peaks around 615 nm. Since each spectral sensitivity curve is a normalized one, one might expect that the higher the sensitivity is, the smaller its variance would become. Actually, how- ever, both DC and AC sensitivity values tended to scatter more at the blue end of the visible spec- trum than at the other end where the average sensitivity dropped somewhat more. Thus, if we evaluate the normalized DC sensitivity in logarith-

mic unit (mean + standard deviation; 0 + 0 at 615 nm), we get - 1.01 + 2,02 at 437 nm based on 17 data points from 5 cells, and - 1.31 + 0.108 at 740 nm based on 8 data points from 4 cells.

R~ G-type horizontal cells The DC spectral sensitivity was measured in 4

R / G type horizontal cells, and the record of Fig. 5 is an example. The voltage clamp was hindered occasionally during the scan over 620-660 nm, because R / G cells reversed their response polarity somewhere in this region but the manual switch- ing by SW of Fig. 1 was made often too late or too early. In this example, the voltage clamp was decent only during those periods indicated by dashed segments, and o ~ a t i o n s and deviations occurred elsewhere due to the problem mentioned

L-type horizontal cell

7~

- 2 . r "

0

- 2

400 500 600 700 Wavelength (nm)

Fig. 4. DC and AC spectral sensitivity curves of an L-type horizontal cell (not the same cell unit as in Fig. 3). The spectral characteristics of the photodiode was taken into account to correct the photodiode output for obtaining the wavelength-de- pendent threshold flux of incident photons. The spectral scan was repeated 5 times in alphabetical order (A . . . . . E) in the direction of the arrow. Dashed curves are the absorption spectra of the carp's red- and green-sensitive cones measured

by MSP (Hhrosi, 1976).

above. By correcting the light-intensity record for the photodiode characteristics, the spectral sensi- tivity of this R / G unit was determined as plotted in Fig. 6. The DC sensitivity of the hyperpolariz- ing component was highest at 500-510 nm, whereas the peak sensitivity of the depolarizing response could not be located due to the lack of spectral resolution. At the transition to the blue end of spectrum, there was a rather abrupt fall of the sensitivity. No such change was observed in L-type cells.

Discussion

Hb.rosi (1976) employed microspectrophotome- try (MSP) to measure the absorption spectra of

69

red- and green-sensitive cones in the goldfish which is closely related to the carp (Fig. 4, broken curves). At the blue spectral region, the carp L-type hori- zontal cells were found to be less sensitive, by roughly 0.5 log unit, than what one might expect from the absorption spectrum of red-sensitive cones (Fig. 4). A similar difference from the MSP result has been found in the carp red-sensitive cones themselves, for which several possible rea- sons have been suggested (Kaneko and Tachibana, 1985). Another departure from Hhrosi's data oc- curred at the other end of spectrum; both DC and AC sensitivities at 760 nm were somewhat lower in comparison with the absorption of red-sensitive cones (Fig. 4). This difference was observed in R / G units as well (Fig. 6). This was unlikely due to infrared contamination because we took pre- cautions against this kind of artifact (see Materials and Methods). As a matter of fact, a similar difference from MSP data in the deep-red region has been reported to exist at the level of ganglion cells (Spekreijse et al., 1981; Van Dijk and Spekreijse, 1984). All this can be understood if the difference in question already occurs at red-sensi- tive cones.

The similarity in shape between DC and AC spectral sensitivity curves (Fig. 4) deserves ex- planation. Thus, we consider the V m - l o g I rela- tion which describes the amplitude of voltage re- sponse, V m, as a function of log I, the relative light intensity in log unit. This curve is sigmoidal as is well known (Naka and Rushton, 1966a). The D C - A C similarity can be understood if it is as- sumed that the sigmoidal curve undergoes a paral- lel shift along the abscissa in accordance with changes in the wavelength (Fig. 7). This parallel shift is implied by Naka and Rushton's (1966b) principle of univariance. Thus, the DC-AC simi- larity is consistent with the notion that L-type horizontal cells receive inputs predominantly from red-sensitive cones.

Blue- a n d / o r green-sensitive cones play some role in producing the L-type cell response (Maksimova et al., 1966; Tauchi et al., 1984; Yasui and Yamada, 1987). Since it was possible to repeat the spectral scan several times due to the present fast method, we were able to observe that the DC and AC spectral sensitivities appear

70

V

• 3 "

.2

R/G-type horizontal cell

photodiode /

S W

419 462 502 533 617 701 701

m Y 437 481 523 580 662 739 662

- 3 0 1 VITI . d a r k l e v e l t , J~ ~

SW

ll:lllltl 617 617 533 502 462 419 n m

580 523 481 437

m V

- 3 0 ; _ , . . . . . . . . . . . . .

- 5 0 :=- ~ . . . . . . . . . . . ~ . . . . . . . . , . . . . .

! I

60 sec

Fig. 5. A record obtained by means of voltage clamp by light for the measurement of DC spectral sensitivity of an R/G horizontal cell. The criterion response was 8 mV from the dark level of -42 mV in either hyperpolarizing or depolarizing direction as appropriate. The response reversed its polarity between 662 and 701 nm where the manual switch (SW) of Fig. 1 needed the proper reset. The switching was made at the time marked by SW during each way of this round-trip spectral sweep. The record is sound only for intervals indicated by dashed segments beneath the middle trace. The voltage clamp for the last portion of scan was preceded by a

complete failure prior to SW as well as by a period of bad oscillations.

vulnerable to greater fluctuation in the blue region than at the other end of the spectrum (Fig. 4). This is of interest since there is no evidence to indicate that carp's red-sensitive cones behave less erratically than do the other two chromatic classes (see Tomita et al.'s 1967 record). The spectral dependence of data scatter may bear relevance to the strong possibility which we are aware of lately; short-wavelength inputs and red signals are per- haps mediated differently for their synaptic trans- mission to second-order retinal cells (Marc and

Lain, 1981; Mangel et al., 1985; Yasui and Yamada, 1987).

One probable role of horizontal cells with their large receptive field is to evaluate the average light intensity of visual scenes (Naka and Rushton, 1966a; Kaneko, 1970; Yasui and Ohtsuka, 1986). In the present application of voltage clamp by light, the stimulus was continuously present rather than as flashes, and the color changed from time to time. The DC spectral sensitivity measured here tells, therefore, how the retinal illuminance should

71

._>

c

o~ o

..3

-1

i

- 2 ,

400

R/G- type horizontal cell

.:-:::-~ ...... / ........ " ..... - ~ . sw ~ . . . . . . . I . . . . . . ~, , :.~,,

A "', I "',

- -~o 6~o Wavelength (nrn)

700

Fig. 6. The DC spectral sensitivity of the R / G cell of Fig. 5. SW indicates where the response reversed its polarity; hy- perpolarization for shorter and depolarization for longer wave -

lengths.

Vm

lime J L'I -I~ L'zl 'L'3' g ~0, I

Fig. 7. A schematic diagram to explain that the DC and AC spectral sensitivity curves are similar in the L-type horizontal cell (Fig. 4). The voltage clamp by light causes V m to change in a triangular-wave fashion about a mean value V0 with a peak-to-peak amplitude of AV. The correspondingly changing monochromatic light intensity is constructed by assuming a sigmoidal curve which relates V to the intensity I in logarith- mic scale. If synaptic inputs to the L-type unit are only from a single class of photoreceptors, then the curve would not alter its shape but should shift in the log 1 direction when the wavelength is varied. Thus, if the DC sensitivity at the wave- length of A is denoted by L in logarithmic unit, then the intensity has a mean of 10 L and a peak-to-peak amplitude of 10 L+'~L/2 - 1 0 L-aLl2. It follows that the DC and AC sensi- tivities are equal except for a constant factor of log (10 zaL/2 - 10-aL/2) . This analysis is legitimate only insofar as the frequency-sensitive kinetics are negligible. The horizontal cell has a bandwidth wide enough in this regard (Yamada et al.,

1985).

be adjusted as a function of wavelength if it is desired that horizontal cells keep producing a constant signal against a color-changing back- ground light. This is an inquiry somewhat analo- gous to the direct matching method in visual psychophysics (Le Grand, 1968).

The present method has been combined with the bridge-balance technique so that the mem- brane resistance of a non-spiking visual neuron can be measured under voltage clamp by light (Yasui and Yamada, 1985). In these extended applications, the fast clamp action of our appara- tus makes it necessary to prevent mutual inter- ference between voltage clamp and bridge bal- ance; the brief voltage deflection caused by each extrinsic current pulse for the resistance measure- ment is kept from being fed back to the illumi- nance control by using masking pulses. With this added feature, we have been investigating a wave- length-dependent difference in the postsynaptic mechanism of L-type horizontal cells (Yasui and Yamada, 1987) and the light adaptation process of crayfish retinula cells (Yasui and Yamada, 1985), by examining constant-voltage membrane resis- tance changes.

Acknowledgement

The authors are grateful to Professor A. Kaneko for his helpful comments.

This research was supported in part by the Grant in Aid for Scientific Research from the Ministry of Education, Science and Culture (Nos. 61890013 and 62490620 to S.Y.).

References

Franceschini, N. (1979) Voltage clamp by light: rapid measure- ment of the spectral and polarization sensitivities of recep- tor cells, Invest. Ophthalmol. Suppl. 18: p. 5 (abstr.).

Hhrosi, F.I. (1976) Spectral relation of cone pigments in gold- fish, J. Gen. Physiol., 68: 65-80.

Kaneko, A. (1970) Physiological and morphological identifica- tion of horizontal, bipolar and amacrine cells in goldfish retina, J. Physiol. (Lond.), 207: 623-633.

Kaneko, A. and Tachibana, M. (1985) Electrophysiological measurements of the spectral sensitivity of three types of cones in the carp retina, Jpn J. Physiol., 35: 355-365.

72

Le Grand, Y. (1972) Spectral luminosity. In D. Jameson and L.M. Hurvich (Eds.), Handbook of Sensory Physiology, Vol. VII/4, Springer, New York, pp. 413-433.

Maksimova, E.M., Maksimov, V.V. and Orlov, O. Yu. (1966) Amplified interaction between the signals from cell recep- tors and sources of S-potentials, Biofizika, 11: 472-477.

Mange1, S.C., Ariel, M. and Dowling, J.E. (1985) Effects of acidic amino acid antagonists upon the spectral properties of carp horizontal cells: Circuitry of the outer retina, J. Neurosci., 5: 2839-2850.

Marc, 1LE. and Lain, D.M.K. (1981) Uptake of aspartic and glutamic acid by photorecqators in the goldfish retina, Proc. Natl. Acad. Sei. U.S.A., 78: 7185-7189.

Mitaral, G., Asano, T. and Miyake, Y. (1974) Identification of five types of S-potential and their corresponding generating sites in the horizontal cells of the carp retina, Jpn. J. Ophthalraol., 18: 161-176.

Naka, IC-I. and Rushton, W.A.H. (1966a) S-potential from colour units in the retina of fish (Cyprinidae), J. Physiol. (Lond.), 185: 536-555.

Naka, K.-I. and Rushton, W.A.H. (1966b) An attempt to analyse colour reception by electrophysiology, J. Physiol. (Lond.), 185: 556-586.

Padmos, P. and Norren, D.V. (1972) The vector voltmeter as a tool to measure electroretinogram spectral sensitivity and dark adaptation, Invest. Opbthalmol., 11: 783-788.

Smakman, J.G.J. and Pijpker, B.A. (1983) An analog-digital feedback system for measttdng photoreceptor properties with an equal response, J. Neurosci., Methods, 8: 365-373.

Spekreijse, H., Mooy, J.E.M. and Van Den Berg, J.J.T.P. (1981) Carp ganglion ceU action spectra, Vision Res., 21: 1601 - 1604.

Tauchi, M., Yang, X.-L. and Kaneko, A. (1984) Depolarizing responses of L-type external horizontal cells in the goldfish retina under intense chromatic background, Vision Res., 24: 867-870.

Tomita, T., Kaneko, A., Murakami, M. and Pautler, E.L. (1967) Spectral response curves of single cones in the carp, Vision Res., 7: 519-531.

Van Dijk, B.W. and Spekreijse, H. (1984) Color fundamentals deduced from carp ganglion cell responses, Vision Res., 24; 211-220.

Yamada, M., Shigematsu, Y. and Fuwa, M. (1985) Latency of horizontal cell response in the carp retina, Vision Res., 25: 767-774.

Yasui, S. and Yamada, M. (1985) Studies of retinal neurons using voltage damp by light, J. Physiol. Soc. Jpn., 47: p. 486 (abstr.).

Yasui, S. and Ohtsuka, T. (1986) Horizontal cell signal is smaller with texture-like non-uniform patterns than with uniform fields of the same space-average illuminance, Vi- sion Res., 26: 583-598.

Yasui, S. and Yamada, M. (1987) Conductance-decrease neu- rotransmitter action on H1 horizontal cells in the carp retina, Neurosci. Res., Suppl. 5: p. s186.