Materials

Adult individuals either of the wild type or of the orange-red variety of the medaka, Oryzias latipes, were used in this study, and leucophores within scales were examined for their motile responses. Motile iridophores on the scales of the dark sleeper goby, Odontobutis obscura obscura, were also tested for the possible application of this method, because, since they are dendritic, these cells show cellular motility similar to the leucophores (Iga and Matsuno, 1986; Fujii et al., 1991b). The medakas and the gobies were obtained from local dealers, and before all experiments, the fish were kept in freshwater aquariums at least for a few days for acclimatization.

With either species of fish, scales were plucked from the anterior, dorso-lateral part of the trunk, and immersed in a physiological saline solution for teleosts which had the following composition: (in mM): NaCl, 125.3; KCl, 2.7; CaCl₂, 1.8; MgCl₂, 1.8; d-(+)-glucose, 5.6; Tris-HCl buffer, 5.0 (pH 7.3). By means of a fine glass needle (diameter: ca. 0.4 mm) glued at both ends to the coverslip, a scale plucked from a medaka or goby was held with its bony scale side in contact with the coverslip. The method of irrigating the scale with physiological saline or with stimulant solutions in a small perfusion chamber was essentially identical to that described elsewhere (Fujii and Miyashita, 1975; Oshima and Fujii, 1984).

Optical system

In preliminary observations, an Olympus dark field epi-illumination microscope (Neopak NJTr, Olympus, Tokyo) was used. For later quantitative measurements of cellular responses, a Nikon industrial light microscope with similar performance (Optiphot XT-BD, with CF-BD Plan objective lenses; Nikon, Tokyo) was employed. Since they were installed with transmission and incident-light illumination systems, these microscopes enabled us to observe the chromatophores either by usual transmission optics or by incident light illumination, or by both. Thus, a comparison of the images of the cells by both optical systems could easily be made. We used a Nikon trinocular assembly of the T type in order to photoelectrically measure the motile response while observing the cell through binocular eye-pieces.

Photosensing

The differential photosensing part of the system developed in this study is schematically displayed in Fig. 1. Small-sized, high quality silicon photodiodes (S1226-18BK, Hamamatsu Photonics, Hamamatsu) were conveniently adopted as the photoelectric transducers, and were installed inside the photographic column of the trinocular assembly. On the rear focal plane of the objective lens, upon which the real image of the object was projected, a finely ground thin glass plate was placed to homogeneously diffuse the light-rays to the photosensors. The concentric separation of the light beams was also performed on this plane. For recording the light scattering from the cell body (S_c), a cut end of a plastic optical fiber (0.5 mm diameter) was set just in contact with the center of the ground surface of the glass plate. The optical fiber was sheathed within a thin stainless steel tube with an inner and outer diameter of 0.5 and 0.8 mm, respectively. The other end of the fiber was attached to a silicon photodiode. Using the 40 × objective lens, we were thus measuring S_c from a central circular area 12.5 μm in diameter, which approximated the size of the cell body of a single chromatophore (Fig. 1).

Fig. 1

Diagram showing the light sensing components of the system for quantitative recording of the response of a single light-scattering dendritic chromatophore. The apparatus was designed for independently measuring lightrays scattered from the central part and from the peripheral part of the cell independently. At the top of the figure, is shown the arrangement of the photodiodes within their housing (PSH) adapted on the photographic column (PCM) of the trinocular assembly of the microscope. At the bottom, the dimensions of the field-restricting diaphragm (FRD) with a projected image of a leucophore are displayed. IVC₁: current-to-voltage converter of channel 1, IVC₂: current-to-voltage converter of channel 2, OF: optical fiber, PD_C: photodiode for sensing light scattering from central part of the cell, PD_P: photodiode for sensing light scattering from peripheral part of the cell, RIP: real image plane, or rear focal plane of the objective lens, ST: stainless steel tubing.

In order to measure the scattering of light from the peripheral, dendritic zone of the leucophore (S_p), the light-rays hitting a circular area 5.6 mm in diameter on the plane of the real image were detected (Fig. 1). By design, the concentric central zone of 0.8 mm in diameter was occupied by the stainless steel tubing, and thus, the light scattering from the zone between two concentric circles with the diameters of 5.6 and 0.8 mm could be measured. Using the 40 × objective lens, the actual area to be measured on the skin was within the circular area of 140 μm in diameter, excluding the central circular zone 20 μm in diameter (Fig. 1). Usually, the domain occupied by a leucophore from Oryzias or by a motile iridophore from Odontobutis was within the 140 μm diameter. For an accurate measurement of light scattering from this area, three photodiodes were placed at equal distances from each other, namely, on the apexes of an equilateral triangle.

An eye-piece micrometer with an inscribed lattice pattern could be used to adjust the position of the cell body of the chromatophore with the mechanical stage adapted to the microscope stage. However, it proved to be more practical to position the cell body at the center of the domain by monitoring the output voltage of the channel to measure the light scattering from the cell body (cf. next section). Such a procedure could be done rather easily, because, when equilibrated in physiological saline, light-scattering organelles in the leucophore were normally aggregated in the cell body.

Electronic circuits

Simplified diagrams of the electronic division of the apparatus designed for processing output currents of the photosensors are shown in Fig. 2. Coupled with the photosensing system described above, these circuits enabled us to measure S_p and S_c differentially and continuously.

Fig. 2

Diagrams of the electronic processing parts employed. (A) Common initial steps consisting of two sets of serial combinations of current-to-voltage converters (IVC) and direct-current amplifiers of the non-inverting type (DCA). Channel 1 was for obtaining V_p, while the channel 2 was for V_c. Rf₁ - Rf₄: feedback resistors for IVC or DCA, SPD: silicon photodiode. (B) Adding circuit (ADD) for obtaining the sum of the output voltage of channel 1 (V_p) and the negative value of the output voltage of channel 2 (−V_c), namely, V_p − V_c. The negative value was obtained by means of a polarity-reversing circuit (PRC). R₁ −R₄: fixed resistors; in the present series of measurement, we adopted the same value (10 kΩ) for them. (C) Circuits for the 2nd and the 3rd treatments in the present series of measurement as desribed in the Results section. By means of adding circuits (ADD), appropriate voltage biases could be independently applied to V_p and V_c. The results were fed to the divider circuit (DIV) for obtaining their division (2nd treatment). In the 3rd treatment, the output voltage of the divider was further introduced to a “rooter” circuit (ROOT) to extract its square root. In the present trial, the same value (10 kΩ) was employed for R₁ −R₄ and also for R₅ − R₈. Rb₁, Rb₂: variable resistors giving appropriate voltage biases to adding circuits. For further details, see text.

Photoelectric current from either of the two channels, namely, from the photosensor for S_c, or from the assembly of photosensors for S_p, was converted separately into voltage change by a current-tovoltage converter using a high quality operational amplifier (IVC, Fig. 2A). In this design, we adopted a MAX421 (CPD-type package; Maxim Integrated Products, Sunnyvale, CA) as the appropriate operational amplifier. This was so designed that the input offset was automatically compensated with the built-in chopper-stabilized mechanism at 400 Hz, in addition to the desirable feature of very high input impedance (10¹² Ω). The efficiency of the conversion was modified by adjusting the value of the feedback resistance (Rf₁, Rf₃) of the operational amplifiers. The voltage outputs were further amplified by means of one-step direct-current amplifiers of the non-inverting type. Again, identical operational amplifiers to those employed in the preceding step were adopted (DCA, Fig. 2A), and again, the amplification was adjusted by changing the feedback resistance (Rf₂, Rf₄) of the operational amplifiers.

In either channel, the final voltage output (V_p or V_c) of the circuits can be expressed as the product of the intensity of light scattering (S), the efficiency of the current-to-voltage conversion, and the amplification factor of the direct-current amplifier. If the product of latter two are designated as A₁ or A₂, as the overall amplification factor, the final voltage output of each channel can be described as: V_p = A₁ × S_p, or V_c = A₂ × S_c.

When leucosomes disperse from the cell body into dendrites, V_p increases and V_c decreases, while the reverse changes occur when the leucosomes aggregate. If we desired, these changes can be separately registered on a chart recorder having more than two channels. During the actual recording of the cellular response, however, these changes are more conveniently integrated into a single channel and recorded.

A₁ or A₂ can be adjusted independently throughout a wide range by adjusting the efficiency of the current-to-voltage converter and/or the amplification factor of the direct current amplifier that follows the converter (Fig. 2A). Appropriate values for A₁ and A₂ were empirically determined beforehand on cells with average sizes and optical characteristics. Since the contribution of the central part to the overall change of light scattering was less than that of the peripheral zone, the range of the net voltage changes of channel 2 was adjusted to 30 − 50% of that for the peripheral zone.

The output voltages of the two channels, V_p and V_c, were then processed in three different ways. First, the difference between V_p and V_c was determined (Fig. 2B). V_c was first converted to −V_c using a polarity reversing circuit. V_p and −V_c were then summed by an adding circuit executed by the operational amplifier of the same type as that described above (MAX421), and recorded on a paper-chart recorder (EPR-10B, or EPR-221, Toa Electronics, Tokyo). Their ratio, V_p/V_c, was calculated continuously by an analog integrated circuit, and was recorded (Fig. 2C). For this purpose, an item especially designed for calculating the division of two input signals (AD535JH, Analog Devices, Tokyo) was adopted. Finally, the V_p/V_c ratio was fed to another integrated circuit (AD533JH, Analog Devices), from which the square root of the input voltage could be obtained. The result, was routed to the paper-chart recorder (Fig. 2C).

The polarity reversing device was put at the final stage of the electronic division of the present system, although not shown in Fig. 2, in order to reverse the polarity of the output signals to the paperchart recorder. Usually, it was set so as to record the progress of the dispersion of light-scattering organelles in the cell as the upward shift of the trace on the chart.

Since the scattering of light from part of a cell was weak, a considerably higher amplification of the signals was needed in either channel. Even though the utmost care in constructing and wiring the circuits had been taken, some noise was inevitably introduced in this system. In order to decrease this noise, an active, low-pass filter was placed between the output of the system and the recorder (not shown in Fig. 2). A second-order Butterworth type circuit was made, which was installed with the part for altering the cutting frequency stepwisely. A cutting frequency of 0.5 Hz was found to yield good results without noticeable deformation of the shape of the cellular response.

Computer graphics for analyses

In order to examine which of the three treatments mentioned above for integrating the two output signals would be more appropriate, computer graphics were ued, in addition to the actual empirical trials on the cells. A program for the simulation was developed using N₈₈-DISK BASIC(86), by which changes could be observed in the form of multiple-ordered algebraic functions by altering the values of any coefficient in the function. In fact, the virtual relationship of the input signals to an integrated output could easily be visualized by changing various parameters of the functions on the display of a personal computer.

Electrical stimulation

In some experiments, the chromatophores were stimulated electrically to induce the dispersion of organelles. The method was fundamentally identical to that employed in earlier studies using isolated fin pieces (Fujii and Novales, 1968, 1969; Fujii and Miyashita, 1975), but modified somewhat for application to the scales. An electronic stimulator (SEN-3201, Nihon Kohden, Tokyo) was used to stimulate nerves controlling the chromatophores. A stimulating electrode, a Pt wire 500 μm in diameter and insulated except at the cut end, was gently placed around the rostral margin of the piece of skin attached on an isolated scale. A leucophore or a motile iridophore located about 500 μm posterior to the electrode was measured for the cellular response. The polarity of the stimulating pulses was set to be negative in reference to this electrode. Another Pt wire dipped in the bottom of the perfusing chamber served as an indifferent electrode. Rectangular pulses of 1-msec duration and 10.0 V in strength were applied through the stimulating electrode at various frequencies. In order to reduce polarization at the surface of the electrodes, the pulses were biphasically modified by inserting a condenser (0.1 μF) between the output of the stimulator and the stimulating electrode. A storage oscilloscope (5111A, Tektronix, Beaverton, Oregon) was used for monitoring the stimulating pulses.

Chemical stimulation

Norepinephrine (hydrochloride salt; racemic modification; Sankyo, Tokyo) was used to induce the dispersion of the organelles either in leucophores or in motile iridophores. The concentration of the amine was given in terms of the physiologically active l-(−)-isomer. As the sympathetic neurotransmitter, norepinephrine has been shown to aggregate chromatosomes in common light-absorbing chromatophores via activation of α-adrenoceptors (Fujii, 1993; Fujii and Miyashita, 1975; Fujii and Oshima, 1986). In Oryzias leucophores, however, the dispersion of leucosomes in response to sympathetic stimuli is mediated through adrenoceptors of the β type (Obika, 1976; Iga et al., 1977). By contrast, the reflecting platelets in the motile iridophores of the dark sleeper have been shown to disperse through the mediation of α-adrenoceptors (Iga et al., 1987).

In some experiments, a K⁺-rich saline solution was employed to stimulate the chromatophore, because an elevated concentration of K⁺ ions is known to act as a sympathetic stimulus via the release of adrenergic transmitter from the postganglionic fibers (Fujii, 1959, 1993; Fujii and Oshima, 1986). In some preliminary observations, such as those shown in Fig. 1, a K⁺-rich saline in which Na⁺ ions in the primary saline were totally replaced with equimolar concentration of K⁺ ions was employed. In later studies made using the photoelectric method, a saline containing 50 mM K⁺ was exclusively used, in which the concentration of Na⁺ ions was also compensatorily decreased so that the final osmolarity was the same as that of the standard saline solution.

All the physiological measurements were carried out at a room temperature between 20 and 25°C.

The first treatment

The sequence of the motile responses of leucophores is shown in Fig. 3. When we used the microscopic system coupled with the electronic processing parts as shown in Fig. 2A, the following events took place: When leucosomes dispersed from the cell body into dendrites of a leucophore, the overall output voltage of channel 1 (V_p) increased, while that of channel 2 (V_c) decreased (Fig. 4). The reverse changes occurred when the leucosomes aggregated. If we assume the progress of the responses to be linear for the sake of simplicity, the output voltages of the two channels can be expressed as:

Fig. 4

Simulated lines by computer graphics for examining the appropriateness of the method described in the 1st treatment described in the Results. A: Postulated change of voltage output of the initial step for channel 1 (V_p). B: Postulated change of voltage output of the initial step for channel 2 (V_c). C: Line for y_l (=V_p − V_c). D: Line shifted parallel from y_l, in order to position the pen around 0.1 within the whole span of the recording chart. E: Line changed of its inclination from line D in order to include the whole range of the cellular response within the range of ca. 0.1 − 0.9 on the recording chart. Abscissa (x): State of leucophore, or the extent of leucosome-dispersing response, expressed as the value between 0 and 1.0, where 0 stands for the state in which leucosomes are completely aggregated in the perikaryon, while 1.0 indicates the state in which leucosomes are completely dispersed in the cytoplasm. Ordinate for lines A and B (y): Changes of V_p, V_c, V_p − V_c, and the modified functions (Region: 0 − 1.0). In this particular exhibition, we set V_p = 0.5x + 0.07, and V_c = −0.2x + 0.35. For further explanations, see text.

In the simulation displayed in Fig. 4, values for a, b, c and d in equations “1” and “2”, were set to be positive for easier analysis. As mentioned above, the actual change of the light-scattering from the central region is usually smaller than that from the peripheral part of the cell. Based on a rough estimation of the light-reflecting properties of the photomicrographic images of leucophores, we recognized that the contribution of the former was usually less than half that of the latter. In the simulation exhibited in Fig. 4, therefore, we adopted the value of 40% for this. Since both V_p and V_c are linear functions of x within the same region, the incline of V_c, namely c, was thus set to be 40% that of V_p, namely a. As practical values for a and c, we used 0.50 and 0.20, respectively. In consideration of the actual situation during the measurement of the response of leucophores, 0.07 and 0.35 were representatively given to b and d, respectively. The two lines denoted as A and B in Fig. 4 are those generated when these values are substituted in equations “1” and “2”.

When the response of a leucophore proceeds, either in the leucosome-dispersing or the aggregating direction, V_p and V_c show an inverse correlation with each other, as described above. Thus, the integrated response (y_l) may be simply expressed as the remainder of these two output voltages as:

As one popular application, an operational amplifier could be employed as the adding circuit, and such a treatment could easily be performed, as described in the Materials and Methods section (see also Fig. 2B). Line C in Fig. 4 indicates this relationship. By adjusting the position of the pen (Fig. 4, line D), and by changing the sensitivity of the recorder, the range of the leucophore response can be recorded within the desirable region on the recording chart (Fig. 4, line E). In this treatment, the recorded voltage change is a first degree equation. Without need of complicated decoding processes, therefore, further analyses can easily be performed as necessary.

As a typical recording using this type of analysis, Fig. 5 shows the responses of a leucophore from medaka to an increase in K⁺ concentration and then to norepinephrine. The dispersion of leucosomes in response to these stimuli are clearly observable.

Fig. 5

Typical recording of the response of a leucophore on a medaka scale, Oryzias latipes, to an elevated concentration (50 mM) of K⁺ ions in the perfusing medium. The recording was made employing the 1st treatment described in the Results. Finally, 2.5 μM norepinephrine solution was applied to bring about the full level of leucosome dispersion.

The second treatment

As a function of integrating the changes in the light scattering both from the central and the peripheral parts of the leucophore, we then considered their ratio:

With the same values for a-d, the lines A and B in Fig. 6 indicate the same functions as those shown in Fig. 4. The obtained relation took a concave form, shown as curve C in the figure. The equation indicates that y_ll diverges towards positive infinit when x approaches d/c. In practice, such a condition is unlikely, since it corresponds to the state in which the light scattering from the central part of the cell is completely lost. The function approaches 0 when x approaches −b/a, but, as described above, negative values for a or b were out of the definition.

Fig. 6

Simulation for examining appropriateness of the methods described in the 2nd and the 3rd treatments described in the Results section. A: Postulated change of V_p. B: Postulated change of V_c. C: Curve for y_ll (= V_p/V_c). D: Curve for y_lll (= ). E: Curve for dy_lll/dx. F: Curve for d²y_lll/dx². In this particular exhibition, too, we set V_p = 0.5x + 0.07, and V_c = − 0.2x + 0.35. The explanations given in the legend for Fig. 4 regarding the abscissa and ordinate also apply to this figure. For further explanations, see text.

The first order differential function of equation “4” is calculated as:

When x = d/c, the equation diverges towards infinit, but such a condition is not practical within the defined region of x.

From equation “5”, we can obtain the second order differential function of the relation as:

Under the present condition, x is smaller than d/c, and the function is positive. In other words, the incline of equation “4” increases within the defined region (0 ≦ x ≦ 1).

The third treatment

Since function “4” took a concave shape, we concluded that the relation could not be employed as such. We thus tried to analyze changes in the shape of the function in which function “4” was further extracted to its square root, as:

The function is shown as curve D in Fig. 6. In order that the value within the root be positive, x must be within the range between −b/a and d/c. Such conditions are naturally satisfied.

The first and the second order derivatives of equation “7” were calculated as:

Curves E and F in Fig. 6 show functions (8) and (9), respectively. Examination of the product members constituting these equations shows that all the terms except for (4acx + 3bc − ad) are positive within the defined region. On the other hand, the points of inflection of the initial function “7” can be obtained when its second-order differential of the initial function equals zero. Thus, the point can be obtained the point by setting the last term of equation “9” to be zero, the point thus obtained being x = (ad − 3bc)/4ac.

This means that the term (4acx + 3bc − ad) is positive or negative, when x is larger or smaller than (ad − 3bc)/4ac, respectively. Within the region where x is smaller or larger than the point, the curve takes a convex or concave shape, respectively, and approximates a straight line in the vicinity of the inflection. Substituting various positive numbers for a, b, c and d into equations “1” and “2”, we examined changes in the shapes of curves of the function as well as of its derivatives. In this way, we could recognize that function “7” becomes straighter as - b/a becomes more negative and/or that d/c becomes more remote from the upper limit of the actual range, namely, unity.

Modification of the 2nd and 3rd treatments

If we want to have the point of inflection around the middle way of the response (x = 0.5), we can do so. In the representative simulation displayed in Fig. 4 and also in Fig. 6, we have already adopted 0.50, 0.07, 0.20 and 0.35 for a, b, c and d, respectively. We can perform the operation, for example, by changing only d. At the point of inflection, the term “4acx + 3bc − ad” should equal zero, as mentioned above. Thus, d = 4cx + 3bc/a. Substituting real numbers into the equation, we can obtain the value of d as 0.484. Changing the value of d means shifting function “2”. By applying a voltage bias to the input for channel 2 of the divider circuit, such a shift can easily be performed electronically. For this purpose, we employed a simple circuit shown as Rb₂ in Fig. 2C. If such precautions are duly taken before recording the leucophore response, the method may be convenient, although such precise and sensitive settings are not needed for practical purposes.

In order to improve the linearity in recording the cellular response, we have tried to further modify the method. Observations on the changes of curves for function “7” and for its derivatives (“8”, “9”), showed that when the intercepts of the original functions (“1” and “2”) became larger, the linearity of the integrated relation (“7”) increased. In the simulation shown in Fig. 7, the intercepts of functions “1” and “2” were increased by 1.0, namely, b = 1.07 and d = 1.35 (cf. lines A (Y_p) and B (Y_c)). Looking at the actual relation and its derivatives (lines D, E and F in Fig. 7), we recognized that the linearity was significantly improved if compared with the case of y_lll. Such shifts of the relations can be easily realized by adding voltage biases to the input of the divider circuit. By adjusting Rb₁ and/or Rb₂, such operation can be easily done electronically.

Fig. 7

Simulated lines and curves for improving the 3rd treatment described in the Results section. In this particular simulation, either V_p or V_c was shifted up by 1. A and B: Line for Y_p = V_p + 1 = 0.5x + 1.07. B: Line for Y_c = V_c + 1 = − 0.2x + 1.35. C: Curve for Y_p/Y_c. D: Curve for . E: Curve for d/dx(). F: Curve for d²/dx²(). The explanations given in the legend for Fig. 4 regarding the abscissa and ordinate also apply to this figure. For further explanations, see text.

As a typical recording obtained by this method, Fig. 8 illustrates the responses of a leucophore from a medaka scale to electrical stimulation. Leucosomes in the cell dispersed in response to pulses of various frequencies. In this way, practically identical recordings to those obtained by the method described as the first treatment could also be obtained by this method.

Fig. 8

Typical recording of the responses of a leucophore of the medaka, Oryzias latipes, to electrical nervous stimulation of chromatic nerves and to norepinephrine. In this recording, the amounts of light scattering from the dendritic zone (S_p) and from the central body (S_c) were processed by the 3rd treatment, as described in the Results. The stimulation was performed by biphasically deformed square pulses (1 msec, 10.0 V) at the various frequencies noted in the figure. Finally, 2.5 μM norepinephrine solution was added to stimulate a maximal level of leucosome dispersion.

Application to motile iridophores

We then applied the developed method to motile iridophores of the dark sleeper goby, Odontobutis obscura obscura. Figure 9 shows the typical responses of a cell to norepinephrine recorded while the apparatus was set to operate under conditions of the first treatment. The upward shift of the trace was due to the dispersion of light-reflecting platelets into the peripheral dendritic zone, while the downward shift indicates their aggregation into the cell body. In describing the optical property of motile iridophores, incidentally, the term “light scattering” as used for leucophores may appropriately be replaced with “light reflectance” because the lightrays are reflected mainly from the flat surfaces of the reflecting platelets (Fujii et al., 1991b; Fujii, 1993).

Fig. 9

Typical recording of the response of a motile iridophore on a scale of the dark sleeper goby, Odontobutis obscura obscura. After equilibration in physiological saline, 2.5 μM norepinephrine (NE) was applied. The recording was made using the electronic circuit for the 1st treatment described in the Results. A gradual dispersion of light-reflecting platelets took place. Upon rinsing the scale with physiological saline again, reaggregation of them occurred.