Photoinactivation of A Portion of A Neurone For Long-Term Studies of Its Role In Behaviour

Late instar nymphs of male cockroaches Periplaneta americana were used in all experiments. Under cold anaesthesia, the ventral cuticle overlying either an abdominal connective or the meso-metathoracic connective was removed, the nerve cord supported by a platform, and an axon impaled. Microelectrodes were filled at the tip with 5% LYin 0·1 mol 1⁻¹ LiCl (tip resistances 35-70 MΩ). The dye was injected using constant hyperpolarizing current of 5–50nA, for 5–30 min, until the axon was brightly filled as viewed in a Wild M5A microscope equipped with a Leitz fluorescence epi-illuminator (peak excitation wavelength 426 nm). After axonal filling, the epi-illuminator beam was stopped down to a 1·5 mm spot on the axonal segment to be inactivated. After irradiation, a few grains each of penicillin, streptomycin and phenyl thiourea were placed in the open wound. The wound was then sealed, either by replacing the cuticle (abdomen) or by covering with a stretched piece of Parafilm (thorax), and closed with the surgical glue Histoacryl Blue. The animal was then kept at room temperature and fed rat chow and water supplemented with penicillin (1 mg ml⁻¹) and sucrose (80 nmol 1⁻¹). The day after the operation the animal was tested for escape turning responses to controlled wind puffs (Camhi & Levy, 1988). We recorded these responses using a Sony Betamax videotape, under strobe illumination.

We first tested the ability of photoinactivation of a small region of an abdominal giant axon to block acutely the conduction of action potentials (Fig. 1A). After filling a giant axon with LY, action potentials evoked by posterior stimulating hook electrodes were recorded intracellularly and, with a more anterior pair of hook electrodes, extracellularly (Fig. 1B, first panel). Upon hyperpolarizing the impaled axon to block its spike, a discrete wave (indicated by the dot in the first panel) was lost from the extracellular trace (Fig. 1B, second panel), representing the contribution of the impaled axon to the recorded compound action potential. Upon releasing the axon from hyperpolarization, the discrete wave returned (Fig. 1B, third panel). Then, after 7–5min of spot illumination centred on the recording microelectrode, both the intracellular action potential and the resting potential disappeared, as did the corresponding wave in the extracellular trace (Fig. 1B, fourth panel).

Fig. 1C shows, from this same axon, the intracellularly recorded action potential at six different moments during the course of the illumination. The spike first broadened, developing a long falling phase, then developed additional bumps, and finally was lost. The resting potential decreased throughout this period (compare heights relative to arrowhead positions), with the greatest jump coming at the very end. In each of three abdominal giant interneurones (GIs) and each of three thoracic interneurones tested, this same sequence of spike blockage and loss of resting potential was seen. The mean time to complete loss of the spike was 9 min (range 4–20min). In each case, as soon as the action potential disappeared, the illumination was turned off. The recording was maintained for a further 15–45 min, and the action potential never recovered.

To examine whether the inactivated region of the axon remained inactive at least 1 day later, we began by testing whether an axon different from the one filled the previous day conducted action potentials normally, without spike blockage in the region of connective previously illuminated. We demonstrated that this axon, impaled one body segment posterior to the previous day’s impalement, was not the previously impaled axon, by filling it very slightly with LY, and noting that it was separate from the still filled GI of the previous day. When we stimulated with anterior hook electrodes we were able to evoke in this impaled axon a spike at a fixed latency (Fig. 1D, left panel). We then stimulated with hook electrodes located posterior to the microelectrode, at a voltage set near threshold for the impaled axon. By recording with hook electrodes anterior to the microelectrode, we found that those stimulus pulses that evoked an intracellularly recorded spike also evoked an extracellular spike (Fig. 1D, middle trace). To arrive at the hook electrodes, the spike had to pass through the previously illuminated region of connective. Finally, intracellular stimulus pulses evoked a spike in both posterior and anterior pairs of hook electrodes (Fig. 1D, right panel). In summary, then this axon conducted normally through the irradiated region of connective.

We next repeated an identical set of experiments on the axon that we had filled on the previous day, which we now reimpaled, also one body segment posterior to the original impalement site. The reimpalement was carried out under dim blue illumination, to reveal the fluorescent target axon. To check that the fluorescent axon had been impaled, we injected a small amount of LY and could see that this new dye clearly merged with the existing fluorescence. Now, stimulating with anterior hook electrodes, we were unable to excite the impaled axon, even with stimulus pulses large enough to excite a great many other axons, as seen from the compound action potential recorded with the posterior hook electrodes (Fig. 1E, left panel). In contrast, stimulus pulses from the posterior hook electrodes did evoke a spike recorded intracellularly. By using a fixed stimulus voltage near threshold for the impaled axon, we saw no difference in the extracellular trace between those simulus pulses that did evoke an intracellularly recorded spike and those that did not (Fig. 1E, middle panel). This verified that the spike of this axon was blocked in the previously illuminated region. As further confirmation, intracellular stimulation produced a spike at the posterior but not at the anterior hook electrodes (Fig. 1E, right panel).

After the above experiment, we filled the reimpaled axon with yet more Lucifer Yellow using 5·10⁻⁸ A for 40 min to achieve a dense staining. We then examined it in a compound epifluorescence microscope (Fig. 2A). In the reimpaled axon, the dye diffused in both directions, towards locations 1 and 3 in Fig. 1 A. However, it accumulated at a discrete blunt end-point, located at the border between the illuminated and unilluminated regions (left limit of dotted circle in Fig. 1A). This suggests that the axon had become sealed at this location, as if by a plasma membrane. A much weaker fluorescence appeared to the right of the seal in Fig. 2A, in the portion of axon that had been irradiated. The dye here is still conlined to a region roughly equal to the diameter of the axon. This could either be the dye remaining from the initial filling session or could represent the leakage of some dye from the second filling session, through the sealed end. In either event, the presence of dye in this region suggests either that the illuminated region of axonal membrane remains impermeable to Lucifer Yellow, even though the axon is physiologically inactive there, or that the dye is held in place by a sheath around the possibly degenerated axonal membrane.

Also seen in this figure is another, less densely filled, axon also terminating in a similar blunt end-point. This corresponds to a second axon that had been filled in this preparation on the first day, prior to photoinactivation, but was not reimpaled on the second day. (It is less brightly filled because it was not filled again on the second day.) A Nomarski differential interference contrast image of the same region of tissue as that in Fig. 2A shows a membrane-like seal in each of the two axons previously filled and irradiated (Fig. 2B). In four of five abdominal axons that were examined on the second day, a blunt end-point similar to that in Fig. 2A,B was seen.

We have also analysed the animals’ escape behaviour after photoinactivating single axons in the meso-metathoracic connective. In each of these experiments, we irradiated for 30 min, 10 min longer than the longest time required to block the spike in any of our previous experiments. Of the 19 animals with successful impalements, nine responded behaviourally the next day to a sufficient number of wind stimuli (between 11 and 50) to permit a test of behavioural normality. In each of these, a discrete LY axon was subsequently localized histologically in the meso-metathoracic connective.

Fig. 2C shows a cross-section of the meso-metathoracic connectives from one of these animals, with a single LY-filled axon. This axon is identifiable by its position and size, as well as by its physiological response to wind stimuli, as giant intemeurone 3 (GI3) (Spira et al. 1969; Westin et al. 1977). Fig. 2D represents graphically the turning behaviour of this animal to wind stimuli. The line graph, representing the linear regression plot for the data of this animal, shows that the cockroach consistently turned away from the wind. The slope of this line, -0-39, is within the range previously shown for normal animals (Camhi & Tom, 1978), as were the slopes for all the animals we tested.

Fig. 2E shows the linear regression plots for all nine preparations tested. The axons killed in these experiments included two GIs3, six cells not previously identified, and one lambda cell (dotted line). Killing GI3 by other methods has been shown not to produce behavioural deficits (Comer, 1985). Thus, the normality of the behaviour in these two animals with GI3 killed, as well as of those with unidentified cells killed, demonstrates that the present experimental procedure (including the surgery, filling and irradiating) did not disturb the escape directionality, and so this method can be used to search for cells involved in the control of turn direction. The normal behaviour following the killing of the lambda cell shows that this neurone, excited by all the ventral GIs known to mediate escape (Ritzmann & Pollack, 1986), does not by itself control the directions of escape turns.

In summary, a discrete photoinactivated region of a Lucifer-filled axon in the cockroach remains inactive for at least 1 day. Just proximal to the site of inactivation, the axon usually forms a blunt end-point that accumulates dye. This technique disturbs the cockroach sufficiently little that normal escape behaviour can be elicited, assuming that the killed cell is not crucial for the turn direction. The method thus permits one to search for neurones that mediate escape turns, or more generally, any behaviour of interest. The procedure should also permit studies of regeneration of a single axon in an environment much more orderly than that following cutting or crushing, the usual methods employed for this purpose. Such studies are now in progress.

This work was supported by grant number 86/152 from the US-Israel Binational Science Foundation to JMC and AS. We thank Hanah Sassoon for histological preparation.