Intern. J. Neuroscience, 116:247–264, 2006 Copyright 2006 Taylor & Francis Group, LLC ISSN: 0020-7454 / 1543-5245 online DOI: 10.1080/00207450500403033
BRAINSTEM INPUT MODULATES GLOBALLY THE TRANSMISSION THROUGH THE LATERAL GENICULATE NUCLEUS
T. OZAKI EHUD KAPLAN The Rockefeller University and The Mount Sinai School of Medicine New York, New York, USA
The transmission of visual information from the retina to the visual cortex through the lateral geniculate nucleus (LGN) is a complex process, which involves several neuronal mechanisms, elements, and circuits. The authors investigated this process in anesthetized, paralyzed cats by recording from LGN relay neurons, together with their retinal input, which appeared as slow (S) potentials. The major findings are: (1) The transfer ratio (LGN firing/retinal firing) fluctuated slowly and (2) these fluctuations in transfer ratio were synchronized across the nucleus, did not depend on visual stimulation, and were highly correlated with neural activity in the parabrachial nucleus of the brainstem (PBN). Electrical stimulation of the PBN increased transmission from retina to cortex through the LGN. It is concluded that the PBN, which is part of the Ascending Arousal System, can modulate globally the transmission of information through the thalamus. Keywords brainstem, information transmission, LGN, modulation, thalamus, vision
Received 4 March 2005. This work was part of T. Ozaki’s Ph.D. thesis work at the Biophysics Laboratory at The Rockefeller University, and was supported by NIH grant EY-4888 to EK, and by an NEI core grant EY-10867. EK was the Jules and Doris Stein Research to Prevent Blindness Professor at the Ophthalmology Department at Mount Sinai School of Medicine. Address correspondence to Ehud Kaplan, The Mount Sinai School of Medicine, New York, NY 10029, USA. E-mail: [email protected]
T. OZAKI AND E. KAPLAN
INTRODUCTION The lateral geniculate nucleus (LGN) is a complex, multilayered structure that is strategically positioned along the neural highway from retina to visual cortex. Most of the synapses within the LGN originate from non-retinal sources (Sherman & Guillery, 1996), and these are believed to influence the transmission of information from the retina to the visual cortex. Although many of these synapses are excitatory, it appears that only retinal ganglion cells (RGCs) are able to elicit nerve impulses in relay cells in the LGN (Kaplan & Shapley, 1984). Thus, the primary role of the LGN is merely to delete nerve impulses from the impulse train carried into it by the RGCs. The process that determines what goes through and what is discarded was the focus of this study. The pattern of innervation of the LGN suggests that transmission control involves a delicate interplay between the excitatory and inhibitory influences that originate in the retina, visual cortex, brainstem, and local interneurons (see, for example, Uhlrich et al., 1988, 1991). The brainstem input appears diffuse and broad, whereas others inputs, especially from the visual cortex, seem more localized. Previous studies have shown that electrical stimulation of brainstem regions increases the response of LGN relay neurons to visual stimuli, but without significant modification of the spatial parameters of the receptive field (Uhlrich et al., 1990, 1995). The present study was initially motivated by the observation of large, slow fluctuations in LGN firing rates under constant visual conditions, with no time varying stimulation (Ozaki et al., 1996). The study reported here investigated the nature of the process of transmission control. Specifically, it was asked whether the process that determines which neural messages will be allowed to continue from retina to visual cortex is local or global. The results show that the input from brainstem to the LGN modulates transmission, and that modulation is global. This study differs from previous studies of the effects of brainstem on LGN function because the S potential recording approach (see later) allowed the authors to focus on the issue of transmission from retina to cortex, and because they recorded simultaneously from more than one LGN location. METHODS Surgical Preparation The experiments were performed on 20 male adult cats. Anesthesia was induced by an intramuscular (IM) injection of xylazine (Rompun, 2 mg/kg)
followed by ketamine hydrochloride (Ketaset, 10 mg/kg). Sites of invasive surgery were shaved and scrubbed with Betadine (povidine-iodine, 10% solution). The local anesthetics Xylocaine (Lidocaine hydrochloride, 20 mg/ml solution) or Novocain (procaine hydrochloride, 20 mg/ml) were also given as needed during surgery. Two femoral veins were cannulated for intravenous (IV) infusion of pharmacological agents during the experiment. After cannulation, 0.5 cc solution of either Pentothal (thiopental sodium, 10 mg/ml solution) or a 20:1 solution of Diprivan (propofol, 10 mg/ml), combined with sufentanil (5 µg/ ml) was given to initiate deeper anesthesia. Additional injections of anesthetic were used as needed during preparatory surgery to maintain surgical anesthesia. Ampicillin (1.5 cc, 30 mg/ml IV) to prevent infection, and 1.5 cc dexamethasone phosphate (4 mg/ml IV), to prevent edema, were also administered daily. Drops of Neo-synephrine (phenylephrine hydrocloride, 10%) and 1% atropine sulfate ophthalmic solution were applied to the eyes, to dilate the pupils and retract the nictitating membranes. The first 17 experiments used pentothal as the long-term anesthetic, infused at 2–6 mg/kg/hr. The last three experiments used a combination of propofol and Sufenta, infused at a rate of 3–6 mg/kg/hr). No systematic differences were seen between the data sets collected under the two different anesthetic regimes. A femoral artery was also cannulated to monitor arterial blood pressure, and a urinary catheter was inserted to monitor fluid output. The trachea was cannulated to allow artificial respiration. Muscular paralysis was induced with an IV injection of 1–1.5 cc of Pavulon (pancuronium bromide, 2 mg/ml) and was maintained throughout the experiment by continuous infusion (0.2–0.4 mg/kg/hr). The animal also received infusions of lactated ringer solution with 5% dextrose at 3–5 ml/kg/hr. The cat’s head was fixed in a stereotaxic apparatus, which rested on a floating air table. The cat’s body was suspended from a vertebral clamp, to reduce respiratory artefacts. Gas permeable hard contact lenses covered the eyes to prevent corneal drying. The lenses were periodically removed and cleaned during the experiment. Artificial pupils (3 mm in diameter) were placed in front of the contact lenses to keep the retinal illumination constant. The optical quality of the eyes was checked regularly by ophthalmoscopy. Core body temperature was maintained at 38.5ºC throughout the experiment by a DC-powered heating pad, controlled by a proportional feedback circuit with a subscapular temperature probe. The blood pressure, heart rate, and expired CO2 level were continuously monitored with a Hewlett-Packard patient monitor (model 78354A). The mean
T. OZAKI AND E. KAPLAN
arterial blood pressure was recorded by the computer as a voltage from the J11 A/D card of the HP patient monitor to which the cat was connected, and maintained between 90–120 mm Hg by regulating the delivery of the Ringer’s solution and anesthetic. In experiments that investigated the effects of blood pressure on LGN transmission, the authors administered an IV injection of a 0.1 cc Aramine (metaraminol bitartrate, 0.5 mg/ml) to raise the blood pressure. Expired CO2 was maintained between 3.5–4.5% by adjusting the respirator stroke volume or rate. The air to the lungs was moistened to prevent drying of the pulmonary tissues. Occasional supplements of O2 were provided during the course of the experiment. These experiments often lasted for 3–5 days, after which the animal was sacrificed with a barbiturate overdose. The experimental protocol described earlier was approved by the Rockefeller University and Mount Sinai Animal Care and Use Committees, and was in accordance with the National Institutes of Health guidelines for the use of higher mammals in neuroscience experiments. Electrophysiological Procedures
Single Electrode Recording. A craniotomy was performed over HorsleyClarke coordinates 6.5A, 9L. The dura mater was then removed, and an electrode was lowered rapidly to a depth of 8 millimeters below the surface of the cortex. The electrode was then advanced by a stepping-motor drive (smallest step size: 1 µ) in steps of a few microns until spike activity from single neurons in response to visual stimuli was detected. This occurred typically at a depth of ~13 millimeters below the surface of the cortex. The authors studied the activity of relay cells only in layers A and A1 of the LGN. Extracellular recordings of single units were made with tungsten-in-glass electrodes, with a 5–10 µm long tip coated in platinum black (Merril & Ainsworth, 1972). The output was then directed to two oscilloscopes, each with its own audio monitor and window discriminator, filtered with adjustable low (30 to 300 Hz) and high frequency cutoffs (3 to 10 kHz). One oscilloscope was set to monitor LGN spikes (LGN output), whereas the other was used to monitor S-potentials (retinal input to the LGN, see Figure 1 and next section). Receptive fields (RF) of single cells were mapped by hand with a flashing spot on a tangent screen. The optic disk was also mapped by reverse ophthalmoscopy on the tangent screen, as a reference for the retinal
position of the RF. Each cell was characterized as ON or OFF center by the polarity of its response to a large flashing stimulus. S-Potential/Spike Pair Recordings. S-potentials are slow synaptic potentials, also called pre-potentials, generated by impulses arriving along the optic tract (Bishop et al., 1953; Cleland et al., 1971). Kaplan and Shapley (1984) reported that injections of the sodium channel blocker tetrodotoxin (TTX) into the eye of the cat caused spontaneous and light evoked S-potentials to disappear, whereas electrical stimulation of ganglion cell axons at the optic chiasm was still able to elicit S potentials. This result proved that ganglion cell impulses were the only source of these potentials, which may therefore be used as a measure of ganglion cell input to the LGN. Kaplan and Shapley also showed that without this retinal input the LGN did not fire action potentials. The recording of the input and output of the relay cell allows the examination of the transmission properties of the LGN. A sample of this type of recording can be seen in Figure 1. With this recording method, action potentials can be easily distinguished from retinal S-potentials by their amplitude and shape.
Figure 1. An example of an S-potential/spike pair recording from an LGN relay cell in the cat. The large spike-like depolarizations with an afterhyperpolarization are the action potentials from the LGN relay cell (denoted with black dots). The smaller depolarizations, which are monophasic and slower than the action potentials, are S-potentials (denoted with the hollow dots). Only 4 of the 10 RGC spikes elicited LGN spikes. The voltage trace was sampled at 12 kHz.
T. OZAKI AND E. KAPLAN
Simultaneous Recording of PBN and LGN Cells. In order to record simultaneously the activity of PBN neurons and LGN relay cells, two recording electrodes were placed within the brain of the animal: one in the LGN and the other within the PBN, at stereotaxic coordinates AP 0, L 2, H –1. Recording sites were confirmed histologically after the experiment. Electrical Stimulation of PBN Cells with Simultaneous LGN Recording. For electrical stimulation of the PBN, two stimulating electrodes were placed at AP 0, L 3, and H –1, one on each side of the head. Electrical stimulation of the PBN consisted of 150 µs pulses (500 µA to 750 µA) at 50 Hz applied for 1 s (Uhlrich et al., 1995). A full stimulus period consisted of these bursts of electrical stimulation, repeated 4–8 times, interleaved with periods of nonstimulation to avoid damage to the brainstem. The bilateral placement of the stimulating electrodes assured that a large portion of the PBN would be activated by the current stimulation.
Visual Stimulation Visual stimuli were presented on the face of a Conrac CRT (60 cd/m2, 135 frames/s, placed 114 or 57 cm from the animal), using specialized hardware and software developed in the laboratory (Milkman et al., 1980). For single unit recordings, the receptive field of a cell was centered in the middle of the CRT with an adjustable mirror. For recordings of pairs of LGN relay cells, the mirror was adjusted so that both receptive fields were centrally located on the CRT. Optimal refraction was determined by placing spherical lenses of varying power in front of the eye in order to maximize the modulated firing response to a drifting sinusoidal grating. Each cell was characterized as either X or Y based on the strength of frequency doubling in the response to contrastreversing gratings of high spatial frequencies (Hochstein & Shapley, 1976). To examine changes in the LGN and retinal response over long periods of time, recordings of 5000 to 8000 s in duration were taken. The cat viewed either a uniform screen at approximately 60 cd/m2, or a grating of optimal spatial frequency for the recorded cell, drifting at 4 Hz.
Data Acquisition and Analysis Long stretches of electrophysiological recordings were obtained with the Discovery data collection system (Datawave Technologies Inc.), which permitted
recordings of nerve impulses from several channels, as well as other signals, such as stimulus markers, time stamps, or blood pressure. The transfer ratio was used to assess the fidelity of information transfer through the LGN. The transfer ratio is simply the fraction of retinal spikes that result in LGN spikes over some period of time. A convenient way to calculate this value is to measure the firing rate for both the LGN and the retina over some period, and then divide the LGN response by the retinal response. The authors have also Fourier analyzed the retinal and LGN responses, and examined the ratios of the fundamental components of these responses. Note that the LGN does not generate spikes without a retinal input (Kaplan & Shapley, 1984). Thus, the transfer ratio is always less than (or equal to) 1.0.
RESULTS The Transmission Through the LGN Fluctuates Slowly
Recording at One LGN Location. Figure 1 illustrates the most fundamental aspect of LGN function: an LGN relay cell passes on to the visual cortex only a fraction of the impulses that it receives from its retinal ganglion cell input. This fraction (always ≤ 1.0) fluctuates with time (Figure 2). When the eyes are stimulated by an effective visual stimulus, such as a drifting grating, these fluctuations are less dramatic (Figure 2, left panel) than they are when the animal is viewing a steady, unmodulated screen (Figure 2, right panel). These fluctuations are slow: their power spectrum is dominated by frequencies lower than 0.01 Hz (Kaplan et al., 1993). Similar results were obtained when the activity measure for stimulated discharge was the fundamental Fourier component of the spike train (not shown). Recording at 2 LGN Locations Simultaneously. Because the fluctuations in transfer ratio are not correlated with the visual stimulus, and appear also when no modulated visual stimulus is presented, they appear to be due to some network property, or some external, non-retinal input to the LGN circuitry. It was wondered whether these fluctuations in transmission ratio are local or global: when transmission is high at one LGN location, is it also high everywhere? To answer this question, simultaneous recordings were made from two extracellular electrodes, with each electrode recording one RGC/LGN cell pair.
T. OZAKI AND E. KAPLAN
Figure 2. Fluctuations in the firing rates of an LGN cell and its retinal input. (Left) Although the cat was viewing a uniform, unchanging screen, the LGN cell (red, lowest trace) shows slow fluctuations in its firing rate. These changes do not appear in the geniculate cell’s retinal input (black, middle trace). The trasfer ratio from RGC to LGN (LGN firing rate/RGC firing rate) is shown in the upper trace. (Right) Same as on the left panel, but here the cat was viewing a sinewave grating, drifting at 4 Hz. The firing rates shown here have been averaged across a sliding 12 s window, to help in visualizing the slow fluctuations. The thin gray “background” traces show the variability of the LGN and retinal firing rates using a 1 s counting window. The correlation between LGN firing rate and trasfer ratio was 0.99 for the unstimulated case and 0.97 for the stimulated case. (See Color Plate II at end of issue.)
Two tungsten-in-glass electrodes were placed simultaneously within the cat LGN. In order to record S-potentials as well as spikes, both electrodes had to be independently controlled, to allow for fine adjustments of each electrode position. However, due to the size of the electrodes and their holders, the electrodes could not be placed straight down into the LGN, and were inserted at a relative angle and distance such that upon reaching the recording area in the LGN, their tips were close together (approximately 1 millimeter). Fine stepping motors (1 µ step) were used to advance each electrode independently until the desired paired recording was obtained. Once spike activity from single neurons in response to visual stimuli was detected, the electrodes were slowly advanced until S-potential/spike pair recordings were seen in both electrodes. Typical results of such a dual recordings that lasted for approximately 90 min is shown in Figure 3. Here the (smoothed) firing rate of two RGCs (black traces) and their LGN targets (red traces) are shown. The fine details
Figure 3. The firing rate of two simultaneously recorded X-ON LGN relay cells and their retinal inputs recorded simultaneously (a and b). The firing rates of the LGN relay cells slowly fluctuate in synchrony, whereas the retinal ganglion cells do not. The receptive fields of the two cells were approximately 18 degrees apart, and were not overlapping. The cat viewed a sinewave grating, drifting at 4 Hz. (See Color Plate III at end of issue.)
of the two firing rate series are not identical, but the slow fluctuations in the LGN firing seem rather similar. This similarity is illustrated in Figure 4, which shows (left panel) the two transfer ratios plotted together as a function of time. The high correlation between the transfer ratios is shown in the right panel of Figure 4. The data in Figure 3 and Figure 4 were recorded while the cat viewed a sine wave grating, drifting at 4 Hz. Similar results were observed (highly correlated transfer ratios for the two cells) when no dynamical stimulus was presented (data not shown). The transfer ratio was highly correlated with the firing rate of the LGN relay cell (r = .99 for the spontaneous discharge, and r = .98 when the cat viewed drifting grating), but poorly correlated with the firing rate of the retinal ganglion cells (r = –.02 for spontaneous activity and r = –.016 during stimulation with drifting grating).
T. OZAKI AND E. KAPLAN
Figure 4. The transfer ratios for the two X-ON LGN relay cells are highly correlated. (Left) The transfer ratios from the data in Figure 3 are plotted together versus time. (Right) The transfer ratios from the left panel are plotted against each other. The correlation coefficient between the two was 0.95. (See Color Plate IV at end of issue.)
These results were observed under various conditions: when the receptive fields of the two LGN cells were overlapping or far apart, and independent of whether the cat was viewing a temporally modulated stimulus or not.
What Role Does the PBN Play in the Fluctuations of the LGN? The results illustrated in Figure 3 and Figure 4 suggested that the slow fluctuations in transmission ratio occur globally, across the entire LGN. The diffuse, modulatory innervation of the LGN by the pathway from the brainstem (Uhlrich et al., 1990, 1995; Lu et al., 1993) suggested that the brainstem might be the source of these global, slow fluctuations in LGN transmission. To study this possibility, two approaches were used: (1) simultaneous recording in the parabrechial nucleus (PBN) and the LGN, and (2) electrical stimulation of the PBN, while recording from a RGC/LGN pair.
(1) Simultaneous PBN-LGN Recordings. Figure 5 shows a cross correlation function between the firing rates of a typical PBN recording and an LGN relay cell. The cross correlation function shows a very broad, slightly offset peak, and the function returns to the baseline after 40–50 s (red trace in Figure 5). When the LGN spike train were shuffled, the correlation disappeared (black trace in Figure 5), validating the notion that it is the temporal
BRAINSTEM INPUT INPUT BRAINSTEM
Figure 5. Cross correlation between the simultaneously recorded LGN and PBN cells furing periodic stimulation. The cross correlation between a simultaneously recorded LGN relay cell and a cell from the parabrachial nucleus is shown in red. Notice that the peak showing positive correlation between the two cells is both broad and has a slight lag. Shuffling the LGN spike train eliminates this positive peak (black). (See Color Plate V at end of issue.)
structure of the LGN firing rate, and its dependence on the PBN firing, that gives rise to the positive peak in the cross correlation. When the PBN firing rate was correlated with the firing of the RGC, the cross correlation was essentially flat (not shown). The very sluggish, broad correlation shown in Figure 5 is consistent with the slow, global fluctuation in LGN firing rate and transmission ratios illustrated in the previous sections, and thus supports the hypothesis that the PBN plays a role in the generation of these fluctuations.
(2) Electrical stimulation of the PBN. The PBN was stimulated electrically as described in the Methods section, while recording S-potential/LGN relay cell pairs. Typical results are shown in Figure 6, which shows the firing rate of an LGN relay cell (red trace), its retinal ganglion cell input (black middle trace), and the LGN/RGC transfer ratio (upper black trace) during electrical stimulation (50 Hz stimulation, 150 µS pulses, 500 mA for 1 s). It is clear that the transfer ratio increases immediately following PBN stimulation, and returns slowly to its previous resting state.
T. OZAKI AND E. KAPLAN
Figure 6. Firing rate and transfer ratio of an LGN relay cell during stimulation of the PBN. (a) The transfer ratio from retinal galgion cell (black trace in b) to an LGN relay cell (red trace in (b) during electrical stimulation of the PBN. The stimulation of the PBN increased the transfer ratio from retina to LGN. The small decreases (dips) in the retinal ganglion cell firing rate, seen in the black trace in b, are artifacts due to the nature of S-potentials recordings, which were affected by the electrical stimulation. (c) shows the periods of electrical stimulation (1 s, 50 Hz pulses, 500 µA), interleaved with periods of no stimulation to avoid damage to the parabrachial nucleus. (See Color Plate VI at end of issue.)
The time course of the transfer ratio return to its resting state after the end of electrical stimulation is very similar to the falling phase of the cross correlation function shown in Figure 5. This similarity provides further support to the notion that PBN neural activity plays a major role in the determination of the transfer ratio from retina to cortex.
Effects of Blood Pressure. The authors noticed that during electrical stimulation of the PBN, the blood pressure of the animal increased slightly. However, it was determined that this increase could not have accounted for the increase in LGN transmission. The authors recorded from a RGC/LGN pair, while injecting the vaso constrinting drug Aramine into the animal’s veins, with no effect on the LGN firing rate, which is highly correlated with the transfer ratio (Figure 2).
DISCUSSION Summary of Results The investigation of transmission from retina to visual cortex through the LGN has shown the following: (1) The transmission is not constant, but rather fluctuates slowly; (2) The fluctuations are not related to the retinal input; (3) The fluctuations are global, and involve much of the LGN; (4) Activity in the Parabrachial Nucleus (PBN) of the brainstem is highly correlated with the fluctuations in LGN transmission. These results, together with the finding that electrical stimulation of the PBN increases LGN transmission of retinal signals, point to the PBN as a significant player in the process that determines the faithfulness with which the LGN passes on to the cortex retinal messages. LGN relay neurons receive a multitude of inputs. These include inputs from the retina, visual cortex, local interneurons, efferents from the reticular nucleus of the thalamus (TRN), and input from the brainstem. It is possible that some input other than that from the brainstem participates in the modulation of information flow through the LGN. However, the characteristic intrinsic fluctuations of relay cells and interneurons (e.g., Zhu et al. 1999) are too fast to be likely candidates. On the other hand, the sluggish dynamics and the global nature of the modulations observed, and the correlation of transfer ratio with the neural activity in the brainstem, all support the notion that the brainstem input is crucial for the large, global fluctuations observed in LGN transmission. The results do not show what cellular or synaptic processes might be involved in the filtering of information through the LGN, or how the PBN input might modulate these processes. The observation (Kaplan & Shapley, 1984) that without retinal input LGN relay neurons are silent, suggests that the LGN can only veto and delete impulses from the retinal spike train (but compare Eysel et al., 1986). However, if the PBN input were to affect the membrane potential of LGN relay neurons, it could easily have a profound effect on their dynamics as well, because of the T type Ca channel that is common in thalamic neurons (Llinás & Jahnsen, 1982; Jahnsen & Llinás, 1984a, 1984b), a cation channel that depends on the membrane potential. Hyperpolarization of relay cells will activate the T channel and shift them into their bursting mode, whereas depolarization, which will inactivate the T channel, will return them to a tonic mode. The neurochemistry of the PBN input suggests the possibility of effects on two different time scales: slow modulatory effects that are due to slow acting cholinergic metabotropic connection (De lima, and Singer, 1987; Smith
T. OZAKI AND E. KAPLAN
et al., 1988; Uhlrich et al., 1988; McCormick, 1992; see also Eysel et al., 1986, for more cholinergic effects in LGN), and faster nicotinic synapses, which are known to have a higher activation threshold than the cholinergic terminals. It is possible that they act cooperatively, with the nicotinic pathway making rapid changes of the overall level that is set by the metabotropic cholinergic pathway. The parabrachial nucleus, which is part of the Ascending Arousal System, has been shown to receive retinal input in mammals (Fite & Janusonis, 2002). However, it is likely that the main influence on PBN activity are various environmental factors, including the state of arousal, rather than the retinal activity, because the fluctuations in the LGN activity and transfer ratio are uncorrelated with the firing rate of retinal ganglion cells. This, of course, applies only to those retinal ganglion cells that innervate the LGN, and not to those that terminate in the PBN. The physiological properties and mode of action of those RGCs that project to the PBN are not yet known.
Sample Size The results reported here were observed in all 16 pairs of LGN cells recorded from. This is a relatively small sample, and its size is due to the extreme technical difficulty involved in these experiments. To begin with, the recording of S-potential/spike pairs within the LGN is challenging. This recording technique becomes extremely difficult when a second electrode is introduced into the same LGN: when the second electrode is advanced, the recording from the first electrode often deteriorates or is lost. Further, in order to perform the analysis in this study, these recordings had to be stable for a long time. Paired recordings of LGN and S-potential provides an independent control on the stability of the recording. Because the slow fluctuations show up in the LGN recordings and not in the more challenging S-potential recordings, there is confidence that the slow fluctuations in the LGN are real and not just degradations of the recording or fluctuations in signal-to-noise ratio or in the health of the cell. Most importantly, the observations presented here (slow, synchronized LGN fluctuations, tightly correlated with PBN activity) were uniform across all cells measured, regardless of their cell type, distance between their receptive fields, or stimulus conditions. To the authors’ knowledge, there are no published accounts of in vivo electrophysiological recordings from parabrachial cell in cat or monkey, probably because of the challenging nature of such an experiment. The PBN has fewer cells than other nuclei (e.g., LGN), and these cells are spread in a
relatively flat, layer-like structure, making vertical penetration particularly inefficient for recording (Uhlrich et al., 1995). The heavy myelination of the many fibers that travel through the brainstem area also adds to the difficulty in recording from PBN cells. Finally, because there is no topographic mapping within the PBN, it is extremely difficult to find and isolate cells of interest. The only reliable way to estimate the success of the recording is by post-experiment histological confirmation. Other Modulatory Inputs to the LGN The PBN is, of course, not the only possible source of modulation of LGN transmission. The perigeniculate nucleus (PGN) and the visual cortex also project to the LGN. However, their input is an unlikely cause of the fluctuations reported here for two main reasons: (1) PGN activity has been shown to be negatively correlated with LGN firing (Funke & Eysel, 1998) and (2) the corticofugal feedback is topographically organized (Tsumoto et al., 1978), unlike the diffuse PBN innervation. There are numerous other inputs to LGN relay cells, but there is not sufficient data about their function to assess confidently their involvement in transmission control. Is LGN Transmission Bi Stable? The existence of the fluctuations in transfer ratio (Figure 2, upper traces) was independent of whether the cat was visually stimulated or not. However, the average transfer ratio was considerably higher during visual stimulation (compare left and right upper traces in Figure 2). In addition, the transfer ratio appears to be bistable, and is typically in either low or high mode. The transfer ratio traces are thus reminiscent of recordings from single ion channels, where the channel goes rapidly from an open to a closed state (or vice versa), and remains in that state for a relatively long time. As a result, a histogram of the transfer ratio time series is clearly bimodal (not shown). It is not clear whether this bistability is related to the up and down states of cortical neurons (see, e.g., Shu et al., 2003), but it is noted that the visual cortex sends a powerful efferent feedback to the LGN.
Previous Studies that Showed Brainstem Involvement The brainstem influence on LGN function has been known for a long time. For example, the locus ceoreleus has been shown to influence LGN response
T. OZAKI AND E. KAPLAN
(Rogawski & Aghajanian, 1980; Kayama et al., 1982). The results of Reppas et al. (2002) are also consistent with a modulation of retina-to-LGN transmission that is related to input from the brainstem. The present results are consistent with previous results on the effects of PBN stimulation on LGN responsivity (Uhlrich et al., 1995), but because not only the LGN relay neuron but also its RGC input were recorded, the author demonstrated that the effect could not be due to a change in retinal activity or to elevated blood pressure. In addition, by recording from two separate locations in the LGN simultaneously, the authors have showed that the PBN’s effect on the LGN was global, consistent with the expectations based on the diffuse nature of the innervation from PBN to LGN. It has also been demonstrated that there was a positive correlation between PBN neural activity and LGN response, even during spontaneous maintained discharge.
Effects of Anesthesia The experiments were conducted under light anesthesia of either a barbiturate of propofol. Slow fluctuations in EEG are typically interpreted as an indication of synchronized firing, which is common under deep anesthesia. However, it is believed that the fluctuations reported here in retina → cortex transmission are not due to the anesthetics used. This is because the anesthetic level was light, as indicated by the EEG, and because no difference was noticed between the barbiturate and propofol anesthesia.
CONCLUSIONS The LGN is no longer viewed as a passive nucleus, which slavishly relays to the cortex the retinal messages. The results of this article show that the brainstem plays an important role in the dynamical balance that determines transmission through the entire LGN. More focused feedback from the visual cortex is probably responsible for fine tuning the global influence of the brainstem, in ways that perhaps reflect the nature of the visual stimulus.
REFERENCES Bishop, P. O. (1953). Synaptic transmission. An analysis of the electrical activity of the lateral geniculate nucleus in the cat after optic nerve stimulation. Proceedings of the Royal Society (London) B, 141, 362–392.
Cleland, B. G., Dubin, M. W., & Levick, W. R. (1971). Simultaneous recording of input and output of lateral geniculate neurons. Nature New Biology, 231, 191–192. De Lima, A. D., & Singer, W. (1987). The brainstem projection to the lateral geniculate nucleus in the cat, identification of cholinergic and monoaminergic elements. Journal of Comparative Neurology, 259, 92–121. Eysel, U. T., Pape, H.-C., & van Schayck, R. (1986). Excitatory and differential disinhibitory actions of acetylcholine in the lateral. Journal of Physiology (London), 370, 233–254. Fite, K. V., & Janusonis, S. (2002) Optic afferents to the parabrachial nucleus. Brain Research, 043, 9–14. Funke, K., & Eysel, U. T. (1998) Inverse correlation of firing patterns of single topographically matched perigeniculate neurons and cat dorsal lateral geniculate relay cells. Visual Neuroscience, 15, 711–729. Hochstein, S., & Shapley, R.M. (1976). Quantitative analysis of retinal ganglion cell classifications. Journal of Physiology (London), 262, 237–264. Jahnsen, H., & Llinás, R. (1984a). Electrophysiological properties of guinea-pig thalamic neurones, an in vitro study. Journal of Physiology (London), 349, 205– 226. Jahnsen, H., & Llinás, R. (1984b). Ionic basis for the electroresponsiveness and oscillatory properties of guinea-pig thalamic neurones in vitro. Journal of Physiology (London), 349, 227–247. Kaplan, E., & Shapley, R. (1984). The origin of the S (slow) potential in the mammalian lateral geniculate nucleus. Experimental Brain Research, 55, 111–116. Kaplan, E., Mukherjee, P., & Shapley, R. (1993). Information filtering in the lateral geniculate nucleus. In R. Shapley & D. M.-K. Lam (Eds.), Contrast sensitivity (pp. 183–200). Cambridge, MA: The MIT Press. Kayama, Y., Negi, T., Sugitani, M., & Iwama, K. (1982). Effects of locus coeruleus stimulation on neuronal activities of dorsal lateral geniculate nucleus and perigeniculate reticular nucleus of the rat. Neuroscience, 7, 655–666. Llinás, R., & Jahnsen, H. (1982). Electrophysiology of mammalian thalamic neurones in vitro. Nature, 297, 406–408. Lu, S.-M., Guido, W., & Sherman, S. M. (1993). The brain-stem parabrachial region controls mode of response to visual stimulation of neurons in the cat’s lateral geniculate nucleus. Visual Neuroscience, 10, 631–642. McCormick, D. A. (1992). Cellular mechanisms underlying cholinergic and noradrenergic modulation of neuronal firing mode in the cat and guinea pig dorsal lateral geniculate nucleus. Journal of Neuroscience, 12, 278–289. Merrill, E. G., & Ainsworth, A. (1972). Glass-coated platinum-plated tungsten microelectrodes. Medical & Biological Engineering, 10, 662–672. Milkman, N., Schick, G., Rossetto, M., Ratliff, F., Shapley, R., & Victor, J. (1980). A two-dimensional computer-controlled visual stimulator. Behavior Research Methods & Instrumentation, 12, 283–292.
T. OZAKI AND E. KAPLAN
Ozaki, T., Everson, R., Kaplan, E., & Purpura, K. (1996) Principal Components Analysis of the transfer of visual information through the LGN Society for Neuroscience Abstracts, 22, 639.7 Reppas, J. B., Usrey, W. M., & Reid, R. C. (2002). Saccadic eye movements modulate visual responses in the lateral geniculate nucleus. Neuron, 35, 961–974. Rogawski, M. A., & Aghajanian, G. K. (1980). Modulation of lateral geniculate neurone excitability by noradrenaline microiontophoresis or locus coeruleus stimulation. Nature, 287, 731–734. Sherman, S. M., & Guillery, R. W. (1996). The functional organization of thalamocortical relays. Journal of Neurophysiology, 76, 1367–1395. Shu, A., Hasenstaub, A., & McCormick, D. A. (2003) Turning on and off recurrent balanced cortical activity Nature, 423, 288–293. Smith, Y., Paré, D., Deschênes, M., Parent, A., & Steriade, M. (1988). Cholinergic and non-cholinergic projections from the upper brainstem core to the visual thalamus in the cat. Experimental Brain Research, 70, 166–180. Tsumoto, T., Creutzfeldt, O. D., & Legéndy, C. R. (1978). Functional organization of the corticofugal system from visual cortex to lateral geniculate nucleus in the cat. (With an Appendix on geniculo-cortical mono-synaptic connections). Experimental Brain Research, 32, 345–364. Uhlrich, D. J., Cucchiaro, J. B., & Sherman, M. S. (1988). The projection of individual axons from the parabrachial region of the brain stem to the dorsal lateral geniculate nucleus of the cat. Journal of Neuroscience, 8(12), 4565–4575. Uhlrich, D. J., Tamamaki, N., & Sherman, S. M. (1990). Brainstem control of response modes in neurons of the cat’s lateral geniculate nucleus. Proceedings of the National Academy of Sciences (USA), 87, 2560–2563. Uhlrich, D. J, Cucchiaro, J. B., Humphrey, A. L., & Sherman, M. S. (1991) Morphology and axonal projections patterns of individual neurons in the cat perigeniculate nucleus. Journal of Neurophysiology, 65, 1528–1541, 1991. Uhlrich, D. J., Tamamaki, N., Murphy, P. C., & Sherman, S. M. (1995). Effects of brain stem parabrachial activation on receptive field properties of cells in the cat’s lateral geniculate nucleus. Journal of Neurophysiology, 73(6), 2428–2447. Zhu, J. J., Lytton, W. W., Xue, J.-T., & Uhlrich, D. J. (1999). An intrinsic oscillation in interneurons of the rat lateral geniculate nucleus. Journal of Neurophysiology, 81, 702–711.