Ординатура / Офтальмология / Английские материалы / Progress in Brain Research Visual Perception, Part I Fundamentals of Vision Low and Mid-Level Processes in Perception_2006

Introduction

The eye moves in the horizontal plane under the action of two antagonist extraocular muscles: the

Corresponding author. Tel.: +34-954-349374; Fax: +34-954-349375; E-mail: jmdelgar@upo.es

lateral and medial recti. The lateral rectus muscle is innervated by motoneurons located in the pontine abducens nucleus, while the medial rectus muscle is innervated by motoneurons located in the mesencephalic oculomotor complex (Bu¨ttner-Ennever and Horn, 1997). As illustrated in Figs. 1 and 2, these extraocular motoneurons are capable of

evoking phasic firing (i.e., high-frequency bursts of action potentials lasting E100 ms) that will produce a strong muscular contraction which is able to generate a fast eye displacement — that is, a saccade or a fast phase of the vestibulo-ocular or opto-kinetic reflexes (Robinson, 1981; Moschovakis et al., 1996; Delgado-Garcı´a, 2000). This fast muscular activation is necessary to overcome the viscous drag of the orbit. In order to maintain a stable position of the eye in the orbit, extraocular motoneurons are also capable of a sustained tonic firing, necessary to counteract the restoring elastic components of orbital tissues (Robinson, 1981; Escudero et al., 1992; Fukushima et al., 1992; Moschovakis, 1997; Delgado-Garcı´a, 2000; Major and Tank, 2004). Thus, horizontal motoneurons encode the necessary velocity and position signals to rotate the eye toward the appropriate visual target and to hold the eye stable in the orbit. In fact, the firing properties of ocular motoneurons can be precisely represented by a first-order linear model (Robinson, 1981). In cats, horizontal motoneurons increase their mean firing rate by E7 spikes/s per degree of eye position, and by 1 spike/s per degree/s of eye velocity in

the pulling direction of the involved muscle (see references in Delgado-Garcı´a, 2000).

In the next few pages we will concentrate on experiments carried out by our group regarding the firing activities of prepositus hypoglossi (PH) neurons during eye movements, and on recent in vitro studies on the functional properties of reticular afferents to these neurons. More-detailed and comparative reviews regarding the integrative properties of PH neurons for the generation of eye-position signals can be found elsewhere (Robinson, 1981; Cannon and Robinson, 1987; Fukushima et al., 1992; Moschovakis et al., 1996; Moschovakis, 1997; Delgado-Garcı´a, 2000; Major and Tank, 2004).

The final common pathway for horizontal eye movements

Abducens and medial rectus motoneurons represent the final common neural pathway interposed between eye-movement-related brainstem centers and extraocular muscles in the horizontal plane. Thus, abducens and medial rectus motoneurons

must be able to translate to the lateral and medial recti muscles the precise neural motor commands corresponding to each type of eye movement (Robinson, 1981; Escudero and Delgado-Garcı´a, 1988; Fukushima et al., 1992; Bu¨ttner-Ennever and Horn, 1997; Moschovakis, 1997). As indicated above, abducens and medial rectus motoneurons occupy widely separated sites in the brainstem, but they have similar firing profiles and gain for eye-position and eye-velocity motor commands (de la Cruz et al., 1989). In this regard, it is well known that horizontal conjugate eye movements are generated in the abducens nucleus (Delgado-Garcı´a et al., 1986). Abducens internuclear neurons convey a signal similar to that present in abducens motoneurons to

the medial rectus motoneurons located in the contralateral oculomotor nucleus. That is, besides eye accommodation and vergence signals produced in midbrain regions close to the oculomotor nucleus, neural signals present in abducens motoneurons represent horizontal oculomotor commands to the corresponding extraocular muscles.

Abducens motoneurons fire a burst of action potentials slightly preceding and during eye movements in the on-direction, and decrease, or even stop, their firing during off-directed saccades. Moreover, abducens motoneurons present a tonic firing proportional to eye positions in the orbit. This tonic firing increases or decreases linearly with eye positions further in the onor

214

off-direction, respectively. Thus, abducens motoneurons are multi-stable, in the sense that they are capable of producing a stable firing for every position of the eye in the orbit (Major and Tank, 2004). Recently, we have shown that the firing variability of abducens motoneurons in alert cats during ocular positions of fixation depends on the balance between inhibitory and excitatory synaptic innervation, but also on intrinsic mechanisms capable of stabilizing motoneuron firing (see Fig. 1; Delgado-Garcı´a et al., 1989; Gonza´lez-Forero et al., 2002).Whereas, the neural origin and functional properties of oculomotor subsystems generating eye saccades and the vestibulo-ocular and opto-kinetic reflexes are relatively well known, the site and (mainly) the mechanism by which eye position is generated has always been somewhat of a mystery (Robinson, 1981; Escudero and Delgado-Garcı´a, 1988; Fukushima et al., 1992; Moschovakis, 1997; Aksay et al., 2001; Major and Tank, 2004). In an initial seminal study, Robinson (1981) proposed the presence of a common eye-position-signal neural integrator subserving all of the eye-movement subsystems. However, available experimental information suggests that there are several integrators, depending on the origin of extraocular motor commands and on the plane of movement (horizontal or vertical). For example, horizontal and vertical eyeposition signals are integrated separately in the PH nucleus and in the interstitial nucleus of Cajal (Fukushima et al., 1992; Moschovakis, 1997; Delgado-Garcı´a, 2000). Moreover, other brainstem (medial vestibular nucleus) and cerebellar (fastigial nucleus) structures seem to carry eye-position signals, at least in the cat (Escudero et al., 1992; Gruart and Delgado-Garcı´a, 1994). Integration is assumed to take place within those nuclei and/or as a result of the functional interactions they establish by their reciprocal connections and connections with other brainstem (vestibular nuclei) and cerebellar (flocculus) structures. That is, neural integration subserving eye positions of fixation could be the result of a network property.

Firing properties of prepositus hypoglossi neurons

Neurons located in the paramedian pontine reticular formation (PPRF), in particular those

called excitatory burst neurons (EBN; Fig. 2), are able to generate bursts of action potentials that encode the amplitude, peak velocity, and duration of eye saccades and fast phases of the vestibuloocular and opto-kinetic reflexes (Igusa et al., 1980; Escudero and Delgado-Garcı´a, 1988; Fukushima et al., 1992; Moschovakis et al., 1996). These neurons project monosynaptically onto abducens motoneurons and PH neurons. The question is: how do PH neurons transform the eye-velocity signals provided by EBN into eye-position ones?

Permanent or transient blockage of the functional processes taking place in the PH nucleus in both cats and monkeys supports the assumption that this structure is the site where eye-velocity signals are integrated into eye-position motor commands. It has also been suggested that eyeposition signals are the result of functional interactions established by the reciprocal connections of PH nucleus with the vestibular nuclei, the contralateral PH, and the cerebellum (Cannon and Robinson, 1987; Cheron and Godaux, 1987; Moreno-Lo´pez et al., 1996, 1998, 2001; Kaneko, 1997). Here, we will firstly consider the firing activities related to eye movements that can be recorded in the PH nucleus of alert cats, and later we will address the issue of how PH neurons generate eye-position signals.

Depending on their linear relationships with eye position and/or velocity, PH neurons can be classified as position, position-velocity and velocityposition cells (Figs. 1 and 2; Lo´pez-Barneo et al., 1982; Delgado-Garcı´a et al., 1989; Escudero et al., 1992). Pure position neurons are activated during ipsilateral eye fixations. Their mean position gain is E7 spikes/s per degree (i.e., similar to positionsignal values present in abducens motoneurons), and they present no noticeable eye-velocity signals. It has been shown in alert cats that position neurons project monosynaptically onto abducens motoneurons (Escudero and Delgado-Garcı´a, 1988; Escudero et al., 1992). Position-velocity neurons located in the PH nucleus seem to encode both eye-position and eye-velocity signals in the horizontal plane, and are activated by eye movements in the ipsilateral direction. The mean position and velocity sensitivity of these neurons are E5 spikes/s per degree and 0.6 spikes/s per degree/s

respectively, with correlation coefficients, rZ0.6 for both parameters. Position-velocity neurons project onto abducens motoneurons and to the vicinity of the oculomotor complex. Finally, velocity-position neurons present a rather irregular tonic firing, not related significantly with the position of the eye in the orbit, but they present a significant relationship with ipsior contra-lateral eye movements (0.75 spikes/s per degree/s, rZ0.6; Figs. 1 and 2). Velocity-position neurons related to ipsilateral eye movements seem to project to the peri-oculomotor area, while those related to contralateral saccades project mainly to the cerebellum (Delgado-Garcı´a et al., 1989).

The variability of interspike intervals for a similar eye position in the orbit decreases from veloc- ity-position to position-velocity and to pure position neurons in relation to the increase in eye-position signals (Figs. 1 and 2). However, it can be noticed in these two figures that the firing rate of abducens motoneurons appears to be more stable than that presented by pure position neurons. In accord to this, extraocular motoneurons have already been reported as having some intrinsic mechanisms that might contribute to a persisting and stable firing rate (Delgado-Garcı´a et al., 1989; Gonza´lez-Forero et al., 2002). Thus, the stabilizing role of the intrinsic membrane properties of extraocular motoneurons in the generation of eye-position motor commands should also be taken into account. The algebraic addition (i.e., another form of integration) of many different sources of eye-position signals upon the distal dendrites of ocular motoneurons (Baker and Spencer, 1981) could be further elaborated by the intrinsic active properties of the motoneuron membrane to produce the stable firing rate they display during eye positions of fixation (Delgado-Garcı´a, 2000).

The cascade model for the generation of eye-position signals

Using available data collected from extracellular recordings of firing activities of PH neurons during eye movements in alert cats, Delgado-Garcı´a et al. (1989) have proposed a neural circuit in cascade to

215

explain the generation of eye-position signals. In this circuit, the three neuronal types described above (velocity-position, position-velocity, and position neurons) were assumed to receive similar inputs from vestibular and reticular origins. This early proposal was modified following data obtained later (Fig. 2; Escudero et al., 1992) showing that only position neurons seem to project monosynaptically onto abducens motoneurons. Nevertheless, it has also been shown that position-velocity neurons project onto extraocular motoneurons, at least in the vertical motor system (Moschovakis et al., 1996; Moschovakis, 1997). In short, the cascade model proposes that eyevelocity signals coming from EBN (Fig. 2) terminate preferentially on velocity-position neurons. Each component of the rest of the chain (positionvelocity and position neurons) projects on both the following and the preceding one, forming an integrated loop, and these chains of neurons would be superimposed upon the shorter, direct pathways carrying eye-velocity motor commands (Escudero et al., 1992). Evidence supporting this proposal is the following: (i) The larger latency of activation of abducens motoneurons presented by positionvelocity versus position neurons (Escudero et al., 1992). (ii) The decreasing onset time with respect to the triggering saccade presented by velocityposition, position-velocity, and position neurons. Indeed, the latency to the beginning of the saccade was higher for those neurons exhibiting larger velocity signals. Peri-event time histograms of the spike activity of identified abducens motoneurons showed activation latencies in the monosynaptic range when triggered by premotor PH position neurons. In contrast, the activation of abducens motoneuron discharge by position-velocity and (mainly) velocity-position PH neurons was in the dior polysynaptic range (Escudero et al., 1992). (iii) The progressive disappearance of velocity signals, as opposed to the increasing position signal, from velocity-position to position neurons (Delgado-Garcı´a et al., 1989; Escudero et al., 1992). (iv) The presence of cascade-like, polysynaptic connections could explain the high susceptibility of eye-position neuronal systems to drugs, anesthetics, conscious state, and attention level (Delgado-Garcı´a, 2000).

216

Experiments carried out in goldfish (Pastor et al., 1994) and cats (Moschovakis et al., 1996) suggest the presence of separate integrator mechanisms to store eye-velocity signals and to generate pure eye-position ones. It is possible, with the help of selected pharmacological tools, to dissociate the two mechanisms related to neural integration in the oculomotor system (Moreno-Lo´pez et al., 1996, 1998). For example, the injection of nitric oxide synthase inhibitors into the PH nucleus in alert cats produces significant changes in eye velocity, but no noticeable effect on eye positions of fixation. Thus, the unilateral injection of the nitric oxide synthase inhibitor L-nitro-arginine methyl ester (L-NAME; Fig. 3) produces ramp-like eye movements in the contralateral direction, interrupted by fast phases directed ipsilaterally

Fig. 3. Records of the right eye position in the horizontal plane (RH, in degrees) obtained from control alert cats and following the injection of the indicated drugs in the left PH nucleus. Recordings were carried out in conditions of darkness (D) and light (L). Control recordings are illustrated at the top. Doses and time after the injection of the illustrated recordings were as follows: L-nitro-arginine methyl ester (L-NAME; an inhibitor of the nitric oxide synthase), 28 nmol, 4 (D) and 5 (L) min (middle set of records); S-nitroso-N-acetylpenicillamine (SNAP; a nitric oxide donor), 20 nmol, 3 (D) and 5 (L) min (bottom set of records). Vertical arrows indicate the direction of eye movements: l, left; r, right. (Reproduced from Moreno-Lo´pez et al. (1998), with permission of Cell Press.)

(Moreno-Lo´pez et al., 1996). In contrast, a group of PH neurons, located laterally in the so-called marginal zone close to the medial vestibular nucleus, seems to be involved in the generation of eye-position signals (Kaneko, 1997; MorenoLo´pez et al., 1998, 2001). Neurons located in the marginal zone are characterized by having a nitric- oxide-sensitive guanylyl cyclase, and the injection of nitric oxide donors in this area significantly affects the neural integrator for eye position. As indicated, the injection of the nitric oxide donor S-nitroso-N-acetylpenicillamine (SNAP; Fig. 3) in the marginal zone produces an exponential ipsilaterally directed drift of eye position.

The sustained firing reported here as a characteristic of PH position neurons seems to be necessary for the generation of a stable eye position in the orbit (Robinson, 1981; Moschovakis, 1997; Delgado-Garcı´a, 2000). In vivo studies carried out in goldfish also reported a persistent firing of selective neuronal brainstem populations related to the generation of eye-position signals (Aksay et al., 2001, 2003). It should be pointed out that the longlasting neuronal responses surpassing the duration of the triggering stimulus described here have evident relationships with the persistent firing underlying working memory and other cognitive processes (Fuster, 1997; Goldman-Rakic, 1995; Major and Tank, 2004). Two recent papers have made it possible to bridge the gap between these two apparently different neural mechanisms. Thus, Egorov et al. (2002) have reported that a muscarinic cholinergic mechanism underlies persisting firing observed in slices of rat entorhinal cortex, while Navarro-Lo´pez et al. (2004) have suggested that a similar muscarinic cholinergic action could be responsible (at least in part) for the neuronal processes that is able to generate eye-position signals. The rest of the present report will be devoted to analyzing the relationships between cholinergic mechanisms and neural integration in the PH nucleus.

In search of a synaptic mechanism for eye fixation

As shown by Aksay et al. (2001), the sustained firing rate observed in the neural integrator subserving eye position does not depend on neuronal

intrinsic properties, but has to be ascribed to the amplitude and rate of the synaptic inputs arriving at the integrator (brainstem area I, where positionrelated neurons are located in goldfish). It has also been proposed that synaptic feedback among neurons located in the brainstem area I is still necessary for temporal integration (Aksay et al., 2003). Recently, we have reported a slightly different mechanism possibly underlying the generation of eyeposition signals (Navarro-Lo´pez et al., 2004). In this case, the interaction between those PPRF neurons (mainly EBN, Igusa et al., 1980) projecting to the PH nucleus and meso-pontine reticular cholinergic neurons (Semba et al., 1990) is an additional mechanism, occurring at the synaptic level, necessary for the generation of persistent firing in PH neurons.

Experiments reported by Navarro-Lo´pez et al. (2004) were carried out in brainstem slices collected from newborn rats. As illustrated in Fig. 4, single pulses applied to the PPRF evoked graded

217

responses in identified PH neurons. If the stimulus reached threshold, prepositus neurons fired a synaptically triggered action potential presenting a characteristic biphasic appearance of the afterhyperpolarization (Fig. 4A, B). The short latency (E2.5 ms) of the evoked excitatory postsynaptic potential and its negligible variability when evoked at a high rate (up to 200 Hz) suggested its monosynaptic nature, confirming that EBN (Fig. 2) located at the rostral PPRF project directly onto prepositus neurons.

Train stimulation (100 ms) of the same reticular formation area evoked sustained depolarizations in the recorded PH neurons, exceeding the end of the stimulus by hundreds of milliseconds (Fig. 4C, D). This is indeed an example of persisting activity, evoked in this case on neurons supposedly endowed with the capacity of neural integration. Both the amplitude and the duration of the evoked sustained depolarization were linearly related with

218

train frequency (Fig. 4E, F). This latter result is very important, since firing frequency in EBN encodes saccade velocity, and the neural signal elaborated at the integrator needs to be proportional to it (Robinson, 1981; Moschovakis, 1997). On many occasions, the sustained depolarization was able to reach threshold and evoke a train of action potentials (Fig. 4C, D). Interestingly, the evoked burst of action potentials decay with a constant time, similar to that measured for orbital mechanics and to that present in abducens and medial rectus motoneurons (i.e., 120–150 ms; DelgadoGarcı´a et al., 1986; de la Cruz et al., 1989).

We have also shown that the synapse of EBN on PH neurons is glutamatergic in nature, acting probably on alpha-amino-3-hydroxy-5-methylisoxazole propionate (AMPA)/kainate receptors (Fig. 5A; Navarro-Lo´pez et al., 2004). However, it is important to indicate here that the sustained depolarization evoked by train stimulation of the PPRF was produced by a different synaptic mechanism, involving cholinergic receptors (Fig. 5B). As shown in Fig. 6, these cholinergic receptors are located

postsynaptically, i.e., on PH neurons, because the application of carbachol (a non-specific cholinergic agonist) was able to depolarize recorded neurons even in the presence of TTX (to remove all possible action potentials spontaneously present in the preparation).

Moreover, pirenzepine (a selective blocker of muscarinic M1 receptors) was able to block both the sustained depolarization evoked by train stimulation of the PPRF and the long-lasting depolarization produced by the addition of carbachol to the bathing solution. In accordance with these results, we have proposed that the interaction between AMPA/kainate and muscarinic M1 receptors underlies a plausible mechanism able to generate persistent activity in PH neurons.

It is known that extraocular muscles present a greater strength for the same length when relaxing than when contracting. For this reason, abducens motoneurons and PH neurons present a higher firing rate for a similar eye position when the position follows on-directed saccades than when it follows off-directed saccades (Delgado-Garcı´a et al. 1986,

1989). It is important to note (Fig. 6D) that when PH neurons are depolarized and repolarized in vitro, they still present the same hysteresis phenomenon (i.e., a different firing rate for the same resting potential, depending on the changing direction of membrane potential values). As shown in vivo, PH neurons presented a higher firing rate when the membrane potential was going in the depolarizing direction than when being repolarized.

From the results reported here, PH neurons recorded in an in vitro preparation presented similar functional properties (sustained firing proportional to train stimulation of the PPRF area, hysteresis, etc.) to those recorded in alert mammals (Delgado-Garcı´a et al., 1989). In order to confirm the cholinergic nature of the synaptic mechanisms involved in the generation of the persistent activity observed in PH neurons, we decided

to carry out a pharmacological study in alert cats. For this aim, we localized the PH nucleus using electrophysiological techniques (Escudero and Delgado-Garcı´a, 1988), and then carried out microinjections of selected drugs during the performance of spontaneous eye movements (NavarroLo´pez et al., 2004).

In the light of this view, the administration of non-specific (carbachol) and specific (McN-A-343) agonists of the muscarinic M1 receptors did not produce any noticeable effect on eye movements (Fig. 7B, D). In contrast, the pharmacological blockage of muscarinic receptors by atropine sulfate and, especially, by pirenzepine (a specific antagonist of the M1 receptor) produced postsaccadic, centripetal drifts of the eye in the horizontal plane (Fig. 7A, C). In the dark, agonists induced a nystagmus with slow phases directed to the

220

Fig. 7. Records of left eye position in the vertical (VP) and horizontal (HP) planes, from alert cats, following the injection of the indicated drugs in the ipsilateral PH nucleus. Recordings were carried out in conditions of darkness (gray lane) and light. Doses and time of recording after injection were as follows: atropine sulfate (A) (a non-specific antagonist of cholinergic receptors), 8 mM and 22.3 min; carbachol

(B) (a cholinergic agonist), 1.5 pM and 12.8 min; pirenzepine

(C) (a specific antagonist of M1 receptors), 0.12 pM and 4.9 min; and McN-A-343 (D) (a specific agonist of muscarinic M1 receptors), 20 pM and 4.5 min. Eye position is plotted as degrees of rotation in the horizontal plane, positive to the left (l) and up (u), and negative to the right (r) and down (d). Zero (0) indicates the central, resting position of the eye in the orbit. Curved and straight arrows indicate important peculiarities of eye movements evoked by the drugs. (Taken from NavarroLo´pez et al. (2004), with permission of the Society for Neuroscience.)

contralateral side, while cholinergic antagonists produced a nystagmus with ipsilaterally curved slow phases. Results obtained with cholinergic agonists (ramp-like displacement of the eyes in the horizontal plane to the contralateral side in darkness) are similar to the effects evoked by blockage of GABA and glycine receptors (Moreno-Lo´pez et al., 2002). The absence of effects of cholinergic agonists in the light can be explained by the fact that in this situation, and because of the activation of the visual system, there is a massive availability of acetylcholine in the PH nucleus (Chan and Galiana, 2005). In contrast, results obtained with cholinergic antagonists (curved slow phases in darkness and postsaccadic drifts in the light) suggest a loss of eye-position signals in the horizontal plane. These results are in agreement with those reported by Moreno-Lo´pez et al. (2002), following the administration of antagonists of glutamate receptors. Thus, these experiments confirm results obtained in vitro suggesting that the activation of muscarinic M1 receptors is necessary for the generation of eye-position signals (Navarro-Lo´pez et al., 2004).

The cholinergic connection

The diagrams illustrated in Figs. 8–10 attempt to summarize the results obtained by our group in a recent series of in vitro and in vivo experiments (Navarro-Lo´pez et al., 2004, 2005).

The electrical stimulation of the PPRF (i.e., of EBN; Fig. 8) by single pulses evokes a monosynaptic depolarization of PH neurons (Igusa et al., 1980). The PPRF synapse is glutamatergic in nature, acting on AMPA/kainate receptors. It has been shown (Navarro-Lo´pez et al., 2004) that the presentation of a single pulse to the PPRF area is unable to open the N-methyl-D-aspartate (NMDA) receptors also present on the membrane of PH neurons. However, in this situation, cholinergic (ACh; Fig. 8) terminals are able to modulate glutamate release by the presynaptic activation of muscarinic receptors. This presynaptic mechanism could act as a high-pass filter (Lisman, 1997), canceling out the disturbing effects of a lowrate firing of EBN.

Train stimulation of the PPRF evokes a sustained depolarization of PH neurons surpassing