The local loop of the saccadic system closes downstream of the superior colliculus

The local loop of the saccadic system closes downstream of the superior colliculus

Neuroscience 143 (2006) 319 –337 THE LOCAL LOOP OF THE SACCADIC SYSTEM CLOSES DOWNSTREAM OF THE SUPERIOR COLLICULUS R. KATO,a A. GRANTYN,a Y. DALEZIO...

2MB Sizes 1 Downloads 82 Views

Neuroscience 143 (2006) 319 –337

THE LOCAL LOOP OF THE SACCADIC SYSTEM CLOSES DOWNSTREAM OF THE SUPERIOR COLLICULUS R. KATO,a A. GRANTYN,a Y. DALEZIOSb,c AND A. K. MOSCHOVAKISb,c*

During saccades in their on-direction, extraocular motoneurons (MNs) emit bursts of discharge followed by tonic discharges proportional to the position reached by the eyes. MN bursts are generated by neural circuits, the burst generators, that once triggered run a preprogrammed course (Ron and Robinson, 1973). As proposed early on, the ability of these circuits to generate bursts and ensure saccadic accuracy is due to their configuration as feedback loop controllers (Robinson, 1973). The closed loop configuration of the burst generator is supported by considerable evidence. Firstly, some saccades (e.g. the slow ones produced by patients suffering from spinocerebellar atrophy) can be modified in mid-flight (Zee et al., 1976). Moreover, saccades pause in mid-flight in response to electrical stimulation of the pontine region containing omnipause neurons (OPNs) but resume their course and can acquire their targets after the end of the stimulation (King and Fuchs, 1977). Also, saccades executed toward a stationary target can vary considerably in terms of duration, yet their size is held within a very limited range of values. This phenomenon is common in normal subjects, and is particularly pronounced after i.v. injections of small doses of valium (Becker et al., 1981; Jürgens et al., 1981). Finally, lidocaine injection into the paramedian pontine reticular formation (PPRF) severely reduces the peak velocity of horizontal saccades, sometimes by as much as 50% (Barton et al., 2003). Their size, however, reduces by much less (10 –20%) due to proportional increases in duration, and their accuracy is thus only moderately compromised. Although there is general agreement regarding the closed loop configuration of the burst generator, the debate is still continuing as regards the signals it uses for its feedback and the location and identity of its comparator (reviewed in Sparks, 2002). Focusing on this latter issue, it is instructive to recall the model proposed by Robinson and his colleagues (van Gisbergen et al., 1981), who assumed that its feedback loop closes locally in the reticular formation, at the level of the medium lead burst (MLB) neurons (Fig. 1A). MLB neurons emit bursts encoding saccade size and they project to MNs, both directly and through the neural integrator (NI). The latter is composed of networks (subdivided into those that prefer horizontal movements and are housed in the nucleus prepositus hypoglossi (PH) (Navarro-López et al., 2004) and those that prefer vertical movements and are housed in the interstitial nucleus of Cajal (Dalezios et al., 1998)) that create the tonic eye position (E) related discharge also encountered in MNs. Additionally, the NI is thought to supply MLBs with an E feedback signal that they compare with a command signal of desired eye position (E=; equal to target position with

a

Lab. de Physiologie de la Perception et de l’Action, C.N.R.S.–College de France, Paris, France

b

Department of Basic Sciences, Faculty of Medicine, University of Crete, P.O. Box 2208, Heraklion 71003, Crete, Greece

c

Institute of Applied and Computational Mathematics, FORTH, Heraklion, Greece

Abstract—Models of the saccadic system differ in several respects including the signals fed back to their comparators, as well as the location and identity of the units that could serve as comparators. Some models place the comparator in the superior colliculus while others assign this role to the reticular formation. To test the plausibility of reticular models we stimulated electrically efferent fibers of the superior colliculus (SC) of alert cats along their course through the pons, in the predorsal bundle (PDB). Our data demonstrate that electrical stimulation of the PDB evokes saccades, even with stimuli of relatively low frequency (100 Hz), which are often accompanied by slow drifts. The velocity and latency of saccades are influenced by the intensity and frequency of stimulation while their amplitude depends on the intensity of stimulation and the initial position of the eyes. The dynamics of evoked saccades are comparable to those of natural, selfgenerated saccades of the cat and to those evoked in response to the electrical stimulation of the SC. We also show that PDB-evoked saccades are not abolished by lesions of the SC and that therefore antidromic activation of the SC is not needed for their generation. Our data clearly demonstrate that the burst generator of the horizontal saccadic system is located downstream of the SC. If it is configured as a local loop controller, as assumed by most models of the saccadic system, our data also demonstrate that its comparator is located beyond the decussation of SC efferent fibers, in the pons. © 2006 IBRO. Published by Elsevier Ltd. All rights reserved. Key words: predorsal bundle, electrical stimulation, cat, burst generator, PPRF, chemical inactivation.

Saccades are stereotypical rapid eye movements generated by the coordinated activation of many cell classes distributed throughout the brain (Moschovakis et al., 1996). *Correspondence to: A. K. Moschovakis, Department of Basic Sciences, Faculty of Medicine, University of Crete, P.O. Box 2208, Heraklion 71003, Crete, Greece. Tel: ⫹30-2810-394509; fax: ⫹30-2810-394530. E-mail address: [email protected] (A. K. Moschovakis). Abbreviations: E, eye position; EI, efficacy index; EPe, eye position error; E=, desired eye position; ⌬E, eye displacement; FOR, fastigial oculomotor region; IR, incidence ratio; LLB, long lead burst; Me, motor error; MLB, medium lead burst; MN, motoneuron; NI, neural integrator; NRTP, nucleus reticularis tegmenti pontis; OPN, omnipause neuron; PDB, predorsal bundle; PH, prepositus hypoglossi; PPRF, paramedian pontine reticular formation; RI, resettable integrator; RPo, nucleus reticularis pontis oralis; SC, superior colliculus; T, current threshold.

0306-4522/06$30.00⫹0.00 © 2006 IBRO. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.neuroscience.2006.07.016

319

320

R. Kato et al. / Neuroscience 143 (2006) 319 –337

Fig. 1. Schematic illustration of models of the local loop of the saccadic system. (A) Robinson’s bang-bang model. (B) Trans-collicular model. Solid and dashed lines symbolize excitatory and inhibitory connections, respectively. Dots indicate a series of intervening transformation steps. See text for further explanations.

respect to the head) supplied by higher order structures. The instantaneous difference between E= and E is the motor error signal (Me) encountered in MLB neurons. This drives the eyes until E matches E=, whereupon Me is reduced to zero, MLB firing stops and the eyes land on target. Neither the intensity nor the duration of MLB bursts needs to be preprogrammed in this model. Instead, they are automatically adjusted for the desired saccade size through the operation of the feedback loop. Although other models of the saccadic system rely on feedback signals of eye displacement (⌬E) to adjust the size of MLB bursts and they use long lead burst (LLB) neurons rather than MLB neurons as comparators (Scudder, 1988; Moschovakis, 1994), they likewise assume that the feedback loop closes locally in the reticular formation. Fig. 1B illustrates an alternative model, which includes the superior colliculus (SC) in the feedback loop of the saccadic system. It was motivated by the observation that the decline of the instantaneous firing frequency curve of a particular class of collicular saccade-related burst neurons (“clipped” cells) is tightly related to the dynamic Me during saccades (Waitzman et al., 1991). This model assumes that the input to the saccadic system is a tonic signal that specifies the eye position error (EPe, equal to the distance between initial and E=). During a saccade, a corollary discharge of instantaneous distance from the initial eye position (⌬E) provided by a resettable integrator (RI) is subtracted from the EPe signal and the result is conveyed by “clipped” cells to their target neurons in the midbrain and the pons. When ⌬E equals EPe, firing rate becomes zero, and the saccade terminates. Evidence in support of this scheme was obtained with the help of electrical stimulation of the OPN area. As predicted by the trans-collicular model, this prolongs the duration of the discharge of saccade related SC neurons till after the end of the stimulation and the acquisition of the target (Keller and Edelman, 1994). This configuration was maintained in a number of more recent models (e.g., Lefèvre and Galiana, 1992; Arai et al., 1994; Grossberg et al., 1997; Arai and Keller, 2005). Such single loop trans-collicular models sug-

gest that the circuitry downstream of the SC would generate normal saccades even if it operated in an open loop configuration (Das et al., 1995). Implicitly, this means that all feedback signals necessary for saccade generation pass through the SC, and any local feedback loop in the lower brainstem is redundant or even nonexistent. It should be possible to decide whether a saccade generating local loop exists in the pons by studying eye movements evoked in response to electrical stimulation of brain stem sites lying downstream of the putative comparator in the SC. If the comparator of the saccadic system is downstream of the SC, eye movements evoked by electrical stimulation of SC efferents should display the kinematic properties of normal saccades. If, on the other hand, a pontine comparator does not exist, electrical stimulation anywhere downstream of the SC should cause the eyes to move for the duration of the stimulus. Further, eye movements evoked in response to stimuli delivered as frequency steps, should be dominated by the dynamics of the oculomotor plant and of structures intervening between the point of stimulation and the plant (such as the NI). Their time course should therefore resemble ramps and/or exponential slow drifts rather than saccades. To test the existence of a postcollicular comparator we electrically stimulated the major tectofugal fiber bundle, the predorsal bundle (PDB in Fig. 1), which conveys the output of the SC to the PPRF. Electrical stimulation of the pontine tegmentum has been long known to evoke eye movements (Bender, 1964) but, surprisingly, there is no conclusive evidence that saccades can be evoked from it including its most medial aspect, i.e. the PPRF (Cohen and Komatsuzaki, 1972). Our results demonstrate that electrical stimulation of the PDB can generate saccades whose dynamics are well in the range of normal saccades in the cat and remain so after chemical inactivation or lesion of the SC. A preliminary version of our results has been presented in an abstract (Moschovakis et al., 2005).

EXPERIMENTAL PROCEDURES We used four purpose-bred adult cats (code named E, F, G and PU), weighing 3– 4 kg, complying with European Union regulations on biosafety and the use of live animals in research (directive 86/609), as well as with the “Principles of laboratory animal care” (NIH publication No. 86 –23, revised 1985) and were approved by the institutional ethics committee. Ethical considerations obliged us to keep to a minimum the number of animals we employed and their suffering. They were surgically prepared for chronic recording and stimulation experiments under pentobarbital anesthesia (induction with 40 mg/kg, i.p., supplemented with 5 mg/kg/h, i.v.) and sterile conditions. Stainless steel bolts were cemented to the skull for painless head fixation, and three turns of Teflon insulated stainless steel wire were passed under the insertions of the extraocular muscles to form a search coil. For craniotomy, the skull was exposed from the bregma to the occipital crest and the center of the desired opening was marked on the occipital bone at the stereotaxic coordinate P13–P16, depending on the size of the skull. Using this mark we drew a circular contour (diameter 12 mm) around which the bone was removed with a dental drill paying particular attention to avoid any local damage to the dura. After verifying that the 10 mm chamber fits the opening, the dura was excised along its perimeter. Using a specially designed carrier the chamber was gently advanced and placed in touch with

R. Kato et al. / Neuroscience 143 (2006) 319 –337 the surface of the cerebellar vermis. The chamber was oriented in rostro-ventral direction, with an inclination of 27° relative to the stereotaxic coronal plane. When extrapolated to the dorso-ventral level of the rhombencephalon, its axis crossed the brainstem surface at the stereotaxic coordinate P3–P4, thus permitting the exploration of the pons from the abducens nucleus caudally (P6) to the pontomesencephalic junction rostrally. The narrow space between the chamber’s outer wall and the edge of the bone was then closed with small pieces of the Gelfoam before applying a small quantity of the fast-curing dental cement (Unifast, GC Corporation, Tokyo, Japan) all around the base of the outer wall of the chamber. The outside of the chamber (10 mm in height) was completely embedded from the bottom to the top with a slowcuring dental cement (Resivy, Prodont-Holliger S.A., Vence, France) which produces less heat during polymerization. After the cement became hard, the carrier head, which occupied the inner space of the chamber, was removed, the cerebellar surface and the inner walls of the chamber were thoroughly rinsed with sterile Ringer, followed by 1% hydrogen peroxide, and the chamber was filled with a gel containing dexamethasone and antibiotic (neomycin) and capped. In four experiments in one cat (cat E) we recorded responses to PDB stimulation after chemical inactivation and/or lesion of the SC. To gain access to the superior colliculi, openings were made bilaterally in the parietal bones at a later stage of the chronic experiment, under ketamine (20 mg/kg, i.m.) anesthesia. Between experiments, the openings were hermetically closed with bi-layer paraffin/plastic sheets after applying the dexamethasone–neomycin gel. The sheets were fixed by exerting pressure on the lateral edges of the upward facing plastic with appropriately oriented screws. Due to its plasticity, paraffin filled the free space created by the craniotomy, thus preventing the swelling of the cortex and the leakage of the liquor. When needed, and for two days after recovery from the anesthesia, postoperative analgesia was provided by SC injections of 2 mg/kg ketoprophen (Populin and Yin, 1998). Experiments started 8 –10 days after the surgery. During experimental sessions body movements were restrained with a cloth bag and the head was rigidly attached to the stereotaxic frame by means of the implanted bolts. Instantaneous horizontal and vertical E was recorded using the search coil method. The device we used (Angle-Meter, Primelec, D. Florin, Switzerland) is precalibrated for a wide range of eye coil parameters and gives linear measures of the E over 360° in azimuth. Our three-turn coils were in the specified range and the precision of the calibration was verified before the beginning of the experiments on each animal. The straight ahead E was calibrated by averaging samples of instantaneous E obtained while the animal was tracking an object moved by a computer-controlled mechanical device, either vertically in the midsagittal plane (horizontal zero) or horizontally in the plane passing through the center of the eyes (vertical zero). Experiments were done every second day and their duration was limited to 4 –5 h. The animals were not trained for making saccades to specific targets. During stimulation trials, as well as during recording of “spontaneous” saccades we attracted the cat’s gaze to different locations by showing a variety of frequently exchanged toys, often producing noise as well. These objects were presented manually as close as possible to the horizontal meridian of the animal’s field of view without any additional control of the vertical E. The novelty of the targets we used permitted us to keep cats active during recording. To avoid habituation and fatigue, cats were allowed to rest at 30 – 45 min intervals and were fed with semi-liquid food during such pauses. We shall use the terms “spontaneous” or “self-generated” saccades without implying anything other than the fact that they were evoked neither by electrical stimulation nor as a consequence of operant conditioning of the animals. The velocity-amplitude characteristics of such saccades were analyzed from records 3–5 min long which provided samples of over 100 (110 –260) saccades.

321

To stimulate the PDB, we applied 0.5 ms monopolar negative pulses in trains of 300 – 850 ms durations and frequencies of 100, 200 or 330 Hz through glass micro-electrodes (2.5 ␮m in diameter) filled with a 3.8 M NaCl solution. In two cats tungsten microelectrodes (Frederick Haer & Co., Bowdoin, ME, USA) were used as well. For each stimulation frequency, we defined current threshold (T) as the lowest intensity at which movements were evoked by about one third of the total number of stimulus trains. At the highest frequency we used T varied between 4 and 35 ␮A (mean⫾S.D.: 12.9⫾5.6 ␮A, median: 10.0 ␮A). Since responses at intensities ⬍5 ␮A were exceptional, we routinely started stimulating each site by setting the intensity at 5 ␮A, and then increased it in steps of 5 ␮A. To avoid excessive stimulation we did not attempt a true threshold straddling with the use of still smaller steps. If no response was evoked at 40 ␮A, the site was classified as negative (no response). Stimulus trains were applied at intervals of 5 s for about 3–10 min, thus allowing us to collect data from 35–120 trials at each site, with a particular combination of stimulus parameters. During each block of trials we attracted the cat’s gaze to several ipsi- and contralateral locations so that evoked movements would start from a wide range of initial Es. We used stereotaxic coordinates to guide stimulation electrodes toward the brain stem through the cerebellum. In each cat, we first recorded the discharge of single MNs to map the location of the abducens nucleus. In addition, we recorded the activity of OPNs to verify the location of the midline. The location of stimulation sites relative to anatomical landmarks was verified histologically. For this, we used biocytin-filled micropipettes (4% solution in 0.75 M NaCl buffered by 0.05 M Tris) in the last experiment in each cat. Biocytin was iontophoresed (⫺5 ␮A, pulse duration 0.6 s, repetition rate 1 Hz) at four or five points, 0.5 mm from the estimated midline and separated rostro-caudally by 0.8 –1.0 mm. All injections were made at the same depth with respect to the stereotaxic reference point. Biocytin pipettes followed the trajectory of some of the tracks explored during the stimulation experiments. The biocytin reaction product was visualized in 50 ␮m serial sections, which were cut parallel to the electrode tracks and processed using conventional procedures (e.g., Moschovakis et al., 1998b). Biocytin deposits had dense cores, 0.2– 0.3 mm in diameter. The 3-D coordinates of their centers were used to correct the stereotaxically estimated medio-lateral position and the depth of stimulation points. In this report we shall present the data collected in the rostral half of the pontine tegmentum, roughly corresponding to the antero-posterior extents of the nuclei reticularis pontis oralis (RPo) and reticularis tegmenti pontis (NRTP). Evoked eye movements were sampled at a rate of 500 Hz. Their horizontal components were analyzed off-line using the Spike2 (Cambridge Electronic Design Ltd., Cambridge, UK) software. The onset of the stimulus trains as well as the onset and offset of saccades was marked as described before (e.g., Grantyn et al., 1996). The onset of horizontal saccade components was determined by velocity threshold 20°/s. The threshold was decreased to 15°/s for saccades smaller than 3° and correspondingly low peak velocities (in the range 20 – 45°/s). Unless they were followed by fast (⬎20°/s) postsaccadic drifts, we used the same velocity threshold to determine the end of saccades. In the former cases, we used the discontinuity of velocity traces at the transition between the rapid deceleration phase of saccades and the slow deceleration during drifts. In four experiments in one cat (cat E) we recorded responses to PDB stimulation after chemical inactivation and/or lesion of the contralateral SC. Injections of lidocaine (5% in 0.9% NaCl) were made in three experiments separated from each other by one week. Muscimol (0.3% in 0.9% NaCl) was injected in the fourth experiment. The injection barrel of three-barreled glass pipettes was connected to a 10-␮l syringe through a needle embedded in its shaft. An adjacent barrel contained an electrolytically thinned tungsten wire. Its 50 ␮m diameter end protruded beyond the glass

322

R. Kato et al. / Neuroscience 143 (2006) 319 –337

pipette for 150 ␮m. A sharp tip was formed by additional etching after fixing and insulating the wire inside the pipette. The third barrel was added to facilitate manufacture of straight, rigid shanks with parallel walls over lengths of 20 –25 mm. It was filled with isotonic NaCl and was not used in our experiments. The overall diameter of our pipettes was 0.2– 0.25 mm. To lower a pipette toward the SC the back end of its embedded needle was fixed to the syringe carried by a micromanipulator. Tungsten electrodes were used to detect the collicular surface by monitoring visually evoked multiunit activity, to search minimal threshold depth for stimulation-induced saccades and to evaluate the effectiveness of the injection. We injected drug solutions by pressure in large volumes (3.5–5 ␮l, see Table 2) and over relatively short injection times (2– 4 min). The first step of the experimental protocol consisted in recording spontaneous eye movements. During the second step we searched for PDB sites suitable for control records of evoked saccades. This search was guided by PDB mapping during the preceding eight experiments on the same animal, so that one track was enough to find the bundle. After recording control responses to stimulation of a selected PDB site, the electrode was left in the same position till the end of the experiment. Similarly, the positioning of injection pipette in the SC during the third step of the protocol was guided by mapping during preceding experiments and by the detection of visual activity in the superficial layers. Mapping in depth of T for eliciting saccades was not attempted because we expected the threshold to increase due to the leakage of drugs. We fixed the pipettes at the depth at which saccade thresholds fell from 30 to ⬍20 ␮A. After recording responses to SC stimulation we made the injections, taking their end as time-zero for subsequent post-injection records. In experiments with lidocaine (Exp. 1L–3L), test records had the following sequence: 0 –30 s-stimulation of the SC; 1-3-4.5 min-spontaneous eye movements; 3-5-9-10 min-stimulation of the PDB. In the experiment with muscimol (Exp. 4M), known for its prolonged effect (Hikosaka and Wurtz, 1985; Martin and Ghez, 1999), spontaneous movements were recorded at 5– 6.5 min and responses to PDB stimulation at 7–14 min. Similar sequences of test records were repeated also at later times (up to 30 – 60 min). Because of the short-lasting effects of lidocaine (Tehovnik and Sommer, 1997; Malpeli, 1999; Martin and Ghez, 1999) the present report is based on the earliest tests, as specified above. Additional information on the time course of recovery is provided in the Results section.

RESULTS Fig. 2 provides an overview of the region we explored with the aim to stimulate collicular axons descending in the PDB. Because stimulation near the midline might activate OPNs present throughout the caudal pontine reticular nucleus we limited the analysis to stimulus sites located in either side of the rostral pons, at the level of RPo (2.0 – 5.0 mm anterior to the abducens nucleus). Here we describe the effects of stimulation applied to points inside or on the borders of the PDB (n⫽64), as delineated by the presence of its compact longitudinal fascicles. Responses from points located in the reticular core, 1.3–2.0 mm from the midline (PPRF points) were analyzed as a separate group. Deeper points, beginning with the dorsal border of the NRTP were discarded. We also discarded points inside the dorsal extension of the NRTP, which could have activated its dispersed cell groups invading the paramedian tracts.

Movements evoked by PDB stimulation Movements evoked from PDB sites were always ipsiversive and usually consisted of a combination of fast and slow components. Fig. 3 illustrates examples of the horizontal component of eye movements evoked in alert cats in response to the electrical stimulation of one PDB site. Its location, at the rostral limit of the pons, is marked with an asterisk in Fig. 2 (plot PU 4.3). The T was 8 ␮A at 330 Hz, it increased to 15 ␮A at 200 Hz and to 20 ␮A at 100 Hz. Saccadic components were clearly identifiable at all three frequencies and they always terminated long before the end of the stimulus train. Whenever E remained stationary after saccades (Fig. 3A), we could use the 20°/s velocity threshold to determine unambiguously saccade onset and offset. In such cases, saccade amplitude could be exactly measured and used to calculate the velocity–amplitude relationship. Evoked movements changed qualitatively with increasing pulse rate; low velocity ramp- or exponential-like movements lasting till the end of the stimulus train often followed saccades. Such movements resemble slow drifts evoked in response to the electrical stimulation of the SC (Grantyn et al., 1996). They have been attributed to relatively direct projections from the SC to extraocular MNs that largely bypass the burst generators (see (Grantyn et al., 1996) and (Moschovakis et al., 1998a) for a list of fibers and neurons likely to mediate them). In the example shown in Fig. 3, postsaccadic slow movements (drifts) appeared first at 200 Hz (Fig. 3D, E) and their amplitude increased at 330 Hz (Fig. 3G, H). In such cases, we could not always use the same 20°/s velocity threshold to determine the end of saccadic components. Instead, we used the transition (inflection) from rapid (saccade offset) to slow (drift onset) deceleration on the velocity trace. In the absence of such inflections, the main sequence could not be determined. Such movements were considered not to conform to the kinematic characteristics of saccades. On the basis of this criterion, four PDB sites were judged not to evoke saccades. Sites evoking movements with saccade-like velocity profiles were subjected to a further selection taking into account the relationship between the peak velocity and the amplitude of their horizontal components. These had to obey the main sequence relationship with minimal slopes defined individually from the spontaneous saccades of each cat. The values calculated from samples of spontaneous saccades were equal to 9.6, 7.9, 10.8 and 14.1°/s/° in cats E, G, F and PU, respectively. For evoked saccades we accepted values as small as 5.2 (cat E), 5.3 (cat G), 6.1 (cat F) and 7.7 (cat PU) °/s/°, which correspond to the slope of the main sequence curve of self-generated saccades with the slowest horizontal velocity (20% of the total sample). Regression slopes of evoked saccades depended on stimulus intensity and frequency. Increasing the stimulus intensity (by 1.25 to two times) while keeping the frequency constant (330 Hz) resulted in steeper slopes (two-tailed paired t-test: t⫽5.88; df⫽14; P⬍0.0001; mean difference⫽3.28°/s/°). The effect of changing the frequency of stimulation is illustrated for one PDB site from

R. Kato et al. / Neuroscience 143 (2006) 319 –337

323

Fig. 2. Location of stimulation sites on the paramedian portions of coronal sections through the brain stem of cats PU, F and E. Distance (in mm) from the rostral pole of the abducens nucleus is indicated at the bottom left of each drawing. Sites of cat G are entered on corresponding antero-posterior levels of cats PU and E. F 4.3– 4.6 is a superposition of sites located at rostrocaudal levels separated by 0.3 mm. The rostralmost level (4.6) corresponds to the caudal end of the decussation of the superior cerebellar peduncle and is about 2.2 mm behind the caudal end of Meynert’s decussation. Thin solid lines delimit pontine nuclei and fiber bundles. Thick solid lines indicate the dorsal and ventral borders of the brain stem. The centers of filled circles correspond to the location of stimulation sites and their diameters are proportional to the efficacy of the sites in evoking saccades (EI) as indicated in the inset. Open circles correspond to ineffective sites. Anatomical scale bar applies to all drawings. Abbreviations: CS, central superior nucleus; DR, dorsal raphe nucleus; PG, pontine gray; PVG, periventricular gray; Py, pyramidal tract; VTN, ventral tegmental nucleus (of Gudden).

cat PU in Fig. 3. The slope of the main sequence curve was equal to 8.1°/s/° at the stimulation frequency of 100 Hz (Fig. 3C) and it increased to 10.3 at 200 Hz (Fig. 3F) and 13.7°/s/° at 330 Hz (Fig. 3I). These values satisfied well the

criterion of the minimal main sequence slope adopted for cat PU (7.7°/s/°). Stimulus sites at which fast movements did not attain this main sequence criterion were considered as “non-saccadic.” On the basis of this criterion, another

324

R. Kato et al. / Neuroscience 143 (2006) 319 –337

Fig. 3. Horizontal components of eye movements evoked in response to the electrical stimulation of the PDB. (A, D, G) Superpositions of raw records of instantaneous E from individual trials at fixed combinations of stimulus parameters indicated above the bars delimiting the stimulus train. Vertical dashed lines mark the beginning of the stimulus train. (B, E, H) Instantaneous velocity traces for the movements shown in A, D and G, respectively. Rightward movements after the end of the stimulus train are truncated. Other conventions are as in A, D and G. (C, F, I) Plots of absolute peak horizontal velocity (ordinate) versus absolute horizontal amplitude (abscissa) of saccades illustrated in the top row of the same column. The solid lines are the least square linear regression lines through the data and obey the equations displayed. Dashed lines indicate the 95% confidence intervals.

two PDB sites were judged not to evoke saccades. Overall, the slopes of the main sequence curves of saccades evoked with optimal combinations of stimulus intensity and pulse rate ranged from 5.9 –19.8°/s/° (mean⫾S.D.: 11.9⫾3.0°/s/°) and values greater than 9°/s/° were observed in 80% (46/58) of the effective PDB stimulation sites. Thus, the majority of our PDB sites evoked saccades with main sequence slopes in the range previously reported for spontaneous saccades in the cat (⬎9°/s/°; Evinger and Fuchs, 1978; Guitton et al., 1984). Frequency dependent increases of the slope of the main sequence curve such as those illustrated in Fig. 3 were significant in all sites of cat PU and other cats from which enough data were collected to evaluate them statistically (e.g. paired t-test: t⫽3.03, df⫽14, P⫽0.009, mean difference⫽1.74°/s/°, when the frequency was increased from 100 to 330 Hz). Having shown that stimulation of the PDB evokes saccades we wished to compare them with saccades evoked

from the SC. To this end we used data obtained from the PDB for the purposes of this report (15 sites in cats E and PU) as well as data obtained from the SC in cat E and in previous experiments (22 sites in nine animals) using identical stimulation parameters (200 Hz and 2⫻T). Table 1 provides a summary description of the main sequence curves of the saccades evoked from all these sites. As shown here, the sample of saccades evoked from the PDB is well matched to that of saccades evoked from the SC in that the size of the characteristic vectors was equal to 8.87° (S.D.: 2.92) when averaged across PDB sites and 7.70° (S.D.: 3.26) across SC sites. Table 1 also documents the only noticeable difference we found between PDB- and SC-evoked saccades, namely that the slope of the main sequence curve of saccades evoked from the SC (mean⫾S.D.: 13.2⫾2.70°/s/°) was, on the average, steeper (P⬍0.05, Student’s t-test) than that of saccades evoked from the PDB (mean⫾S.D.: 11.3⫾2.3°/s/°). Other-

R. Kato et al. / Neuroscience 143 (2006) 319 –337 Table 1. Comparison of saccades evoked in response to stimulation of the SC and of the PDB with stimulation at pulse rate 200 Hz and relative intensity 2⫻T Subject

Site

SH

MS Slope

r2

MS Y-int

Staircases (%)

1-inj 1-inj 1-inj 1-inj 1-inj 1 1 2-inj 1-3 1-4 1-5 2-1 12-1-3 12-1-4 12-1-5 13-1-1 13-1-3 14-1-1 16-1-1 16-1-3 18-1-1 18-1-3

6.2 13.5 1.3 6.7 4.8 4.5 8.3 9.6 2.7 3.4 6.9 3.9 8.2 8.1 7.9 10.3 11.4 9.4 11.5 12.0 7.9 10.4

13.9 15.4 12.3 11.3 14.6 14.4 14.7 16.0 21.5 15.3 14.4 15.3 11.5 10.2 9.4 10.7 11.4 12.4 11.5 12.0 10.3 12.2

0.99 0.98 0.99 0.97 1.00 0.92 0.96 0.98 0.86 0.87 0.85 0.95 0.92 0.94 0.85 0.88 0.94 0.96 0.93 0.98 0.91 0.96

8.0 15.2 3.8 12.7 8.5 22.9 11.2 31.4 2.7 25.3 35.2 17.2 0.0 13.6 31.8 7.1 3.9 5.6 14.1 16.1 30.6 16.6

75 24 1 63 2 14 0 — 31 57 19 67 0 5 0 0 5 0 15 19 13 34

4-3-3 4-3-5 4-3-6 5-2-7 6-1-5 6-1-6 6-1-7 6-2-6 9-1-4 9-1-5 9-1-6 10-1-4 11-1-3 14-1-3 8-1-5

8.5 10.7 11.3 12.8 10.5 11.6 9.9 11.6 5.4 6.6 7.1 4.8 4.2 6.0 12.0

15.1 11.9 11.2 12.7 13.5 12.8 11.4 11.2 8.2 9.4 12.5 6.5 9.0 9.9 13.7

0.97 0.92 0.93 0.98 0.96 0.98 0.98 0.97 0.90 0.93 0.95 0.90 0.89 0.96 0.98

0.1 24.2 0.6 4.4 24.4 2.7 9.1 35.6 21.9 15.7 25.3 22.5 20.1 13.5 4.5

0 0 0 0 0 0 0 30 0 0 0 30 0 0 0

SC NB2 NB3 NB4 NB5 NB7 STAN BN2 BN2 BN4 BN4 BN4 BN4 E E E E E E E E E E PDB E E E E E E E E E E E E E E PU

Columns contain the following data (from left to right): 1, code name of experimental animals; 2, code names of stimulation sites; 3, characteristic vector (SH) of the horizontal component of saccades; 4, slope of the main sequence curve (MS-Slope, °/s/°); 5, goodness of fit (r 2) of the linear regression describing the main sequence; 6, intercept of the main sequence curve (MS Y-int, deg); 7, percentage of saccadic staircases (Staircases (%)) encountered in stimulation trials evoking saccades.

wise, the main sequence curves of PDB- and SC-evoked saccades were similar. The intercepts of those obeyed by SC-evoked saccades (mean⫾S.D.: 15.2⫾10.40°/s) did not differ significantly (P⬎0.8, Student’s t-test) from those obeyed by PDB-evoked saccades (mean⫾S.D.: 15⫾10.92°/s) and both accounted equally well for the variance (r 2) of the dependent variable (SC: 0.88⫾0.09; PDB: 0.9⫾0.06).

325

Measurements of latency also provide a criterion for comparing PDB- and SC-evoked saccades. It should be noted that the latency of PDB evoked saccades depended on the intensity and frequency of stimulation. For example, the mean latency dropped by 32 ms (paired t-test: P⬍0.0001, t⫽7.09, df⫽14) when the intensity of 300 Hz stimuli was augmented by factors ranging ⫻1.25 to ⫻2.0. Similar drops of latency accompanied increments of intensity tested at 100 Hz and 200 Hz. They also accompanied increments of frequency; for example, mean latency dropped by 43.8 ms (paired t-test: P⬍0.0001, t⫽9.48, df⫽14) when the frequency of stimuli of the same intensities (1.15–2.0⫻T) was raised from 100 to 330 Hz. When saccades evoked in response to identical stimulation parameters are considered (frequency: 200 Hz, intensity: 2⫻T), the mean latency of PDB-evoked saccades was statistically indistinguishable from the one we previously reported for SC-evoked saccades (mean⫾S.D.: 50⫾19 ms; Grantyn et al., 1996). As with SC-evoked ones, saccades evoked from the PDB were often oblique probably due to the activation of the ventral ascending (Av) branches of tectofugal axons which bifurcate before and, less frequently, after crossing to the opposite side in Meynert’s decussation (Grantyn and Grantyn, 1982; Moschovakis and Karabelas, 1985; Moschovakis et al., 1988a,b). Rostrally, these branches reach the fields of Forel, including its medial portion that is homologous to the rostral interstitial nucleus of the medial longitudinal fasciculus (riMLF) of primates (Harting et al., 1980). This nucleus houses both upward and downward pre-MNs (Moschovakis et al., 1991a,b) in both sides of the brain. Because we could not ensure the selective activation of axons whose Av collaterals project to one or the other antagonistic group, we did not expect that our PDB stimulation would evoke saccades with an obvious vertical directional preference. Indeed, vertical components satisfying our saccade identification criteria were observed in only 22 of 58 stimulation sites. Their characteristic vectors were small (range: 0.01– 4.6°; mean⫾S.D.: 1.6⫾1.19°; n⫽22) and inconsistent, tending to drive the eyes to vertical zero. Accordingly, we did not feel that stimulation of the PDB is suitable for evaluating the contribution of postcollicular structures to the generation of vertical saccade components, and we thus did not study them in any quantitative detail. Electrical stimulation of the SC with long trains of pulses is known to evoke sequences of similar saccades which follow closely the one after the other and are for this reason called saccadic staircases (Robinson, 1972). Fig. 4 demonstrates that staircases of saccades can be evoked in response to stimulation of the PDB as well. However, it was more difficult to evoke such staircases from the PDB than it is to evoke them from the SC. Table 1 includes information about the frequency with which they were observed in response to stimulation of the same 15 PDB sites and 21 of the 22 SC sites tested, expressed in terms of the number of trains evoking staircases over the total number of trains that evoked primary saccades. As shown here, a significant percentage of staircases (⬎10%) could be evoked from about half of the

326

R. Kato et al. / Neuroscience 143 (2006) 319 –337

Fig. 4. Staircases of saccades evoked in response to the electrical stimulation of the SC (A, B) and the PDB (C, D) of cat E. Layout and conventions as in Fig. 3.

SC sites tested (12/21) but only from two of the 15 PDB sites. The lower efficacy of the PDB stimulation was also confirmed by counts in records from another 39 PDB sites, tested with various combinations of stimulation parameters. In this sample, seven sites were effective in inducing saccadic staircases with probabilities of 12–57%. Movements evoked by PPRF stimulation Although we did not attempt a systematic analysis of the oculomotor responses evoked from the reticular core, we collected data from 28 PPRF sites separated from the paramedian tracks or from the dorsal border of the NRTP by at least 0.5 mm (Fig. 2) along 10 tracks we made at the same antero-posterior levels as those used to explore the effects of PDB stimulation, and at distances 1.3–2.1 mm

from the midline. No saccade-like movements could be evoked from 18 of these 28 PPRF sites. As shown in Fig. 5C–F, eye movements evoked from these sites lasted for the duration of the stimulus train. The time course of E followed ramp-like (E, F) or exponential (C, D) trajectories. In the case of exponential movements, peak velocities could reach values as high as 50°/s but ramp-like movements rarely reached 20°/s. In no such case was it possible to detect a discontinuity during the decremental phase of the velocity profiles. Discontinuities were sometimes seen in the responses obtained from the remaining 10 reticular sites and were analyzed to reveal the presence of saccadic components. At a stimulation frequency of 330 Hz, saccade-like velocity profiles (e.g. Fig. 5A, B) were observed in four sites only (all in cat PU). In two cases, the

R. Kato et al. / Neuroscience 143 (2006) 319 –337

327

Fig. 5. Horizontal components of eye movements evoked in response to the electrical stimulation of the PPRF in cat PU. (A, C, E) Superpositions of raw records of instantaneous E from individual trials at fixed combinations of stimulus parameters indicated above the bars delimiting the stimulus train. Vertical dashed lines mark the beginning of the stimulus train. (B, D, F) Instantaneous velocity traces for the movements shown in A, C and E, respectively. Movements after the end of the stimulus train are truncated. Other conventions are as in A, C and E. PPRF saccades illustrated in A, B obey the main sequence curve Vmax⫽27.2⫹8⌬E (r 2⫽0.83).

incidence ratios (IR, see below for definition) and hence the number of such responses were too low for a quantitative analysis (n⫽7 and n⫽10 at a stimulus intensity of 30 and 20 ␮A, respectively). Movements with saccade-like velocity profiles were evoked, somewhat more frequently (n⫽18 and n⫽15 at 25 ␮A), from the other two sites. The slopes of the regression lines relating peak velocity to amplitude were 6.74 and 8.36°/s/°, only the latter (examples of which are shown in Fig. 5A, B) passing the criterion adopted in cat PU (7.7°/s/°). The amplitudes of saccades evoked from these two sites depended on initial E (kH⫽0.39, 0.40; see Eq. 1 below for definition) and had characteristic vectors equal to 2.25 and 4.71°. Because all four of the reticular sites from which we could evoke saccadelike movements were separated from the PDB by only 0.5– 0.6 mm and because test currents ranged from 20 to 30 ␮A, stimulus spread to the PDB cannot be excluded. At a stimulation frequency of 100 Hz, movements with saccade-like velocity profiles were evoked from six PPRF sites (four in cat G and two in cat PU). IR were low (IR⫽0.22– 0.59), so that regression analysis could be applied to the data from four sites only. The slopes of their main sequence curves were shallow (5.82– 6.61°/s/°) and passed the criterion in cat G (two sites) but not in cat PU. In addition, responses evoked from these four reticular sites were characterized by high thresholds (25–35 ␮A) and

small characteristic vectors (0.44 –2.35°). As with the sites tested at 330 Hz, and because of the proximity to the PDB and the intensity of the current used (30 – 40 ␮A) its spread to the PDB cannot be excluded, except for one low efficacy site at a distance of 1.0 mm from the PDB (Fig. 2, E 3.25). To summarize, we encountered considerable difficulty in evoking saccades from the reticular formation with the prolonged stimuli we employed. In most sites that evoked eye movements satisfying the criteria we adopted for identification of saccades, current spread to the PDB could not be excluded. Sites located further away, consistently gave rise to either slow, clearly non-saccadic, movements or to unreliable saccade-like responses of low peak velocities. Location of stimulation sites and their relative efficacy As shown in previous studies, the size of saccades evoked by electrical stimulation of the feline SC depends on the initial position of the eyes (e.g., McIlwain, 1986; Grantyn et al., 1996). In studies, such as the present one, which are limited to their horizontal components, this relationship obtains the form, DH⫽SH⫺kH H1

(1)

where ⌬H is the amplitude of the horizontal component of saccades, H1 is the initial horizontal position of the eyes,

328

R. Kato et al. / Neuroscience 143 (2006) 319 –337

Fig. 6. Plot of horizontal ⌬E (⌬H, ordinate) versus initial horizontal E (H1, abscissa) of saccades evoked in response to the electrical stimulation of a PDB site (20 ␮A, 2⫻T, 330 Hz). Vertical dashed line marks the straight ahead initial E. The solid line is the least square linear regression line through the data (open circles) and obeys the equation displayed. Oblique dashed lines indicate the 95% confidence intervals.

SH is the horizontal component of the “characteristic” vector (McIlwain, 1986) and kH is a coefficient characterizing their position sensitivity. As shown in Fig. 6, the size of the horizontal component of saccades evoked in response to stimulation of the PDB also depended on the initial horizontal E. In this particular case, SH was equal to 8.16°, kH to 0.65 and 94% of the variance of the dependent variable can be accounted for by changes of initial E. Except for one site in which the regression was not statistically significant, similar expressions describe the position sensitivity of saccades evoked from all our PDB sites, albeit with different SH and kH values. Absolute values of SH ranged between 1.4 and 12.8°, and therefore correspond to the characteristic vectors of saccades evoked from the rostral half of the SC (Grantyn et al., 1996). The amplitude of evoked saccades depended on the intensity of stimulation as indicated by the fact that mean SH was 2.9° bigger (paired t-test: P⬍0.0001, t⫽7.2, df⫽14) when relative intensities were increased 1.25-two times while the frequency was kept constant at 330 Hz. Also consistent with the analysis of responses to SC stimulation (Moschovakis et al., 1998a), values of kH tended to increase together with the size of the horizontal component of characteristic vectors (kH⫽0.39⫹0.03SH, r 2⫽0.26, P⬍0.0001) and ranged between 0.28 and 0.88. The ratio SH/kH defines the initial position beyond which the direction of saccades should reverse (the x-intercept of the plot illustrated in Fig. 6). These initial positions of expected reversals could be as small as 2° in our sample of PDB sites. Although we explored sufficiently large and symmetrical ranges of H1 we never observed overt reversals of saccade directions, but the probability to evoke saccades decreased as H1 approached the midline, and in some cases we were unable to evoke centrifugal saccades when the eyes were initially deviated into the ipsilateral half of the oculomotor range (Fig. 3A, D). Accordingly, to compare the efficacy of different PDB stimulation sites we first calculated the IR, defined as the ratio of the number of movements we could

accept as saccadic (based on their velocity profile and the slope of their main sequence curve) and starting from H1 in the contralateral half of the oculomotor range over the total number of stimulation trials used to test a certain combination of stimulus parameters at each site and limited to the same range of H1. The values IR obtained with sitespecific optimal combinations of stimulus parameters ranged from 0.13–1.0 and over half of them (53.4%) surpassed 0.9. Typically, IR increased at higher stimulation intensity and frequency. For example, it increased from 0.86 at 100 Hz to 1.0 at 200 and 330 Hz for the site that gave rise to the movements illustrated in Fig. 3. We felt that IR would not suffice to characterize the efficacy of a site because it disregards the site’s threshold for saccades (T) and the relative intensity (I/T) at which its optimal IR was obtained. Accordingly, we decided to correct it by weighing with a coefficient (K), calculated as K⫽(40⫺T)/(I/T), where I is intensity (in ␮A) at which the optimal IR was obtained and 40 corresponds to the upper limit of T at which a site was still accepted as an effective one (40 ␮A, see Experimental Procedures). The numerator of the formula we used to calculate K weighs IR in favor of sites with low threshold (e.g. it will be equal to 36 for T⫽4 ␮A, and 5 for T⫽35 ␮A). Its denominator penalizes sites where the optimal IR was obtained at higher relative intensities (I/T). We normalized K to obtain a new estimate of the sites’ efficacy, the efficacy index, EI⫽IR⫻K/28.8, where 28.8 is the highest value of K we encountered. Fig. 2 shows (as symbols of size proportional to EI) the EI values we obtained in all PDB and PPRF sites we studied, thus reflecting their efficacy in generating saccades. Of the 64 PDB sites tested, six were assigned a value equal to 0 (Fig. 2, open circles). They include points from where no responses could be evoked (n⫽2) or which evoked slow drifts only (another two sites). While movements with velocity profiles corresponding to saccades could be induced from the remaining two sites, the slope of their main sequence curve was too shallow to classify them as saccades (cat PU: 4.2 and 6.2°/s/°). Fifty percent of the EI values we encountered were confined to a range of 0.46 – 0.59, and values equal to or above 0.7 were exceptional (five of 58 sites). In the one site where we encountered a value of EI equal to 1.0, T was equal to 4 ␮A, and IR obtained a value of 1.0 already at a current intensity of 5 ␮A (1.25⫻T). The main effect of using EI rather than IR was to reduce the number of sites with maximal efficacy rating (1.0). Based on IR such high rating would be given to 41.4% of the sites we studied. In contrast, the EI values of these sites were distributed between 0.4 and 1.0. Thus, taking in consideration the threshold and the relative intensity usually moderated our judgment of a site’s efficacy, as illustrated by the example of a site in which T⫽15 ␮A, and IR⫽1.0 was obtained at 30 ␮A. Because of its fairly high threshold and the high relative intensity (2⫻T) we employed, the rating of its efficacy was reduced to 0.43 with the use of EI. On the other hand, the rating of the least effective sites (IR⬍0.5) remained low when EI was used instead (0.056 – 0.25).

R. Kato et al. / Neuroscience 143 (2006) 319 –337

329

the SC due to the antidromic activation of recurrent collaterals. Furthermore, in case the feedback loop from the pontine saccade generator includes the SC, the latter could be activated by the PDB stimulation. The effect of axon reflex or other feedback signals on the collicular circuits could be mediated by output neurons at the subliminal fringe of the stimulation focus, i.e. those escaping the occlusion by antidromic invasion. To address this concern, we chemically inactivated the SC to test if, after this, saccades could still be evoked from the PDB and if so whether their dynamics and metrics would be altered. Injection sites in the left SC (Exps. 1L and 2L; see Table 2 for results) were verified histologically from the glial reaction surrounding the pipette tracks. In our last experiment (4M), injection in the right SC was followed by a hemorrhage that destroyed a large portion of the nucleus (see below). The two sites in the right SC (3L and 4M) were

Deactivation of the SC In the monkey, most of the SC efferent fibers entering the PDB deploy recurrent collaterals near the cell body they originate from in the SC (Moschovakis et al., 1988a,b). Although this is much less frequently the case in the cat, PDB fibers emitting recurrent collaterals have been encountered in this species as well (Grantyn and Grantyn, 1982; Moschovakis and Karabelas, 1985). Because PDB fibers are glutamatergic, and are thus likely to exert excitatory influence on their targets (Mooney et al., 1990; Büttner-Ennever and Horn, 1994) it might be argued that the saccades that we evoked from the PDB are in fact due to the antidromic activation of the SC. If this were the case, evoked saccades might not be due to bursts generated by a pontine comparator receiving step signals (e.g. the pulse trains we employed) orthodromically propagated by PDB axons. Instead, they might be due to bursts generated in

Table 2. Effects of unilateral inactivation of the SC on spontaneous saccades and on saccades evoked by stimulation of the PDB Variable

Exp 1L

Exp 2L

Exp 3L

Exp 4M

Side of SC inactivation Drug used Injected volume (␮l)

Left Lidocaine 3.5

Left Lidocaine 5.0

Right Lidocaine 5.0

Right Muscimola 4.0

Spontaneous saccades Variable Time after injection end (min) Main sequenceb (°/s/°) R2 Difference of slopes (P) Saccade frequencyb (1/min) Mean E shiftd (°)

Control

Test

Control

Test

1.5–3 10.94 0.72

7.06 0.89

Test

1–3.5 9.52 0.77

5.91 0.66

0.2303 24.6

Control

Control

Test

3–4.5 11.48 0.75

0.03676

5.47 0.90

5–6.5 xc xc

11.11 0.79 xc

0.03295

3.6 ⫺4.7

24.6

7.8 ⫺3.6

26.4

9.0 4.9

Test

Control

Test

Control

21.6

0 15.4

Control

Test

Saccades evoked by stimulation of the PDB Variable Side of stimulation Stimulation frequency (Hz) Intensity (␮A) Intensity (⫻T) Time after injection end (min) Incidence ratioe Mean latency (ms) Difference of latencies (P) Characteristic vector SH (°) Difference of SH (P) Main sequence (°/s/°) R2 Difference of slopes (P) a

Control Right 200 10 ⫻2.0 0.88 75.1

Right 200 9 ⫻1.5 3–10 0.94 65.8

0.98 82.3

5.84

6.43

10.24 0.74 0.6198

4–9 1.00 63.0

0.83 86.6

6.81

⫺4.79

9.59 0.90

Left 200 12 ⫻1.7 5–10 0.88 84.5

0.66 84.0

⫺3.69

⫺5.08

0.0149 ⫺2.20 ⬍0.0001

6.93 0.83

10.81 0.88

7.67 0.76

0.7140

xxf

0.0089 10.69 0.90

Left 100 18 ⫻2.0

⬍0.0001

0.0542 6.32

Test

10.94 0.95 0.00095

xxf 8.78 0.80 0.00727

7–14 0.83 94.8

⬍0.0001

Muscimol injection was accompanied by hemorrhage and large lesion. Included are only data on spontaneous saccades in the same direction as that of saccades evoked by PDB stimulation, in corresponding SC-inactivation experiments. c During the indicated recording time the cat was unable to generate leftward (contra-lesional) spontaneous saccades. d Difference between mean horizontal E during post- and pre-injection records. Negative shifts are to left, positive to right. e Incidence ratio is calculated by dividing the number of evoked saccades by the number of stimulus trains when initial horizontal Es are in the hemifield opposite to the side of stimulation. f Statistical significance of the difference between control and test SH cannot be tested because the slopes of regression lines relating amplitude to the initial eye position differ significantly (experiment 2L: P⬍0.0001; experiment 3L: P⫽0.0019). b

330

R. Kato et al. / Neuroscience 143 (2006) 319 –337

therefore reconstructed from known positions of stereotaxically placed electrolytic lesions. In projections on the horizontal plane, two sites (Exp. 2L and 3L) were located near the center of the SC. The medio-lateral locations of the other two sites (Exps. 1L and 4M) were also in the middle of the SC but they were placed more caudally, at the border between the rostral two thirds and the caudal third of the SC. In Exps. 1L and 3L, injections were centered on a point in the lower half of the intermediate gray layer, and in Exps. 2L and 4 M in the deep gray layer near its border with the midbrain reticular formation. Histological examination of the left SC, 3– 4 weeks after the injection revealed dense glial reaction and disappearance of neurons in zones surrounding the calculated positions of pipette tips. In Exp. 1L and 2L the diameters of these zones measured 2.0 and 1.3 mm medio-laterally and 2.4 and 3.0 mm dorso-ventrally, respectively. Clearly, injections of large volumes of liquid resulted in local damage of the SC. Not surprisingly, electrical stimulation of nearby SC sites in three lidocaine experiments did not evoke any response immediately after the end of the injection, even when tested with high currents (50 – 60 ␮A) and no recovery was seen at a later time (after 10 –30 min). In the muscimol experiment, local SC stimulation became ineffective during the adjustment of the final position of the pipette, due either to drug leakage or to mechanical lesion. Histological evaluation of the material two weeks after this experiment showed extensive hemorrhage that destroyed all layers of the caudal two thirds of the SC, as well as the dorsal mesencephalic reticular formation laterally to the central gray. To judge the severity of collicular inactivation we analyzed the spontaneous eye movements evoked 1– 4.5 min after injection of lidocaine. These served as a control for the responses to PDB stimulation that followed immediately afterward (Table 2). Soon after lidocaine injection, the mean horizontal E shifted by 4 –5° to the side of the injected SC. Fixation at greater eccentricities in the ipsilesional hemifield became unstable. It was always accompanied by slow centripetal drifts and occasionally by nystagmus-like patterns with ipsi-lesional quick phases. Contra-lesional saccades could cross the midline but terminated at small eccentricities (⬍5°). Fixations in the contra-lesional hemifield were of brief duration and sometimes unstable, showing centripetal drifts. The total time of fixation in the contra-lesional field occupied only 17–21% of the total duration of the records, in contrast to the roughly symmetrical occupancy of both hemifields prior to injections. Effects on spontaneous saccades in the same direction as those evoked by PDB stimulation, i.e. contraversive to the inactivated SC, are summarized in Table 2. In spite of our active efforts to provoke orienting movements, the frequency of saccades was substantially reduced. The ranges of saccade amplitudes and peak velocities decreased, as did the slope of the main sequence curve (by 35–52%). Unexpectedly, ipsi-lesional saccades were depressed in a roughly similar manner. In two lidocaine experiments, no signs of recovery of spontaneous saccades were seen 15–25 min after the injection,

but it was complete 1 h later. In the third experiment (Exp. 3L, Table 2) recovery was almost complete 20 min after the injection. Accordingly, we used the earliest post-injection responses to PDB stimulation (3–10 min) to evaluate the effect of SC-inactivation. It should be noted that none of our lidocaine experiments resulted in long-term changes of the cat’s posture and oculomotor behavior. As noted above, muscimol injection (Exp. 4M, Table 2) was accompanied by vascular damage and a large lesion of the SC and subjacent tegmentum. The effect on spontaneous eye movements was devastating. Mean horizontal E shifted by more than 15° into the ipsi-lesional hemifield, and remained there for the duration of the experiment (2 h). Contra-lesional saccades disappeared soon (5– 6.5 min) after the injection, and the few ipsi-lesional ones were of small amplitude (⬍5°) and low velocity (⬍40°/s). Slow centripetal drifts and short periods of nystagmoid movements rendered stable fixation impossible. These symptoms improved little when the animal was examined at later times. For example, contra-lesional saccades were seen again 45 min after the injection but only very infrequently. Starting from far eccentric Es, they never crossed the midline, so that all eye movements were made between about 5 and 20° of eccentricity in the ipsi-lesional hemifield. After release from the experimental setup the cat displayed curving of the body toward the lesioned side and intermittent circling in the same direction. These symptoms disappeared about 20 h later but neglect of contra-lesional space persisted for at least three days, and recovery was incomplete during the remaining two weeks of observation. Fig. 7 compares pre- and post-injection (made between 4 and 9.5 min after the end of the injection) records of saccades evoked from the right PDB after lidocaine injection (5 ␮l) in the left SC. In this, and all other SC injection experiments we used the same stimulus parameters to obtain pre- (control) and post-injection (test) records. We did not measure thresholds after injection, assuming that their change should be reflected in the change of the IR of evoked saccades. A comparison of control (A–C) and test (D–F) responses does not reveal any systematic effect of SC-inactivation on the time course, amplitudes and velocities of PDB-evoked saccades (see also Fig. 8, 2L). If anything, the reproducibility of movements was improved in this particular example. Effects of SC-inactivation on the main sequence relationships obtained from four experiments are illustrated in Fig. 8, and Table 2 summarizes the changes of several additional parameters. Data presented in Fig. 8 permit us to qualitatively conclude that effects of lidocaine inactivation of the SC on the main sequence of PDB-evoked saccades are very weak and inconsistent. The slope of the regression line did not change in Exp. 1L, showed a slight but statistically significant (P⬍0.001) increase in Exp. 2L and was depressed in Exp. 3L (P⬍0.01). It should be noted, however, that in the latter case the slope of the main sequence relationship of spontaneous saccades was depressed by 52%, as compared with 21% for PDB-evoked saccades. The probability to evoke saccades from contralateral initial Es did not change appreciably, and laten-

R. Kato et al. / Neuroscience 143 (2006) 319 –337

331

Fig. 7. Horizontal components of eye movements evoked in response to the electrical stimulation of the PDB before (A–C) and after (D–F) injection of lidocaine in the left SC. Records are from experiment 2L (see Table 2 for additional quantitative data). Superimposed traces (A, B) and main sequence equation (C) are obtained from one of the two control records (n⫽39). Equation parameters are not identical to those in Table 2 because the latter is based on two control records (n⫽126). Layout and conventions as in Fig. 5.

332

R. Kato et al. / Neuroscience 143 (2006) 319 –337

Fig. 8. Main sequence relations of saccades evoked by stimulation of the PDB before (open circles) and after (dots) injections of lidocaine (1L, 2L and 3L) and muscimol (4M) in the contralateral SC. Quantities of injected substances and timing of control and test records are given in Table 2.

cies either did not change or became shorter. A stronger depression of response to PDB stimulation was observed in experiment 4 M only, in which a combination of muscimol injection and an extensive vascular lesion led to the complete absence of self-generated contra-lesional saccades. Nonetheless, PDB stimulation still evoked saccades with a high probability and the depression of the slope of the main sequence curve (by 29%) was smaller than the effect on spontaneous saccades in lidocaine experiments. The E sensitivity of PDB-evoked saccades was also impervious to SC lesions since the slope of regression lines such as that illustrated in Fig. 6 was not altered appreciably by either lidocaine or muscimol injections in the contralateral SC.

DISCUSSION Our data demonstrate that electrical stimulation of the PDB evokes saccades, often accompanied by slow drifts. The latency of evoked saccades is influenced by the intensity and frequency of stimulation while their amplitude depends on the intensity of stimulation and the initial position of the eyes. The dynamics of evoked saccades differ little from those of spontaneous or visually triggered saccades of the same species and from those evoked in response to the electrical stimulation of the cat SC. Since we were generally unable to evoke saccades from nearby regions of the PPRF, our data demonstrate that saccade generation relies on a specialized circuit (such as the burst generator). Our study is one of few that have combined electrical microstimulation with localized inactivation in alert behaving animals. It demonstrates that PDB-

evoked saccades are not abolished by the reversible chemical inactivation of the SC and thus leads us to conclude that activation of the SC, either antidromic or orthodromic, is not needed for their generation. Our data clearly demonstrate that the burst generator of the saccadic system is located downstream of the SC If the burst generator is configured as a local loop controller, as assumed by most models of saccadic control, our data also demonstrate that its feedback path closes downstream of the SC. PDB-evoked saccades survive lesions of the SC Our data demonstrate that the SC is not needed to generate saccades with normal characteristics in response to stimulation of the contralateral PDB. We shall discuss separately the results obtained with lidocaine and muscimol injections because the former is known to exert a weaker effect on saccade related SC processes (Hikosaka and Wurtz, 1985, 1986). Lidocaine injections did not alter the likelihood of evoking saccades from the PDB and did not prolong their latencies. The slope of the velocity-amplitude relationship was significantly reduced in only one of three experiments, and the reduction was modest. Such small and poorly reproducible effects invite us to ask whether the local anesthesia involved a sufficiently large volume of the SC. This was evidently the case as shown by the elimination of responses to SC stimulation and by changes in the cat’s spontaneous oculomotor behavior shortly before testing the responses to PDB stimulation. In all lidocaine experiments we observed a strong suppression of the readiness

R. Kato et al. / Neuroscience 143 (2006) 319 –337

to make spontaneous saccades and a restriction of their amplitude and velocity together with significantly shallower main sequence slopes. Neglect of the contra-lesional hemifield, typical of the unilateral surgical ablation of the SC (Sprague and Meikle, 1965; Flandrin and Jeannerod, 1981), was observed as well. It was expressed in the shift of the mean horizontal E to the side of the inactivated SC and in the virtual absence of orienting saccades to visual and auditory stimuli in the contra-lesional hemifield, starting from about 5° off the midline. All this indicates that a major portion of the SC was inactivated. The muscimol experiment is of particular interest as it illustrates the effect of an unquestionably extensive inactivation. Inadvertent vascular lesion resulted in the destruction of 59% of the SC together with the underlying dorsal mesencephalic tegmentum. In this case saccades evoked by PDB stimulation showed an obvious reduction of peak velocities and shallower main sequence slopes (from 10.81–7.67°/s/°; Fig. 8 4M). However, at the same time spontaneous contra-lesional saccades were completely eliminated, mean horizontal E was strongly shifted toward the side of the lesion (15°), stable fixation was absent, and, at longer post-injection times, rare contra-lesional saccades failed by far (10 –13°) to reach the midline. We are not able to separately estimate the effects due to the injected muscimol, tissue ablation, and the compression of adjacent regions due to bleeding. Nonetheless, this experiment demonstrates that PDB-evoked saccades can be evoked with high probability, that they obey the main sequence relationship, and that the overall reduction of their velocity is modest in spite of the severe impairment of the SC and the inability of the animal to generate spontaneous saccades. Comparison with previous studies To our knowledge this is the first study to demonstrate that saccades can be evoked in response to stimulation of SC output fibers in the PDB. Cohen and Komatsuzaki (1972) stimulated the PPRF of alert monkeys, just lateral to the PDB, and the typical responses they reported consisted of constant velocity movements. They also illustrated fast movements with exponential trajectories (their “inconstant velocity” movements) but few examples of saccade-like movements. Keller (1974) as well as Sparks and colleagues (1987, 2002) have used the PPRF stimulation paradigm and they mention only slow ramp-like evoked movements terminating only after the end of the stimulus train. In general, our observations from stimulation sites located in the core of the pontine reticular formation agree with earlier studies in that slow movements with variable time course were the predominant variety of the responses we observed. However, when electrodes were placed close to or inside the paramedian tracts, saccades could be reliably evoked. In several respects, saccades evoked from the PDB were similar to self-generated ones of the cat and to those evoked in response to stimulation of the SC. Firstly, their velocity profiles were nearly symmetrical, and they terminated long before the end of the stimulus train. Also, their

333

peak velocities increased together with their amplitude following the main sequence curve of the cat (Evinger and Fuchs, 1978; Guitton et al., 1984). Finally, their size depended on the initial position of the eyes, in the manner repeatedly documented for SC-evoked saccades in the same species (Guitton et al., 1980; McIlwain, 1986; Grantyn et al., 1996; Moschovakis et al., 1998a) and their latencies were indistinguishable from those of SC-evoked saccades (Grantyn et al., 1996). There are, however, certain differences between the PDB- and the SC-evoked saccades even when the two structures are stimulated with identical parameters. Firstly, the slopes of the main sequence curve of saccades evoked from some (n⫽11) PDB sites were relatively shallow (⬍9°/s/°). Other sites were much more effective as indicated by the fact that 26% (15/58) of them displayed main sequence slopes of 14.0 –19.8°/s/° which corresponds to and even exceeds the top of the ranges reported for visually triggered saccades in the cat (Guitton et al., 1984). Also, it was more difficult to evoke staircases of saccades in response to long duration stimulation of the PDB than it is to evoke them from the SC. Both could be due to differences in the efficacy with which SC stimulation and PDB stimulation activate OPNs. Excitatory SC projections to OPNs are strong from the rostral pole, the socalled fixation zone, and decline toward the central and caudal locations in the SC (Paré and Guitton, 1994; Sugiuchi et al., 2005). Because our collicular stimulation sites were outside the rostral pole, co-activation of OPNs should be weak. This is not the case with PDB stimulation where fibers issuing from the rostral and caudal poles may have similar chances to be recruited, thus increasing the fraction of co-excited OPNs. If OPN discharge was clamped to a nonzero, even if low, value in response to PDB stimulation they could counteract the excitatory drive to burst neurons and account, at least in part, for the shallower main sequence slopes of PDB-evoked saccades. A similar explanation could be proposed for the relative difficulty to evoke saccadic staircases from the PDB. During SC stimulation, the disynaptic inhibition, originating from both the rostral and the caudal SC (Yoshida et al., 2001) and maybe mediated by “latch” LLBs, apparently predominates at the beginning of the train, thus silencing OPNs and triggering the primary saccade. Secondary and higher order saccades would then be elicited every time the RI is reset, i.e. as soon as OPNs resume firing (Moschovakis, 1994). In view of the mixed, excitatory and inhibitory, input transmitted to OPNs by collicular axons, PDB stimulation might clamp the rate of firing of OPNs to some low value for the whole stimulus duration, thus preventing the resumption of OPN firing and disabling the resetting mechanism of the RI and the generation of saccadic staircases (Moschovakis, 1994). Evidently, not all PDB sites were equally effective in generating saccades. Besides six PDB sites from which we were unable to evoke saccades, our sample of 58 effective sites contains five sites from which we could obtain saccades with a probability lower than 0.5, even when the starting position of the eyes was deviated con-

334

R. Kato et al. / Neuroscience 143 (2006) 319 –337

tralaterally. The EI of these sites was also low (⬍0.25) because of elevated thresholds (20 – 40 ␮A). On the other hand, we were able to evoke centripetal saccades with a probability (IR) of 1 from 41% of our effective sites with optimal combinations of stimulation parameters. Typically, such sites were characterized by low thresholds (ⱕ10 ␮A); they did not need strong currents to obtain reliable responses, and obtained thus high EI values (⬎0.5). This leads us to conclude that the precise location of the electrode tip is a major source of the observed variability of stimulation efficacy. The probability of evoking saccades also depended on the initial position of the eyes; in some sites it proved very difficult to evoke centrifugal saccades. This could be due to the relatively small size of the characteristic vectors of evoked saccades in particular when combined with relatively high position sensitivities. The small size of the characteristic vectors could, in turn, be due to our occasional inability to recruit a large enough number of PDB fibers because of the small tips and weak stimulation currents we employed. The choice of our stimulation technique was dictated by the need to obtain the best possible spatial selectivity of our stimuli, in particular since we were targeting a narrow fiber bundle (⬍0.5 mm in mediolateral extent). Restricting the current range to 4 – 40 ␮A was needed to reduce the effective current spread while the upper limit was imposed by the current passing capacity of the capillary or tungsten microelectrodes we employed. Besides saccades, electrical stimulation of the PDB was seen to generate slow drifts akin to those evoked in response to the electrical stimulation of the SC (Grantyn et al., 1996). These have been attributed to relatively direct projections from the SC to extraocular MNs that largely bypass the burst generators (Grantyn et al., 1996; Moschovakis et al., 1998a). These authors provided a substantial list of fibers and neurons likely to mediate them. Briefly, they include collicular efferents deploying terminal fields in the abducens nucleus (Grantyn and Grantyn, 1982; Grantyn and Berthoz, 1985; Olivier et al., 1993), reticulospinal neurons which receive input from the SC and project to the abducens nucleus (Grantyn et al., 1980, 1987), neurons of the PH which also receive input from the SC and project to the abducens nucleus (Grantyn and Grantyn, 1982; Grantyn and Berthoz, 1985; Olivier et al., 1993) and tectorecipient reticulospinal neurons which discharge during orienting eye–neck synergies and give substantial collateral projections to both the PH and the medial vestibular nucleus (Grantyn et al., 1980, 1987, 1992). Since, most of the tectal efferents participating in these projections pass through the PDB, their stimulation and the generation of slow drifts were unavoidable in a majority of our stimulation sites. Implications for models of the saccadic system Our experiments were designed to test the existence of a post-collicular saccade generating mechanism rather than distinguish between alternative post-collicular mechanisms. They are thus consistent with any post-collicular pulse generator and not just those that have been pro-

posed for the brain stem. Such an alternative is a transcerebellar pulse generator (Quaia et al., 1999). Engagement of this circuit in our experiments is consistent with the fact that the PDB projects to the NRTP (e.g., Graham, 1977; Grantyn and Grantyn, 1982; Huerta and Harting, 1984) which then projects to the fastigial oculomotor region (FOR). It is also consistent with the well-known involvement of both the NRTP (Keller and Crandall, 1981) and the FOR (Ohtsuka and Noda, 1991; Goffart et al., 2004) in saccade generation. However, there are several reasons to doubt that the cerebellum houses the local feedback loop of the saccadic system (summarized in Scudder et al., 2002). For example, the discharge of FOR neurons is too variable to entrust them with the control of saccade metrics. More importantly, one of the predictions of a transcerebellar feedback loop model has been experimentally refuted. Rather than last for the duration of the stimulation (as predicted by the trans-cerebellar model of Quaia et al., 1999), SC-evoked saccades stop well before the end of the stimulation and their velocity profiles do not change markedly when the FOR is lesioned (Guillaume and Pélisson, 2001). In light of this evidence and our present findings, it is parsimonious to conclude that saccades evoked in response to the electrical stimulation of the cerebellum are due to the engagement of the same circuit that is responsible for the generation of saccades after PDB, SC and possibly cortical stimulation as well, i.e. a burst generator located downstream of all these structures in the brainstem. Several models place the SC inside the local feedback loop of the saccadic system (Waitzman et al., 1991; Lefèvre and Galiana, 1992; van Opstal and Kappen, 1993; Arai et al., 1994; Grossberg et al., 1997). Consistent with this notion, the instantaneous firing rate of “clipped” presaccadic SC cells decreases together with instantaneous Me, for a wide range of saccade sizes (Waitzman et al., 1988). Moreover, the discharge of SC neurons and the trajectory of the eyes can be disrupted in parallel. For example, saccades are known to be interrupted in midflight in response to OPN stimulation and to resume their course after the end of the stimulation (Keller, 1977; King and Fuchs, 1977) and the same is true of the discharge of SC burst cells (Keller and Edelman, 1994; Munoz et al., 1996). Also, the duration of the discharge of presaccadic SC cells increases together with the duration of saccades and their intensity decreases with saccade velocity following muscimol injections in the OPN area (Soetedjo et al., 2002) and when saccades are perturbed by blinks (Goossens and van Opstal, 2000). Our experiments were not intended to test the existence of a feedback loop operating at the level of the SC, and therefore they do not disprove it. However, there are reasons to doubt that the trans-collicular feedback loop could perform all operations usually assigned to the local feedback loop of the saccadic system. For example, to satisfy the normal amplitude– duration relationships, models that place the SC inside the local loop predict that SC burst neurons firing for bigger saccades should fire longer. This is difficult to reconcile with the fact that the duration of

R. Kato et al. / Neuroscience 143 (2006) 319 –337

the bursts of SC presaccadic cells is poorly correlated to the size of the saccades they prefer (Sparks and Mays, 1990). Also, the frequency of the discharge of SC neurons is reduced considerably more than expected from the reduction of the velocity of saccades accompanied by vergence (Walton and Mays, 2003). Moreover, the discharge of SC presaccadic neurons is poorly related to dynamic Me in several circumstances, such as when saccades are executed in the absence of a precise target (e.g., to the center of a large rectangle; Edelman and Goldberg, 2003) or when they are perturbed with OPN stimuli (Keller and Edelman, 1994) or blinks (Goossens and van Opstal, 2000). The same is true for saccades slowed down following muscimol injections in the OPN area (Soetedjo et al., 2002). In all these circumstances, a local feedback loop that, consistent with our data, operates below the SC would generate saccades that follow a normal time course despite receiving SC signals that are not tightly correlated with saccade dynamics. It might be argued that electrical stimulation of the SC suffices to decide if the local loop of the saccadic system closes through the SC or below it. This is based on the assumption that stimulation of the SC preferentially activates its output fibers, which thus carry a signal dominated by the time course of the stimulation rather than the results of computations performed by a comparator residing in it (Gnadt et al., 2001). However, incoming preterminal fibers are more likely to be activated by the electrical stimulus (Gustafsson and Jankowska, 1976). This is consistent with the preferential activation of axonal branching points (the relative density of which increases in and near axonal terminal fields) rather than cell bodies or axonal initial segments following electrical stimulation of cortical gray layers (Nowak and Bullier, 1998a,b). Accordingly, when electrical stimulation is applied to the SC, afferent pathways (providing the SC with feedforward as well as feedback information) and intrinsic SC connections are unavoidably activated in addition to SC output fibers and it is as likely to engage the input to its putative comparator as to engage its output. In contrast, the electrical stimulation of the PDB used in this study eliminates the ambiguity of intracollicular stimulation, in particular when combined with SC inactivation. Our data thus provide conclusive evidence that saccades can be readily evoked even if the output of the SC is a step function of time and, consequently, that an independent burst generator of the saccadic system is located downstream of the SC. Acknowledgments—We gratefully acknowledge the financial support of the French Ministry of Research (ACI 2003 N°03 5 45) the skillful assistance of M.-A. Thomas with histological procedures and the help of M. Ehrette in adjusting the mechanical setup for simultaneous work in two different regions of the brain.

REFERENCES Arai K, Keller EL (2005) A model of the saccade-generating system that accounts for trajectory variations produced by competing visual stimuli. Biol Cybern 92:21–37.

335

Arai K, Keller EL, Edelman JA (1994) Two-dimensional neural network model of the primate saccadic system. Neural Netw 7:1115–1135. Barton EJ, Nelson JS, Gandhi NJ, Sparks DL (2003) Effects of partial lidocaine inactivation of the paramedian pontine reticular formation on saccades of macaques. J Neurophysiol 90:372–386. Becker W, King WM, Fuchs AF, Jürgens R, Johanson G, Kornhuber HH (1981) Accuracy of goal-directed saccades and mechanisms of error correction. In: Progress in oculomotor research (Becker W, Fuchs AF, eds), pp 29 –37. Amsterdam: Elsevier North Holland. Bender MB (1964) The oculomotor system. New York: Harper & Row. Büttner-Ennever JA, Horn AKE (1994) Neuroanatomy of saccadic omnipause neurons in nucleus raphe interpositus. In: Contemporary ocular motor and vestibular research: a tribute to David A. Robinson (Fuchs AF et al., eds), pp 489 – 495. New York: Thieme Medical Publishers. Cohen B, Komatsuzaki A (1972) Eye movements induced by stimulation of the pontine reticular formation: evidence for integration in oculomotor pathways. Exp Neurol 36:101–117. Dalezios Y, Scudder CA, Highstein SM, Moschovakis AK (1998) Anatomy and physiology of the primate interstitial nucleus of Cajal. II. Discharge pattern of single efferent fibers. J Neurophysiol 80: 3100 –3111. Das S, Gandhi NJ, Keller EL (1995) Open-loop simulations of the primate saccadic system using burst cell discharge from the superior colliculus. Biol Cybern 73:509 –518. Edelman JA, Goldberg ME (2003) Saccade-related activity in the primate superior colliculus depends on the presence of local landmarks at the saccade endpoint. J Neurophysiol 90:1728 –1736. Evinger C, Fuchs AF (1978) Saccadic, smooth pursuit, and optokinetic eye movements of the trained cat. J Physiol (Lond) 285:209 –229. Flandrin JM, Jeannerod M (1981) Effects of unilateral superior colliculus ablation on oculomotor and vestibulo-ocular responses in the cat. Exp Brain Res 42:73– 80. Gnadt JW, Jackson ME, Litvak O (2001) Analysis of the frequency response of the saccadic circuit: System behavior. J Neurophysiol 86:724 –740. Goffart L, Chen LL, Sparks DL (2004) Deficits in saccades and fixation during muscimol inactivation of the caudal fastigial nucleus in the rhesus monkey. J Neurophysiol 92:3351–3367. Goossens HHLM, van Opstal AJ (2000) Blink-perturbed saccades in monkey. II. Superior colliculus activity. J Neurophysiol. 83: 3430 –3452. Graham J (1977) An autoradiographic study of the efferent connections of superior colliculus in the cat. J Comp Neurol 173:629 – 654. Grantyn A, Berthoz A (1985) Burst activity of identified tectoreticulospinal neurons in the alert cat. Exp Brain Res 57:417– 421. Grantyn A, Grantyn R (1982) Axonal patterns and sites of termination of cat superior colliculus neurons projecting to the tecto-bulbospinal tract. Exp Brain Res 46:243–256. Grantyn A, Hardy O, Olivier E, Gourdon A (1992) Relationship between task-related discharge patterns and axonal morphology of brainstem projection neurons involved in orienting eye and head movements. In: Vestibular and brain stem control of eye, head and body movements (Shimazu H, Shinoda Y, eds), pp 255–273. Tokyo: Japan Scientific Societies Press. Grantyn A, Ong-Meang J, Berthoz A (1987) Reticulospinal neurons participating in the control of synergic eye and head movements during orienting in the cat. II. Morphological properties as revealed by intra-axonal injections of horseradish peroxidase. Exp Brain Res 66:355–377. Grantyn AA, Dalezios Y, Kitama T, Moschovakis AK (1996) Neuronal mechanisms of two-dimensional orienting movements in the cat. I. A quantitative study of saccades and slow drifts produced in response to the electrical stimulation of the superior colliculus. Brain Res Bull 41:65– 82. Grantyn R, Baker R, Grantyn A (1980) Morphological and physiological identification of excitatory pontine reticular neurons projecting

336

R. Kato et al. / Neuroscience 143 (2006) 319 –337

to the cat abducens nucleus and spinal cord. Brain Res 198: 221–228. Grossberg S, Roberts K, Aguilar M, Bullock D (1997) A neural model of multimodal adaptive saccadic eye movement control by superior colliculus. J Neurosci 17:9706 –9725. Guillaume A, Pélisson D (2001) Gaze shifts evoked by electrical stimulation of the superior colliculus in the head-unrestrained cat. II. Effect of muscimol inactivation of the caudal fastigial nucleus. Eur J Neurosci 14:1345–1359. Guitton D, Crommelinck M, Roucoux A (1980) Stimulation of the superior colliculus in the alert cat. I. Eye movements and neck EMG activity evoked when the head is restrained. Exp Brain Res 39:63–73. Guitton D, Douglas RM, Volle M (1984) Eye-head coordination in cats. J Neurophysiol 52:1030 –1050. Gustafsson B, Jankowska E (1976) Direct and indirect activation of nerve cells by electrical pulses applied extracellularly. J Physiol (Lond) 258:33– 61. Harting JK, Huerta MF, Frankfurter AJ, Strominger NL, Royce JG (1980) Ascending pathways from the monkey superior colliculus: an autoradiographic analysis. J Comp Neurol 192:852– 882. Hikosaka O, Wurtz RH (1985) Modification of saccadic eye movements by GABA-related substances. I. Effect of muscimol and bicuculline in monkey superior colliculus. J Neurophysiol 53:266–291. Hikosaka O, Wurtz RH (1986) Saccadic eye movements following injection of lidocaine into the superior colliculus. Exp Brain Res 61:531–539. Huerta MF, Harting JK (1984) The mammalian superior colliculus: Studies of its morphology and connections. In: Comparative neurology of the optic tectum (Vanegas H, ed), pp 687–773. New York: Plenum Press. Jürgens R, Becker W, Kornhuber HH (1981) Natural and drug-induced variations of velocity and duration of human saccadic eye movements: Evidence for a control of the neural pulse generator by local feedback. Biol Cybern 39:87–96. Keller EL (1974) Participation of medial pontine reticular formation in eye movement generation in monkey. J Neurophysiol 37:316 –332. Keller EL (1977) Control of saccadic eye movements by midline brainstem neurons. In: Control of gaze by brainstem neurons (Baker R, Berthoz A, eds), pp 327–336. Amsterdam: Elsevier. Keller EL, Crandall WF (1981) Neural activity in the nucleus reticularis tegmenti pontis in the monkey related to eye movements and visual stimulation. Ann N Y Acad Sci 374:249 –261. Keller EL, Edelman JA (1994) Use of interrupted saccade paradigm to study spatial and temporal dynamics of saccade burst cells in superior colliculus in monkey. J Neurophysiol 72:2754 –2770. King WM, Fuchs AF (1977) Neuronal activity in the mesencephalon related to vertical eye movements. In: Control of gaze by brain stem neurons (Baker R, Berthoz A, eds), pp 319 –326. Amsterdam: Elsevier. Lefèvre P, Galiana HL (1992) Dynamic feedback to the superior colliculus in a neural network model of the gaze control system. Neural Netw 5:871–900. Malpeli JG (1999) Reversible inactivation of subcortical sites by drug injection. J Neurosci Methods 86:119 –128. Martin JH, Ghez C (1999) Pharmacological inactivation in the analysis of the central control of movement. J Neurosci Methods 86:145–159. McIlwain JT (1986) Effects of eye position on saccades evoked electrically from the superior colliculus of alert cats. J Neurophysiol 55:97–112. Mooney RD, Bennet-Clarke CA, King TD, Rhoades RW (1990) Tectospinal neurons in hamster contain glutamate-like immunoreactivity. Brain Res 537:375–380. Moschovakis AK (1994) Neural network simulations of the primate oculomotor system. I. The vertical saccadic burst generator. Biol Cybern 70:291–302. Moschovakis AK, Dalezios Y, Petit J, Grantyn AA (1998a) New mechanism that accounts for position sensitivity of saccades evoked in

response to electrical stimulation of superior colliculus. J Neurophysiol 80:3373–3379. Moschovakis AK, Kitama T, Dalezios Y, Petit J, Brandi AM, Grantyn AA (1998b) An anatomical substrate for the spatiotemporal transformation. J Neurosci 18:10219 –10229. Moschovakis AK, Karabelas AB (1985) Observations on the somatodendritic morphology and axonal trajectory of intracellularly HRPlabeled efferent neurons located in the deeper layers of the superior colliculus of the cat. J Comp Neurol 239:276 –308. Moschovakis AK, Karabelas AB, Highstein SM (1988a) Structurefunction relationships in the primate superior colliculus. I. Morphological classification of efferent neurons. J Neurophysiol 60: 232–262. Moschovakis AK, Karabelas AB, Highstein SM (1988b) Structurefunction relationships in the primate superior colliculus. II. Morphological identity of presaccadic neurons. J Neurophysiol 60: 263–302. Moschovakis AK, Kato R, Dalezios Y, Grantyn A (2005) Saccades evoked in response to microstimulation of the predorsal bundle (PDB) in the alert cat. Proc Soc Neurosci 31:167–169. Moschovakis AK, Scudder CA, Highstein SM (1991a) Structure of the primate burst generator. I. Medium-lead burst neurons with upward on-directions. J Neurophysiol 65:203–217. Moschovakis AK, Scudder CA, Highstein SM, Warren JD (1991b) Structure of the primate burst generator. II. Medium-lead burst neurons with downward on-directions. J Neurophysiol 65:218–229. Moschovakis AK, Scudder CA, Highstein SM (1996) The microscopic anatomy and physiology of the mammalian saccadic system. Progr Neurobiol 50:133–254. Munoz DP, Waitzman DM, Wurtz RH (1996) Activity of neurons in monkey superior colliculus during interrupted saccades. J Neurophysiol 75:2562–2580. Navarro-López Jde D, Alvarado JC, Márquez-Ruiz J, Escudero M, Delgado-García JM, Yajeya J (2004) A cholinergic synaptically triggered event participates in the generation of persistent activity necessary for eye fixation. J Neurosci 24:5109 –5118. Nowak LG, Bullier J (1998a) Axons but not cell bodies are activated by electrical stimulation in cortical gray matter. I. Evidence from chronaxie measurements. Exp Brain Res 118:477– 488. Nowak LG, Bullier J (1998b) Axons but not cell bodies are activated by electrical stimulation in cortical gray matter. II. Evidence from selective inactivation of cell bodies and axon initial segments. Exp Brain Res 118:489 –500. Ohtsuka K, Noda H (1991) Saccadic burst neurons in the oculomotor region of the fastigial nucleus of macaque monkeys. J Neurophysiol 65:1422–1434. Olivier E, Grantyn A, Chat M, Berthoz A (1993) The control of slow orienting movements by tecto-reticulo-spinal neurons in the cat: behaviour, discharge patterns and underlying connections. Exp Brain Res 93:435– 449. Paré M, Guitton D (1994) The fixation area of the cat superior colliculus: effects of electrical stimulation and direct connection with brain stem omnipause neurons. Exp Brain Res 101:109 –122. Populin LC, Yin TCT (1998) Behavioral studies of sound localization in the cat. J Neurosci 18:2147–2160. Quaia C, Lefèvre P, Optican LM (1999) Model of the control of saccades by superior colliculus and cerebellum. J Neurophysiol 82:999 –1018. Robinson DA (1972) Eye movements evoked by collicular stimulation in the alert monkey. Vision Res 12:1795–1808. Robinson DA (1973) Models of the saccadic eye movement control system. Biol Cybern 14:71– 83. Ron S, Robinson DA (1973) Eye movements evoked by cerebellar stimulation in the alert monkey. J Neurophysiol 36:1004 –1022. Scudder CA (1988) A new local feedback model of the saccadic burst generator. J Neurophysiol 59:1455–1475.

R. Kato et al. / Neuroscience 143 (2006) 319 –337 Scudder CA, Kaneko CRS, Fuchs AF (2002) The brainstem burst generator for saccadic eye movements. A modern synthesis. Exp Brain Res 142:439 – 462. Soetedjo R, Kaneko CRS, Fuchs AF (2002) Evidence that the superior colliculus participates in the feedback control of saccadic eye movements. J Neurophysiol 87:679 – 695. Sparks DL (2002) The brainstem control of saccadic eye movements. Nat Rev 3:952–964. Sparks DL, Barton EJB, Ganhi NJ, Nelson J (2002) Studies of the role of the paramedian pontine reticular formation in the control of head-restrained and head-unrestrained gaze shifts. Ann N Y Acad Sci 956:85–98. Sparks DL, Mays LE (1990) Signal transformations required for the generation of saccadic eye movements. Annu Rev Neurosci 13: 309 –336. Sparks DL, Mays LE, Porter JD (1987) Eye movements induced by pontine stimulation: Interaction with visually triggered saccades. J Neurophysiol 58:300 –318. Sprague JM, Meikle TH (1965) The role of the superior colliculus in visually guided behavior. Exp Neurol 11:115–146. Sugiuchi Y, Izawa Y, Takahashi M, Na J, Shinoda Y (2005) Physiological characterization of synaptic inputs to inhibitory burst neurons from the rostral and caudal superior colliculus. J Neurophysiol 93:697–712.

337

Tehovnik EJ, Sommer MA (1997) Effective spread and timecourse of neural inactivation caused by lidocaine injection in monkey cerebral cortex. J Neurosci Methods 74:17–26. van Gisbergen JAM, Robinson DA, Gielen S (1981) A quantitative analysis of generation of saccadic eye movements by burst neurons. J Neurophysiol 45:417– 442. van Opstal AJ, Kappen H (1993) A two-dimensional ensemble coding model for spatial-temporal transformation of saccades in monkey superior colliculus. Network 4:19 –38. Waitzman DM, Ma TP, Optican LM, Wurtz RH (1988) Superior colliculus neurons provide the saccadic motor error signal. Exp Brain Res 72:649 – 652. Waitzman DM, Ma TP, Optican LM, Wurtz RH (1991) Superior colliculus neurons mediate the dynamic characteristics of saccades. J Neurophysiol 66:1716 –1737. Walton MMG, Mays LE (2003) Discharge of saccade-related superior colliculus neurons during saccades accompanied by vergence. J Neurophysiol 90:1124 –1139. Yoshida K, Iwamoto Y, Chimoto S, Shimazu H (2001) Disynaptic inhibition of omnipause neurons following electrical stimulation of the superior colliculus in alert cats. J Neurophysiol 85:2639 –2642. Zee DS, Optican LM, Cook JD, Robinson DA, Engel WK (1976) Slow saccades in spinocerebellar degeneration. Arch Neurol 33:243–251.

(Accepted 18 July 2006) (Available online 23 August 2006)