RESEARCH ARTICLE

Victor M. Henriquez Æ Craig Evinger

Modification of cornea-evoked reflex blinks in rats

Received: 26 May 2004 / Accepted: 29 October 2004 / Published online: 23 March 2005� Springer-Verlag 2005

Abstract Although maintaining the tear film on thecornea is the most important role of blinking, infor-mation about the organization and modification ofcornea-evoked blinks is sparse. This study character-izes cornea-evoked blinks and their modification inurethane-anesthetized rats. Cornea-evoked blinks typ-ically begin 16.2 ms after an electrical stimulus to thecornea and last an average of 50.2 ms. In anesthetizedrats, the blink only occurs ipsilateral to the stimulus.In response to cornea stimulation, the orbicularis oculiEMG activity typically exhibits two bursts that cor-relate with the arrival of Ad and C-fiber inputs to thespinal trigeminal complex. In the paired-stimulusparadigm, suppression of the blink evoked by thesecond cornea stimulus occurs for interstimulus inter-vals less than 300 ms and is exclusively unilateral.Stimulation of the contralateral cornea does not affectsubsequent blinks evoked from stimulation of theipsilateral cornea. To determine whether activation ofcornea-related neurons in the border region betweenthe spinal trigeminal caudalis subdivision and the C1spinal cord (Vc/C1) inhibits the second blink in thepaired-stimulus paradigm, we examine the suppressionof cornea-evoked blinks caused by microstimulation inthis region. This suppression of orbicularis oculi EMGactivity begins 8.3 ms after Vc/C1 stimulation. Acti-vation of this region, however, is unlike suppression inthe paired-stimulus paradigm because Vc/C1 activationbilaterally inhibits cornea-evoked blinks. Thus, acti-vation of Vc/C1 is a previously unidentified mecha-nism for modulating cornea-evoked blinks.

Keywords Cornea Æ Reflex blinks Æ Rats

Introduction

Because the primary role of the eyelids is to protect thecornea by maintaining the tear film bathing the cornea,cornea-evoked blinking is the linchpin of the eyelidsystem. Nevertheless, the characteristics and neuralbasis of cornea-evoked blinks are relatively unknown.Because cornea stimulation is painful, there are only afew studies in humans characterizing blinks evokedexclusively by cornea stimulation (Accornero et al.1980; Mukuno et al. 1983; Ongerboer de Visser 1983b;Ongerboer de Visser 1983a; Berardelli et al. 1985;Manfredi et al. 1985; Cruccu et al. 1986; Agostino et al.1987; Cruccu et al. 1997; Acosta et al. 2001a; Acostaet al. 2001b; Aramideh and Ongerboer de Visser 2002).Electrical stimulation of the human cornea elicits asingle, bilateral contraction of the lid closing orbicu-laris oculi (OO) muscle with a latency of 40.7±2.5 mslasting 94±27 ms (Accornero et al. 1980). Althoughthere are numerous studies of blinks evoked by airpuffs directed at the cornea, this stimulus invariablyconfounds cutaneous blinks caused by activation of Abafferents innervating the skin and hair around the eyewith cornea blinks evoked solely from Ad and C-fibercornea afferents. Electrical stimulation of the supraor-bital branch of the trigeminal nerve (SO), an innocuouscutaneous stimulus, evokes a short latency, unilateralOO activation, R1, and a longer latency, bilateralactivation, R2 (Kimura et al. 1969). In humans, the R1occurs 10.8±3.2 ms after the SO stimulus and lastsonly 9.3±2.3 ms. The larger R2 component begins34.4±4.9 ms after the stimulus and lasts 34.6±4.9 ms(Shahani and Young 1972; Tackmann et al. 1982;Powers et al. 1997). Despite similar latencies for the R2component of SO-evoked blinks and cornea-evokedblinks, interacting SO and cornea stimuli demonstratesthat cornea and cutaneous trigeminal blink circuits aredifferent (Berardelli et al. 1985). One goal of the pres-ent study is to characterize cornea-evoked blinks andtheir modification in anesthetized rats.

V. M. Henriquez Æ C. Evinger (&)Departments of Neurobiology and Behavior and Ophthalmology,SUNY Stony Brook, Stony Brook,NY 11794–5230,USAE-mail: levinger@notes.cc.sunysb.edu

Exp Brain Res (2005) 163: 445–456DOI 10.1007/s00221-004-2200-y

The most frequently used blink modification proce-dure is the paired-stimulus paradigm (Powers et al.1997). In this paradigm the amplitude of the blink elic-ited by the second of two identical stimuli is less thanthat evoked by the first stimulus. Determining the depthof the second blink’s suppression as a function of theinterval between the two stimuli provides an estimate ofblink circuit excitability. Our studies determine rodentcornea reflex blink excitability and attempt to identifythe neural basis for blink suppression in the paired-stimulus paradigm.

Cornea afferents primarily terminate in two distinctregions of the rat spinal trigeminal complex, the inter-polaris-caudalis transition zone (Vi/Vc) and more cau-dally, in the caudalis-C1 spinal cord border region (Vc/C1) (Marfurt and Del Toro 1987; Meng et al. 1997).Physiological and clinical data suggest that these tworegions have different roles in the cornea blink reflex. Vi/Vc neurons may be the second-order neurons in thethree-neuron cornea reflex blink circuit, whereas Vc/C1neurons may inhibit cornea reflex blink amplitude. An-other goal of the current study is to investigate the roleof the Vc/C1 region in modifying cornea-evoked reflexblinks.

Several lines of evidence support the hypothesis thatneurons in the Vi/Vc region are second-order compo-nents of the cornea blink reflex circuit. There is a directinput to this region from cornea primary afferents(Marfurt and Del Toro 1987). Neurons in the Vi/Vcregion respond to cornea stimulation and approximatelyhalf of these cornea-responsive neurons project to thefacial nucleus and/or the superior salivatory nucleus, thelacrimal gland preganglionic neurons (Hirata et al. 2000;Hirata et al. 2004). Finally, studies on humans (Ong-erboer de Visser 1983b; Ongerboer de Visser 1983a) androdents (Pellegrini et al. 1995) report that the Vi/Vc re-gion is essential for generation of cornea-evoked reflexblinks.

Physiological evidence indicates that cornea-responsive neurons in the Vc/C1 region inhibit theactivity of the rostral Vi/Vc neurons. Intravenouslyadministered morphine inhibits the response of thecaudal Vc/C1 neurons to cornea stimulation, butfacilitates most of the Vi/Vc neurons (Meng et al.1998; Hirata et al. 2000). One interpretation of thesedata from opiate injections is that Vc/C1 neurons di-rectly inhibit Vi/Vc neurons. Consistent with thisproposal, Hirata et al. (2000) report that Vi/Vc stim-ulation antidromically activates some cornea respon-sive Vc/C1 neurons. Vc/C1 inhibition of Vi/Vcneurons may also be indirect. Approximately 80% ofcornea-responsive Vc/C1 neurons project to the par-abrachial area (Hirata et al. 2000). Parabrachial areaactivation inhibits the activity of Vi/Vc neuronsevoked by cornea stimulation (Meng et al. 2000). Ifthe Vc/C1 region inhibits the activity of the Vi/Vcneurons in the three-neuron reflex blink circuit, thenactivation of the Vc/C1 region should suppress cornea-evoked reflex blinks. We test this hypothesis directly

by investigating the effect of Vc/C1 microstimulationon cornea-evoked reflex blinks.

The cornea afferent termination pattern in the spinaltrigeminal complex and inhibition of Vi/Vc by Vc/C1neurons is a possible explanation of paired-stimulussuppression of cornea-evoked blinks. The first corneastimulus in the paired-stimulus paradigm would gener-ate a blink by exciting Vi/Vc neurons in the blink reflexcircuit and inhibit subsequent cornea-evoked blinks byactivating Vc/C1 inhibition of Vi/Vc blink reflex neu-rons. We test the hypothesis that Vc/C1 neurons pro-duce cornea blink suppression in the paired-stimulusparadigm by replacing the first cornea stimulus in thepaired-stimulus paradigm with Vc/C1 stimulation. If Vc/C1 neurons create blink suppression in the paired-stimulus paradigm, then the effects of Vc/C1 activationand the condition stimulus in the paired-stimulus para-digm should be equivalent.

Methods

Thirty-two male, 175–400 g, Sprague–Dawley rats weremaintained on a 12-hour light/dark cycle and fed adlibitum. All procedures adhered to Federal, State, andUniversity guidelines concerning the use of animals inresearch and received approval from the UniversityIACUC.

Experimental setup

Rats were anesthetized with Urethane (1.2 g kg�1 ip)and Xylazine (1 mg kg�1 im) anesthesia, placed in astereotaxic frame and prepared for orbicularis oculielectromyogram (OOemg) recording and trigeminalnucleus stimulation. To access the medullary brainstem,the dorsal brain stem was exposed and a portion of theC1 vertebral bone removed. In some animals, theoccipital bone was also partially removed. The brainstem was kept moist with warm saline.

For both eyelids, the lid-closing orbicularis oculi(OO) muscle EMG activity was recorded using a pair ofTeflon-coated, stainless steel wires (0.003’’ bare, 0.0055’’coated; AM Systems) implanted in the OO muscle at themedial and lateral margins of the upper eyelid. A silverwire (0.015’’ bare; AM Systems) was placed in the neckmuscles to serve as a reference ground.

A pair of Teflon-coated, silver wires (0.008’’ bare,0.011’’ coated; AM Systems) were used to stimulate thecornea electrically. The insulation was removed fromone end of each wire and bent into a 1.5-mm diameterloop. The centers of the two loops were separated by 2–3mm and rested lightly on the cornea surface. The blink-evoking stimulus was a single (0.1 ms duration),monophasic electrical current pulse ranging from 0.3–3.0 mA. Excess tears were periodically removed fromthe cornea with a cotton wick to reduce current shortingbetween the electrodes. If a blink moved the stimulating

446

electrodes, the electrodes were repositioned to evoke areliable blink.

A glass microelectrode (Corning #7740, AM Systems;capillary filament 2.0 mm O.D.·1.16 mm I.D.) filledwith 2 mol L�1 NaCl saturated with fast green FCF(Fischer Scientific) and its tip broken to produceimpedance between 5 and 30 MW was used to micr-ostimulate the Vc/C1 region. The pipette was located inthe caudal Vc/C1 border region 2.0 to 4.0 mm posteriorto the obex, 1.0 to 1.5 mm lateral from midline and 0.5to 1.5 mm below the brainstem surface using a posteri-orly directed 40 degree angle.

Experimental protocols

Blinks were investigated using the paired-stimulus par-adigm. In this paradigm, identical pairs of blink-evokingstimuli were given to either the same cornea (Ipsi–Ipsi)or to one cornea preceded by stimulation of the con-tralateral cornea (Contra–Ipsi). The interstimulusintervals between pairs of stimuli were 75, 100, 125, 150,200 or 225 ms. Paired-stimulus trials were presentedevery 40 s.

The first stimulus in the paired-stimulus paradigmwas replaced with Vc/C1 stimulation to test the role ofVc/C1 in paired-stimulus modulation of cornea-evokedblinks. Vc/C1 microstimulation was a 70 ms train(200 Hz, 0.1 ms duration stimuli) with a current inten-sity between 5 and 25 lA, presented 75 ms before theonset of the cornea stimulus. This Vc/C1 stimulus trainterminated 5 ms before presentation of the blink-evok-ing cornea stimulus (Basso et al. 1996). To determine thelatency of Vc/C1 effects on cornea-evoked blinks, asingle, 80 lA stimulus (0.1 ms duration) was deliveredduring the blink. In both the single stimulus and trainmicrostimulation conditions, trials with and withoutmicrostimulation were alternated, with 20 s betweeneach trial.

Five rats received a 6 mg kg�1 subcutaneous injec-tion of the 5-HT receptor antagonist metergoline (SigmaRBI, 0.1 mol L�1 in a 5.0% ascorbic acid solution;(Middlemiss and Tricklebank 1992) to investigate therole of serotonin in Vc/C1 blink suppression. Aftercollecting pre-drug data, each rat was injected withmetergoline. Data collection began 30 min after druginjection and continued for 30 min.

At the end of each experiment, fast green wasdeposited at the microstimulation site (Thomas andWilson 1965) to confirm electrode placement. The dee-ply anesthetized animal was perfused intracardially with6% dextran in 0.1 mol L�1 phosphate buffer followedby 10% formalin in 0.1 mol L�1 phosphate buffer. Thebrainstem, cerebellum and spinal cord to the C2 levelwere removed and placed in 30% sucrose in 0.1 mol L�1

phosphate buffer for 24 h. The brains were frozen, cutinto 100-lm sections, mounted on subbed slides andstained with cresyl violet. Vc/C1 microstimulation sites

were identified and transferred to a standardized seriesof brain and spinal cord drawings.

Data analysis

Blinks were measured from the rectified OOemg signalusing laboratory-developed software. OOemg signalswere amplified (AM systems, model 1700, 4-channeldifferential amplifier), filtered at 0.3–5 k Hz, collected ateither 10 or 4 kHz (Data Translation DT 2831, 12-bitA/D resolution) and stored for off-line analysis. TheOOemg latency, duration, and amplitude were deter-mined by marking the start and end of the blink andthe stimulus artifact (Fig. 1). Blink amplitude wasdetermined by integrating the rectified OOemg activitybetween the beginning and end of the blink.

In the paired-stimulus experiments, changes in blinklatency, duration, and amplitude were determined bycomparing values for the blink elicited by the second,Test (T) stimulus to the values for the blink evoked bythe first, Condition (C) stimulus in a single trial. In theContra–Ipsi condition, the Test blink was compared tothe Condition blink evoked ipsilateral to the Test stim-ulus in the preceding trial. To quantify the effect of thepaired-stimulus paradigm on blink amplitude, Condi-tion blink amplitude was subtracted from the Testamplitude and the difference divided by the sum of theCondition and Test blink amplitude ((T�C)/(T+C)) tocreate a normalized difference score. If the Conditionstimulus and/or blink completely suppressed the Testblink, then the normalized difference score would equal�1. If there were no effect on Test blink amplitude, thenthe value would be 0. A positive value indicated that theTest blink was larger than the Condition blink. Toquantify the effect of the paired-stimulus paradigm on

Fig. 1 The rectified OOemg activity of a blink evoked by a corneastimulus (solid triangle). Blink latency is the time from the corneastimulus until the onset of OOemg activity (Latency). Blinkduration is the period of OOemg activity evoked by the corneastimulus (Duration). The trace is a single trial

447

blink duration and latency, the Condition blink valuewas subtracted from the Test blink value (T�C). Theeffects of the Condition cornea stimulus on Test blinkswere tested for significance with a MANOVA repeatedmeasures with Bonferroni post-hoc analysis for intervaleffects.

Vc/C1 microstimulation effects were compared be-tween pairs of sequential trials in which a response waselicited by identical cornea stimulation in both trials. Vc/C1 microstimulation occurred on only one of a pair oftrials. If data from either the trial with or the trialwithout Vc/C1 microstimulation were unusable (e.g. thelid moved the electrodes or tears shorted the cornea-stimulating electrodes), both trials of the pair werediscarded. The same analyses employed for the paired-stimulus paradigm were used for these microstimulationexperiments in which the blinks without precedingmicrostimulation (Control) were compared with theblinks with microstimulation (Test). Significance of thechanges was evaluated with a MANOVA repeatedmeasures with Bonferroni post-hoc analysis comparingthe values of the blink with and without Vc/C1microstimulation.

The latency of blink suppression produced by Vc/C1stimulation was determined by comparing changes in theaveraged OOemg activity for paired trials with andwithout a single 80 lA Vc/C1 stimulus delivered duringthe blink (Fig. 6, below). Each point of the averagedOOemg activity was smoothed using a weighted averageof the 10 values on either side of that point. Thus, withthe 10 kHz sampling rate in this experiment, each OO-emg value was smoothed over 2 ms. To establish thebaseline level of difference between the two OOemgsmoothed averages with and without Vc/C1 stimulation,the OOemg activity from the time after the corneastimulus until two ms before delivering the Vc/C1stimulation were compared. For this control period, thedifference between the OOemg record with and withoutthe Vc/C1 stimulus was calculated. These differenceswere averaged and the standard deviation calculated.The latency of a significant change in blink amplitudeproduced by Vc/C1 stimulation was defined as the timewhen the difference between the two blink averages afterVc/C1 stimulation was continuously two standarddeviations above or below the average control differencefor at least 3 ms.

Results

Characteristics of cornea-evoked blinks

Electrical stimulation of the cornea in urethane-anes-thetized rats evoked a burst of OOemg activity with anaverage latency of 16.2±0.2 ms (SEM; n=753) and anaverage duration of 50.2±0.5 ms (n=753; Figs. 1, 2, 3,4). In 98% of the trials, the cornea stimulus only elicitedOOemg activity ipsilateral to the stimulated cornea.Because contralateral OOemg activation was only pres-

ent on 2% of the trials and the OOemg amplitude wasvery small in those trials, contralateral blinks were notquantified. For the ipsilateral blink, there was consid-erable variability in the pattern of OOemg activity pro-duced by identical cornea stimuli.

The averaged pattern of OOemg activity evoked by acornea stimulus varied among rats, but was typicallybipartite (Figs. 1, 2, 3, 4). The averaged OOemg re-sponse could be two nearly equal amplitude bursts ofOOemg activity (Fig. 2A), an initial small burst followedby a large burst of OOemg activity (Fig. 2B) or an initiallarge burst of OOemg discharge followed by a smallerburst of OOemg activity (Fig. 2C). For each animal, thepattern of OOemg activity evoked by identical corneastimuli also varied from trial to trial (Figs. 2D–H). Someblinks had the appearance of ragged peaks of OOemgactivity spread out over the entire response (Figs. 2D,H). Other blinks exhibited double (Figs. 2E, F) or eventriple (Fig. 2G) peaks of OOemg activity. Althoughbipartite, the averaged OOemg activity evoked by aconstant intensity cornea stimulus (Fig. 2I) did not re-veal the variability among individual responses.

Paired-stimulus paradigm

Presentation of a pair of identical cornea stimuli with aninterstimulus interval less than 300 ms resulted in sup-pression of the blink evoked by the second, Test, stim-ulus relative to the blink elicited by the first, Condition,stimulus (Figs. 3, 4A). Blink amplitude suppression wasmaximum when there was 100 ms between the stimuli.At longer intervals, the condition stimulus produced lesssuppression of the Test blink (Fig. 3C, solid circles). Thesuppression of the Test blink in the paired-stimulusparadigm was not uniform across the entire period ofOOemg activity. At short intervals, the initial OOemgactivity of the Test blink was typically less suppressedthan later activity (Figs. 3A). This suppression of lateractivity caused a significant decrease in the duration ofthe Test blink relative to the Condition blink for inter-stimulus intervals less than 175 ms (Fig. 3D, solid cir-cles). Consistent with the restricted effect of theCondition stimulus on the initial component of TestOOemg activity, changes in Test blink latency weresmall and inconsistent (not shown).

In this urethane-anesthetized preparation, suppres-sion of the Test blink in the paired-stimulus paradigmonly occurred with delivery of both stimuli to the samecornea (Fig. 4). For example, suppression of the leftOOemg blink evoked by the Test stimulus only tookplace when the left cornea received the Condition stim-ulus (Fig. 4A). This result held true for all interstimulusintervals tested (Fig. 3C, empty circles). Similarly, Testblink duration did not change in the Contra–Ipsi para-digm (Fig. 3D, empty circles). These data demonstratedthat the inhibition produced by the paired-stimulusparadigm was only ipsilateral to the stimulated cornea inurethane-anesthetized rats. To investigate whether Test

448

blink suppression in the paired-stimulus paradigm re-sulted from activation of Vc/C1 neurons by corneaafferents, we replaced the Condition cornea stimuluswith a train of ipsilateral Vc/C1 stimulation.

Vc/C1 microstimulation

To determine whether Vc/C1 microstimulation sup-pressed cornea-evoked blinks, we presented a 70 msmicrostimulation train of 200 Hz, 0.1 ms durationstimuli in the Vc/C1 region that terminated five ms be-fore the cornea stimulus. This Vc/C1 stimulus train re-duced the amplitude of cornea-evoked blinks (Figs. 5, 6,7, 8). In four rats tested over a range of stimulusintensities, increasing current intensity from 5 to 20 lAreduced blink amplitude relative to blinks evoked with-out preceding Vc/C1 microstimulation (Fig. 5A–D, da-shed lines). Five to twenty-five lA microstimulationtrains significantly reduced blink amplitude (Fig. 5E).For the rat illustrated in Figs. 5A–D, Vc/C1 microsti-mulation also evoked ipsilateral OOemg activity at 15

and 20 lA (Figs. 5C, D). Nevertheless, OOemg activa-tion with Vc/C1 microstimulation only occurred in six oftwenty-seven rats tested at 20 lA. Vc/C1 microstimula-tion effects on blink duration and latency were variable.For example, for the four animals tested over the entirestimulation intensity range, a significant decrease inblink duration occurred only at the 15 lA stimulusintensity (15 lA; Fig. 5F). In another group of twentyrats only tested with the 20 lA stimulation train, how-ever, this stimulus intensity significantly reduced blinkduration (Fig. 8D). Vc/C1 microstimulation at 20 and25 lA produced small, but statistically significant in-creases in blink latency relative to blinks without Vc/C1stimulation (Fig. 5G). In the experiment with only20 lA stimulus trains, the latency shift was still small(–3.13±0.44 ms; not illustrated), but was over 2 msmore than the increase in the intensity series experiment.

The Vc/C1 stimulus train experiment did not estab-lish the latency of Vc/C1 effects on OOemg activity. Todetermine the latency of blink suppression, a singlestimulus (80 lA, 100 ls duration) was delivered duringthe blink (Fig. 6). This stimulus transiently increased

Fig. 2 Patterns of OOemgactivity evoked by corneastimulation (solid triangle).(A–C) Each trace is the averageof ten rectified OOemg recordsfrom a different rat. (D–H) Fora single rat, records ofindividual blinks evoked by aconstant intensity corneastimulus and (I) the average oftrials D–H

449

ipsilateral OOemg activity (Fig. 6 solid squares, Ipsi-lateral) and subsequently suppressed the blink bilaterally(Fig. 6, empty circles). The ipsilateral blink activationbegan an average of 3.4±0.5 (n=5) ms after Vc/C1stimulation. After the Vc/C1 stimulus, blink suppressionbegan an average of 8.3±0.3 ms (n=5) ipsilateral and8.5±0.9 ms contralateral to the site of brainstem stim-ulation. The latency of suppression for ipsilateral andcontralateral blinks was not significantly different.Nevertheless, the excitatory response might have ob-scured the initiation of the ipsilateral inhibitory re-sponse. It was possible that the inhibitory responsebegan earlier than the contralateral inhibition. In eithercase, a single Vc/C1 stimulus delivered during the blinkreplicated the effects of 70 ms Vc/C1 stimulation trainsdelivered before the cornea stimulus.

One factor confounding the conclusion that activat-ing neurons in the Vc/C1 produced blink suppressionwas that Vc/C1 microstimulation may have antidromi-cally activated serotonergic, nucleus raphe magnus(NRM) axons terminating in the Vc/C1 region (Li et al.1997) which can suppress cornea-evoked blinks (Bassoand Evinger 1996). To test whether Vc/C1 microstimu-lation suppresses blinks by antidromically activating theserotonergic NRM axons identified by Basso and Ev-inger (1996), 5-HT neurotransmission was blocked witha systemic injection of 6 mg kg�1 of the 5-HT receptor

antagonist, metergoline, and the effect of Vc/C1 stimu-lation on cornea-evoked blinks was tested. This dose ofmetergoline blocks NRM blink suppression (Basso andEvinger 1996), but did not affect blink suppressioncaused by Vc/C1 stimulation (Fig. 7A). The normalizeddifference scores for blink amplitude with and withoutVc/C1 microstimulation were not significantly altered by5-HT receptor blockade. It was possible to ensure thatmetergoline effectively blocked 5-HT receptors byexamining blink amplitude before and after drugadministration. Because activation of facial motoneuron5-HT receptors increases motoneuron excitability(McCall and Aghajanian 1979; Hattox et al. 2003), 5-HTreceptor blockade should have reduced blink amplitudebilaterally. As expected, metergoline significantly re-duced blink amplitude by 67.6% ipsilateral and by58.6% contralateral to the Vc/C1 stimulation electrode(Fig. 7B). Thus, blink suppression caused by Vc/C1microstimulation did not involve activation of 5-HTfibers described by Basso and Evinger (1996).

Consistent with the hypothesis that paired-stimulussuppression of cornea-evoked blinks uses Vc/C1 neuronsto suppress Test blinks, 20 lA Vc/C1 stimulation trainsproduced approximately the same level of Test blinksuppression (�0.62; Fig. 5E) as the maximal paired-stimulus suppression (-0.57; Fig. 3C, solid circles,100 ms interstimulus interval). If blink suppression in

Fig. 3 Suppression of Testblinks in the paired-stimulusparadigm. (A, B) The blinkevoked by the Condition corneastimulus (dotted line)superimposed on the blinkevoked by an identical Teststimulus (solid line) at 75 ms (A)and 175 ms (B) interstimulusintervals. The triangle indicatesthe cornea stimulus. Each traceis the average of five rectifiedOOemg records from one rat.(C) The effect of the Conditionblink stimulus on Test blinkamplitude (T–C)/(T+C) as afunction of the interstimulusinterval, and (D) the meandifference between Test andCondition blink latency as afunction of interstimulusinterval for blinks evoked in theIpsi–Ipsi (solid circles) andContra–Ipsi (empty circles)paradigm. Error bars are±SEM. Significant differencesbetween Test and Conditionblinks determined withMANOVA and Bonferronipost-hoc analysis are indicatedwith asterisks (*P<0.05,**P<0.005, ***P<0.001)

450

the paired-stimulus paradigm results from activation ofVc/C1 neurons, then replacing the Condition blinkstimulus with a 20 lA Vc/C1 microstimulation trainshould have produced identical modulation of a Testblink as the paired-stimulus paradigm. Unlike thepaired-stimulus paradigm (Fig. 4), however, Vc/C1microstimulation suppressed cornea-evoked blinksbilaterally (Fig. 8). Although a blink only occurredipsilateral to the cornea stimulus, Vc/C1 microstimula-tion suppressed blinks evoked by stimulation of eithercornea (Fig. 8). For all 20 rats tested, the 20 lA Vc/C1microstimulation train significantly reduced the ampli-tude of blinks evoked from stimulation of the corneaipsilateral (Fig. 8C, black bar) and contralateral(Fig. 8C, gray bar) to the microstimulation site. Con-comitant with the reduction in blink amplitude, Vc/C1microstimulation significantly decreased blink durationipsilateral (Fig. 8D black bar) and contralateral(Fig. 8D, gray bar) to the Vc/C1 stimulation site. Blink

latency for ipsilateral and contralateral cornea-evokedblinks exhibited a small but significant increase (notillustrated). Vc/C1 microstimulation produced a signifi-cantly greater effect on blinks evoked ipsilateral to theVc/C1 stimulus than on contralateral blinks for ampli-tude, duration, and latency. The bilateral effect of Vc/C1activation on cornea-evoked blinks was a qualitativedeparture from blink modulation created by the paired-stimulus paradigm.

Vc/C1 stimulation sites

There was a rough topography of the Vc/C1 regionregarding the effectiveness of stimulation in suppressingcornea-evoked reflex blinks (Fig. 9). The most effectivemicrostimulation sites (normalized differencescore>�0.85, solid and empty squares) were primarilylocated ventrally in layer III near the border of layer II,throughout the region 2 to 4 mm caudal to the obex.Consistent with the explanation for the short-latencyipsilateral OOemg activity generated by a 80 lA stimu-lus (Fig. 6), five out of the six locations where 20 lAmicrostimulation elicited OOemg activity during thestimulation train (Figs. 5C, D) were located laterally,near the spinal trigeminal tract (Fig. 9, solid squares andtriangles).

Discussion

Cornea-evoked blinks

With urethane anesthesia, there is variability in thepattern of OOemg activity evoked by an electrical cor-nea stimulus, but cornea stimulation elicits two peaks ofOOemg activity on average (Figs. 1, 2, 3, 4). Thisbipartite pattern of OOemg activity may reflect the ar-rival of Ad and C-fiber cornea afferent volleys to secondorder neurons in Vi/Vc. Based on their conductionvelocity and a 29 mm distance from the cornea to Vi/Vc(personal observations and Paxinos and Watson 1998),the initial Ad action potentials will reach the Vi/Vc re-gion 8.3 to 19.3 ms after the cornea stimulus. The initialC-fiber action potentials, however, will not arrive until16.1 to 116 ms after the cornea stimulus. The 20.1 msdifference between the mean arrival times for Ad and C-fiber afferents, 10.7 and 30.8 ms, is nearly equal to thetime between the two peaks in OOemg activity of cor-nea-evoked blinks. For the four rats illustrated in Fig. 2,the average time between the two peaks of OOemgactivity is 22 ms. Thus, the arrival of Ad and C-fibercornea afferents can provide the framework for thebipartite OOemg pattern of cornea-evoked blinks. Tri-geminal neurons projecting to the facial nucleus receiveinputs from both Ad and C-fiber cornea afferents (Hi-rata et al. 2000). The variability in OOemg patternsbetween and within rats (Fig. 2) may reflect inconsis-tency in the number and type of cornea afferents

Fig. 4 The Conditioning cornea stimulus and/or blink onlysuppresses ipsilateral blinks in the paired-stimulus paradigm. (A)The OOemg activity elicited by pairs of identical cornea stimuli(solid triangle) to the same cornea (Ipsi–Ipsi; Left OOemg, Leftcornea) showed suppression of the second blink. (B) The OOemgactivity evoked by pairs of cornea stimuli (solid triangle) with theCondition stimulus presented to the contralateral cornea (Contra–Ipsi; Left OOemg, Right cornea, Left cornea) showed no suppres-sion of the Test response. Each record is the average of five rectifiedOOemg records from the same rat

451

activated by a cornea stimulus and modulatory influ-ences acting on the blink circuit.

Cornea-evoked blinks are exclusively unilateral inurethane-anesthetized rats (Fig. 4), whereas cornea-evoked blinks in alert humans are bilateral (Accorneroet al. 1980; Ongerboer de Visser 1983b; Ongerboer deVisser 1983a; Berardelli et al. 1985; Cruccu et al. 1986;Cruccu et al. 1997; Esteban 1999; Acosta et al. 2001b;Aramideh and Ongerboer de Visser 2002). This dis-crepancy between humans and rodents probably reflectsthe effects of anesthesia rather than intrinsic differencesin the organization of cornea-evoked blinks in rodentsand primates. Urethane-anesthetized rats exhibit bilat-eral cornea-evoked blinks when the trigeminal complexexcitability is elevated (Basso et al. 1996). Thus, similar

circuitry for bilateral cornea blinks is in place in rodentsas well as in humans, but urethane anesthesia suppressesthe contralateral response in rodents.

Modification of cornea-evoked blinks

Although we predicted that Vc/C1 activation wouldproduce blink suppression in the paired-stimulus para-digm, different neural mechanisms seem to underlie thetwo forms of blink suppression. Both paradigms pro-duce similar effects on cornea-evoked blink duration andlatency. Nevertheless, the difference between the twoprocedures in terms of bilateral, Vc/C1 stimulation

Fig. 5 OOemg activity evokedby a cornea stimulus with (solidlines) or without (dotted lines) apreceding 70 ms Vc/C1stimulus train(microstimulation) at a (A) 5,(B) 10, (C) 15, or (D) 20 lAstimulus intensity. The triangleindicates cornea stimulation.Each trace is the averagerectified OOemg activity of 10trials from one rat. The effect ofVc/C1 microstimulationintensity on blink (E) amplitude(T–C)/(T+C), (F) duration(T–C) and (G) latency (T–C)relative to blinks on pairedtrials without Vc/C1microstimulation. Error barsare ±SEM. Significantdifferences between Test andCondition blinks determinedwith MANOVA andBonferroni post-hoc analysisare indicated with asterisks(*P<0.05, **P<0.01,***P<0.001)

452

(Figs. 6 and 8) or unilateral, paired-stimulus (Figs. 3and 4) suppression of blinking is compelling.

A possible explanation for the bilateral effect of Vc/C1 microstimulation is that Vc/C1 stimulation anti-dromically activated NRM axon collaterals. Vc receivesa dense serotonergic input from NRM (Li et al. 1997)and NRM microstimulation produces bilateral sup-pression of cornea-evoked blinks similar to that causedby Vc/C1 microstimulation (Basso and Evinger 1996).The results of the metergoline experiment, however, areinconsistent with this explanation of the bilateral effectof Vc/C1 stimulation. Systemic metergoline treatmentblocks NRM blink suppression (Basso and Evinger1996), but does not affect Vc/C1 blink suppression(Fig. 7). The difference in the latency of NRM sup-pression of OOemg, 4.4 ms (Basso and Evinger 1996)and Vc/C1 suppression, 8.3 ms (Fig. 6) also shows that

these are distinct mechanisms. Although the metergolineexperiment rules out the nucleus raphe magnus sup-pression of blinks identified by Basso and Evinger (1996)as a source of the Vc/C1 suppression, the experimentdoes not rule out a novel mechanism of raphe blinksuppression because metergoline does not block all ofthe neurotransmitters released by raphe neurons.

The 8.3 ms latency of Vc/C1 suppression of cornea-evoked blinks implies that Vc/C1 activation does notinhibit Vi/Vc neurons directly. If Vc/C1 suppressedblinks by directly inhibiting Vi/Vc neurons, then blinksuppression should occur within approximately 4 ms.Combining 0.75 ms synaptic latencies for the Vc/C1synapse on Vi/Vc neurons and for the Vi/Vc synapse onOO motoneurons with the 2 ms latency from an OOmotoneuron spike until OOemg activity (Pellegrini et al.1995), predicts a 3.5 ms latency for blink suppression.

Fig. 6 The effect of a single 80 lA Vc/C1 stimulus on a cornea-evoked blink. (A) A left Vc/C1 stimulus (Vc/C1 stimulus, firstdashed line) evoked a short latency facilitation (solid square)followed by blink suppression (empty circle) in a blink (solid line)evoked by left cornea stimulation (solid triangle) relative to theblink evoked without Vc/C1 stimulation (dotted line). (B) A left Vc/C1 stimulus (Vc/C1 stimulus, first dashed line) suppressed theOOemg activity (empty circle) in a blink (solid line) evoked by rightcornea stimulation (solid triangle) relative to the blink evokedwithout Vc/C1 stimulation (dotted line). Each trace is the filteredaverage of 20 rectified OOemg records

Fig. 7 (A) The blink suppression produced by Vc/C1 stimulationbefore (black bar) and after (gray bar) metergoline for blinksipsilateral (left columns) and contralateral (right columns) to theside of Vc/C1 stimulation. (B) The effect of 5-HT receptor blockadewith metergoline (gray bars) on Control blink amplitude relative toControl blinks before metergoline (black bars) ipsilateral (leftcolumns) and contralateral (right columns) to the side of Vc/C1stimulation. Error bars are ±SEM. Significant differences betweenTest and Condition blinks determined with MANOVA andBonferroni post-hoc analysis are indicated with asterisks (NS notsignificant; **P<0.01)

453

The ipsilateral OOemg activity evoked 3.4 ms after Vc/C1 supports this estimate (Fig. 6). Production of thisburst of OOemg activity probably occurs because cur-rent spread from the 80 lA Vc/C1 stimulus antidromi-cally excites primary cornea afferents in the adjacentspinal trigeminal tract. Activation of these afferents ex-cites Vi/Vc neurons projecting to OO motoneurons,which should produce OOemg activity approximately3.25 ms later. Thus, bilateral suppression generated byVc/C1 microstimulation seems to result from activationof a site outside the cornea blink circuit that, in turn,inhibits the cornea blink circuit.

One possible site is the parabrachial area. Eightypercent of Vc/C1 neurons project to the contralateralparabrachial area (Meng et al. 2000) and the parabra-chial area projects bilaterally to Vi/Vc (Yoshida et al.1997) and OO motoneurons (Morcuende et al. 2002).Because activation of the parabrachial area suppressesthe response of Vi/Vc neurons to cornea stimulation(Meng et al. 2000), Vc/C1 microstimulation may exciteparabrachial neurons that, in turn, bilaterally inhibit Vi/Vc pre-motoneurons.

Thus, the data in the current study identify Vc/C1activation as a novel modifier of cornea-evoked blinks.Cornea blink suppression caused by microstimulation ofVc/C1 neurons is different from blink inhibition causedby the paired-stimulus paradigm and NRM activation.It is too simplistic, however, to view the Vc/C1 regiononly as a site of blink suppression. Hirata et al. (2003)demonstrate that microinjecting muscimol to suppressVc/C1 neuron activity blocks most of Vi/Vc neurons’response to cornea stimulation but enhances theresponsiveness of others. Together, these data imply thatthe Vc/C1 region exerts multiple effects on blink circuits.The present data reveal that microstimulation of the Vc/C1 region suppresses blinks. This suppression probablyoccurs by activating regions outside the trigeminalcomplex that inhibit Vi/Vc neurons. The Hirata et al.(2003) data show that the Vc/C1 region appears tofacilitate Vi/Vc neurons through intratrigeminal con-nections to blink circuit neurons and inhibitory inter-neurons. Thus in alert animals, some kinds of corneainput to the Vc/C1 region may cause facilitation of Vi/Vc blink circuit neurons, whereas other kinds of cornea

Fig. 8 Vc/C1 microstimulationsuppresses cornea-evokedblinks bilaterally. (A)Stimulation of the left cornea(solid triangle) with (solid lines)and without (dotted line)preceding left Vc/C1microstimulation. (B)Stimulation of the right cornea(solid triangle) with (solid lines)and without (dotted line)preceding left Vc/C1microstimulation. Each trace isthe average of rectified OOemgactivity from ten trials from onerat. (C) The effect of Vc/C1microstimulation on the blinksuppression ((T–C)/(T+C))and (D) duration (T�C) ofblinks evoked ipsilateral (blackbars) or contralateral (graybars) to Vc/C1 stimulation.Error bars are ±SEM.Significant differences betweenTest and Condition blinksdetermined with MANOVAand Bonferroni post-hocanalysis are indicated withasterisks (***P<0.001 basedon MANOVA and Bonferronipost-hoc analysis)

454

afferents may activate Vc/C1 neurons that inhibit blinkcircuits.

Acknowledgements This work was supported by a grant from theNational Eye Institute to CE (EY07391).

References

Accornero N, Berardelli A, Bini G, Cruccu G, Manfredi M (1980)Corneal reflex elicited by electrical stimulation of the humancornea. Neurology 30:782–785

Acosta MC, Belmonte C, Gallar J (2001a) Sensory experiences inhumans and single-unit activity in cats evoked by polymodalstimulation of the cornea. J Physiol 534:511–525

Acosta MC, Tan ME, Belmonte C, Gallar J (2001b) Sensationsevoked by selective mechanical, chemical, and thermal stimu-lation of the conjunctiva and cornea. Invest Ophthalmol Vis Sci42:2063–2067

Agostino R, Berardelli A, Cruccu G, Stocchi F, Manfredi M(1987) Corneal and blink reflexes in Parkinson’s disease with‘‘on-off’’ fluctuations. Mov Disord 2:227–235

Aramideh M, Ongerboer de Visser BW (2002) Brainstem reflexes:electrodiagnostic techniques, physiology, normative data, andclinical applications. Muscle Nerve 26:14–30

Basso MA, Evinger C (1996) An explanation for reflex blinkhyperexcitability in Parkinson’s disease. II. Nucleus raphemagnus. J Neurosci 16:7318–7330

Basso MA, Powers AS, Evinger C (1996) An explanation for reflexblink hyperexcitability in Parkinson’s disease. I. Superior col-liculus. J Neurosci 16:7308–7317

Berardelli A, Cruccu G, Manfredi M, Rothwell JC, Day BL,Marsden CD (1985) The corneal reflex and the R2 componentof the blink reflex. Neurology 35:797–801

Cruccu G, Agostino R, Berardelli A, Manfredi M (1986) Excit-ability of the corneal reflex in man. Neurosci Lett 63:320–324

Cruccu G, Leardi MG, Ferracuti S, Manfredi M (1997) Cornealreflex responses to mechanical and electrical stimuli in comaand narcotic analgesia in humans. Neurosci Lett 222:33–36

Esteban A (1999) A neurophysiological approach to brainstemreflexes. Blink reflex. Neurophysiol Clin 29:7-38

Fig. 9 Locations of Vc/C1stimulating electrodes thatproduced suppression of corneablinks sorted by OOemgnormalized difference values(large empty squares and largesolid squares, <�0.85; emptytriangles and solid triangles,�0.75 to �0.84; empty circlesand solid circles, �0.65 to –0.74;star, �0.55 to �0.64; plussymbol, -0.45 to –0.54; andsmall solid square, no significantsuppression). Locations for Vc/C1 stimulation electrodes forexperiments that producedOOemg activity during themicrostimulation train aremarked as solid symbols (solidsquare, solid triangle, solidcircle). The location of eachdrawing is relative to the obex

455

Hattox A, Li Y, Keller A (2003) Serotonin regulates rhythmicwhisking. Neuron 39:343–352

Hirata H, Okamoto K, Tashiro A, Bereiter DA (2004) A novelclass of neurons at the trigeminal subnucleus interpolaris/cau-dalis transition region monitors ocular surface fluid status andmodulates tear production. J Neurosci 24:4224–4232

Hirata H, Takeshita S, Hu JW, Bereiter DA (2000) Cornea-responsive medullary dorsal horn neurons: modulation by localopioids and projections to thalamus and brain stem. J Neuro-physiol 84:1050–1061

Kimura J, Powers JM, van Allen MW (1969) Reflex response of theorbicularis oculi muscle to supraorbital nerve stimulation: studyin normal subjects and in peripheral facial paresis. Trans AmNeurol Assoc 94:288–290

Li JL, Kaneko T, Nomura S, Li YQ, Mizuno N (1997) Associationof serotonin-like immunoreactive axons with nociceptive pro-jection neurons in the caudal spinal trigeminal nucleus of therat. J Comp Neurol 384:127–141

Manfredi M, Berardelli A, Cruccu G, Fabiano F (1985) Cornealreflex in normal and pathological subjects. Rev Neurol (Paris)141:216–221

Marfurt CF, Del Toro DR (1987) Corneal sensory pathway in therat: a horseradish peroxidase tracing study. J Comp Neurol261:450–459

McCall RB, Aghajanian GK (1979) Serotonergic facilitation offacial motoneuron excitation. Brain Res 169:11–27

Meng ID, Hu JW, Benetti AP, Bereiter DA (1997) Encoding ofcorneal input in two distinct regions of the spinal trigeminalnucleus in the rat: cutaneous receptive field properties, re-sponses to thermal and chemical stimulation, modulation bydiffuse noxious inhibitory controls, and projections to theparabrachial area. J Neurophysiol 77:43–56

Meng ID, Hu JW, Bereiter DA (1998) Differential effects of mor-phine on corneal-responsive neurons in rostral versus caudalregions of spinal trigeminal nucleus in the rat. J Neurophysiol79:2593–2602

Meng ID, Hu JW, Bereiter DA (2000) Parabrachial area and nu-cleus raphe magnus inhibition of corneal units in rostral andcaudal portions of trigeminal subnucleus caudalis in the rat.Pain 87:241–251

Middlemiss DN, Tricklebank MD (1992) Centrally active 5-HTreceptor agonists and antagonists. Neurosci Biobehav Rev16:75–82

Morcuende S, Delgado-Garcia JM, Ugolini G (2002) Neuronalpremotor networks involved in eyelid responses: retrogradetransneuronal tracing with rabies virus from the orbicularisoculi muscle in the rat. J Neurosci 22:8808–8818

Mukuno K, Aoki S, Ishikawa S, Tachibana S, Harada H, HozumiG, Saito E (1983) Three types of blink reflex evoked bysupraorbital nerve, light flash and corneal stimulations. Jpn JOphthalmol 27:261–270

Ongerboer de Visser BW (1983a) Anatomical and functionalorganization of reflexes involving the trigeminal system in man:jaw reflex, blink reflex, corneal reflex, and exteroceptive sup-pression. Adv Neurol 39:727–738

Ongerboer de Visser BW (1983b) Comparative study of cornealand blink reflex latencies in patients with segmental or withcerebral lesions. Adv Neurol 39:757–772

Paxinos G, Watson C (1998) The rat brain in stereotaxic coordi-nates. Academic Press, San Diego

Pellegrini JJ, Horn AK, Evinger C (1995) The trigeminallyevoked blink reflex. I. Neuronal circuits. Exp Brain Res107:166–180

Powers AS, Schicatano EJ, Basso MA, Evinger C (1997) To blinkor not to blink: inhibition and facilitation of reflex blinks. ExpBrain Res 113:283–290

Shahani BT, Young RR (1972) Human orbicularis oculi reflexes.Neurology 22:149–154

Tackmann W, Ettlin T, Barth R (1982) Blink reflexes elicitedby electrical, acoustic and visual stimuli. I. Normal valuesand possible anatomical pathways. Eur Neurol 21:210–216

Thomas RC, Wilson VJ (1965) Precise localization of Renshawcells with a new marking technique. Nature 206:211–213

Yoshida A, Chen K, Moritani M, Yabuta NH, Nagase Y,Takemura M, Shigenaga Y (1997) Organization of thedescending projections from the parabrachial nucleus to thetrigeminal sensory nuclear complex and spinal dorsal horn inthe rat. J Comp Neurol 383:94–111

456