B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te /b ra in res

Research Report

General anesthesia as a factor affecting impulse activity andneuronal responses to putative neurotransmitters

François Windels, Eugene A. Kiyatkin⁎

Cellular Neurobiology Branch, National Institute on Drug Abuse-Intramural Research Program, National Institutes of Health, DHHS,333 Cassell Drive, Baltimore, MD 21224, USA

A R T I C L E I N F O

⁎ Corresponding author. Fax: +1 410 550 5553.E-mail address: ekiyatki@intra.nida.nih.go

0006-8993/$ - see front matter © 2006 Elsevdoi:10.1016/j.brainres.2006.02.064

A B S T R A C T

Article history:Accepted 19 February 2006Available online 5 April 2006

Although it is evident that general anesthesia should affect impulse activity andneurochemical responses of central neurons, there are limited studies in which theseparameters were compared in both awake and anesthetized animal preparations. We usedsingle-unit recording coupledwith iontophoresis to examine impulse activity and responsesof substantia nigra pars reticulata (SNr) neurons to GABA, glutamate (GLU), and dopamine(DA) in rats in awake, unrestrained conditions and during chloral hydrate anesthesia. SNrneurons in both conditions had similar organization of impulse flow, but during anesthesia,they have lower mean rates and discharge variability than in awake conditions. Inindividual units, discharge rate in awake, quietly resting rats was almost three-fold morevariable than during anesthesia. These cells in both conditions were highly sensitive toiontophoretic GABA, but the response was stronger during anesthesia. In contrast tovirtually no responses to GLU in awake conditions, most SNr neurons during anesthesiawere excited by GLU; the response occurred preferentially in slow-firing units, which wereatypical of awake conditions. Consistent with no postsynaptic DA receptors on SNr neurons,iontophoretic DA was ineffective in altering discharge rates in awake conditions, but ofteninduced weak excitations during anesthesia. Although SNr neurons are autoactive,generating discharges without any excitatory input (i.e., in vitro), their impulse activityand responses to natural neurochemical inputs are strongly affected by general anesthesia.Some alterations appear to be specific to the general anesthetic used, while others probablyreflect changes in the activity of afferent inputs, brain metabolism and neurotransmitteruptake that are typical to any type of general anesthesia. Therefore, an awake, freelymovinganimal preparation appears to be advantageous for studying impulse activity andneurochemical interactions at single-neuron level during physiologically relevantconditions.

© 2006 Elsevier B.V. All rights reserved.

Keywords:General anesthesiaSingle-unit recordingMicroiontophoresisGlutamateGABADopamineRat

1. Introduction

General anesthesia is commonly used in neurophysiologicalresearch to maintain stable single-unit recordings. While it

v (E.A. Kiyatkin).

ier B.V. All rights reserve

is known that most general anesthetic drugs powerfullyinhibit brain metabolism, block animal responses to envi-ronmental challenges, and affect numerous homeostaticfunctions (see Michenfelder et al., 1988 for review), the

d.

Table 1 – Impulse activity of SNr neurons in rats in awake,unrestrained conditions and during chloral hydrate-inducedanesthesia

Parameter Condition Difference(t value)

Control Anesthesia

n cells 74 60n tests 457 383X, imp/sRange 2.20–87.36 0.45–98.7Mean 25.79 22.16Mode 33 15ln-mean 3.02 ± 0.03 2.69 ± 0.05 t = 5.67 (***)

SD, imp/sRange 1.50–21.63 0.75–10.84Mean 5.78 3.38Mode 4.5 3.7ln-mean 1.63 ± 0.02 1.11 ± 0.03 t = 14.44 (***)

CV, %Range 5.64–136.67 2.75–247.09Mean 28.81 29.51Mode 27 25ln-mean 3.21 ± 0.03 3.01 ± 0.04 t = 4.00 (**)

** and *** refer to statistical significance P < 0.001 and P < 0.0001,respectively.

Fig. 1 – Distributions of major parameters of impulseactivity of SNr neurons recorded in awake, unrestrainedrats in drug-free conditions and during chloral hydrateanesthesia. Panel A, discharge rate, imp/s; Panel B, standarddeviation of impulse activity, imp/s; Panel C, coefficient ofvariation of impulse activity, %. Hatched lines showdistribution modal values.

105B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

impact of general anesthesia on neuronal activity andresponsiveness remains unclear due to a limited numberof parallel comparative studies in awake, unrestrainedanimals. To address this issue, single-unit recording coupledwith microiontophoresis was used in rats to compare theactivity of substantia nigra reticulata (SNr) neurons and theirresponses to GABA, glutamate (GLU), and dopamine (DA) inawake, freely moving conditions and during chloral hydrateanesthesia.

SNr neural cells serve as a good substrate for evaluatingthe influence of general anesthesia on activity and neuronalresponsiveness to endogenous transmitter substances. Al-though these cells are autoactive, maintaining spontaneousactivity without any excitatory input in vitro (Atherton andBevan, 2005; Hajos and Greenfield, 1994; Richards et al.,1997; Rick and Lacey, 1994), their activity state is modulatedby both GLU input mainly from the subthalamic nucleusand abundant GABA inputs from the striatum, pallidum,and intrinsic axonal collaterals (Celada et al., 1999; Collin-gridge and Davies, 1981; Deniau et al., 1982; Hammond et al.,1978). These cells are also exposed to the action ofdopamine (DA) that is released within the SNr fromdendrites of dorsally located SNc dopamine neurons (Gau-chy et al., 1987; Nieoullon et al., 1977). While SNr is aprimary output structure of basal ganglia that plays animportant role in movement regulation (Deniau et al., 1982;Fallon et al., 1995), most available data on neurochemicalresponses of these cells were obtained either in vitro or inanesthetized animal preparations (Hammond et al., 1978;Robledo and Feger, 1990; Schmitt et al., 1999; Waszczak andWalters, 1983, 1986). Therefore, in addition to clarifying theimpact of general anesthesia on SNr neurons, this study willbridge these analytical data with the results of monitoringtheir activity under behavioral conditions.

2. Results

Data were obtained from 134 spontaneously active cellshistologically verified to be located in SNr; 74 wererecorded in awake, unrestrained conditions and 60 duringanesthesia. In both conditions, recorded units had biphas-ic, relatively short-duration spikes (range, 1.23–2.23 ms;mean, 1.80 ± 0.14 ms) with no evident between-statedifferences.

106 B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

2.1. Impulse activity

Table 1 shows parameters of impulse activity of SNr neuronsrecorded in two conditions, and Figs. 1 and 2 depict the resultsof statistical analysis.

As can be seen in Figs. 1 and 2, under both conditions, SNrneurons had relatively high and variable rates and regularpattern of discharges, showing strong correlations betweenindividual parameters of impulse activity. Units with highermean rates (X) had larger standard deviations (SD) and smallercoefficients of variations (CV) of impulse activity, suggestingthat fast-firing cells have more regular pattern of discharges.In both conditions, all parameters of impulse activity weredistributed in the population according to ln natural law withrelatively close mean and modal values.

Therewere, however, between-state differences in impulseactivity in each analyzed parameter (Table 1 and Fig. 2). Inawake conditions, most units had a relatively high dischargerate (mode 32 imp/s) and the distribution slightly skewedtoward lower values. In contrast, during anesthesia, mostunits had a lower discharge rate (mode 15 imp/s), and therewere more units with slow-firing than in control conditions.The difference in discharge rate was highly significant. Incontrast, distribution of SD was symmetrical in both condi-tions and the largest between-group differences were in thisparameter (see Table 1). Significant but minimal between-group differences were seen in CV; most units in bothconditions had a regular discharge rate (mode 20–40%).

Strong between-group differences were also found inspontaneous variability of neuronal discharge rate. To char-acterize this parameter, we calculated basal discharge rates

Fig. 2 – Relationships between major parameters of impulse actin drug-free and anesthetized conditions. Each graph shows the rcoefficient of correlation. Hatched lines show mean values.

preceding each iontophoretic application and comparedthem as pairs (378 and 323 pairs for awake and anesthe-tized conditions, respectively). As shown in Fig. 3, dischargerate in most units slightly decreased or increased duringrepeated evaluations; these spontaneous fluctuations inboth conditions were equally distributed around zero line,suggesting variability of discharge rate as a feature typicalof both conditions. The amplitude of fluctuations, however,was two- or three-fold larger in awake conditions comparedto anesthesia (SD for absolute change: 10.13 imp/s vs.4.63 imp/s, P < 0.001; SD for relative change: 74.76% and25.24%, P < 0.001; Student's t test).

2.2. GLU responses

Responses to GLU were evaluated in 26 and 22 units recorded,respectively, in awake and anesthetized conditions (116 and110 tests). Typically, each cell was tested with several GLUapplications at the same or increasing currents (10–60 nA).

As shown in Figs. 4A and 5, most SNr units in anesthetizedanimals showed excitatory responses to GLU. Although thisresponse was dose-dependent in individual units (Fig. 5,original examples), excitations at each current were seen inabout 80% of the tests and the effect of GLU on discharge ratewas significant at each current (20 nA: F1,97 = 72.45 and 40 nA:F1,87 = 38.44; both P < 0.0001). As shown in Figs. 4B and C (alltests at 20 and 40 nA), excitations prevailed in slow-firing cellsand in these cases their relative magnitude was maximal[r = (−)0.817; P < 0.001]. The results of regression analysissuggest that the response to GLU disappears at ∼25–35 imp/s.While the response to GLUwas significant at each current, the

ivity of SNr neurons assessed in awake, unrestrained ratsegression lines, the line of no effect, regression equation, and

Fig. 3 – Spontaneous variability of discharge rate of SNr neurons in awake and anesthetized conditions. Top and bottom graphsshow absolute and relative changes in discharge rate following repeated evaluations in two conditions. As can be seen,discharge rates during anesthesia were much more stable than in awake conditions. See text for other explanations.

107B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

response magnitude increased with increase in ejectioncurrent. At 20 and 40 nA current, mean magnitude ofexcitation was 147 and 189%, respectively. GLU-inducedincrease in activity developed with ∼3 s onset latency,relatively saturated at the final stage of ejection, and hadafter-effects of about 5 s (Fig. 6A).

Surprisingly, GLU iontophoresis had virtually no effecton SNr neurons in awake animals (see Fig. 5). Although thenumber of significant increases was about two-fold largerthan the number of significant decreases (26 vs. 12), nosignificant changes were seen in most of tests at eachcurrent (Figs. 4A and B). The difference in the number ofexcitations in awake vs. anesthetized conditions wassignificant at each current (10 and 60 nA, P < 0.05; 20 and40 nA, P < 0.001). GLU-induced decreases were seen in 7units (12/84 tests), they alternated with no responses, andtheir incidence and magnitude were within a range seenfor spontaneous fluctuations in discharge rate (see Section2.1). Correlative analysis revealed that GLU in awakeanimals has no significant effect on impulse activity ofSNr neurons (Figs. 4B and C). This difference in responsewas especially evident with respect to relative magnitudeof GLU-induced activity (Fig. 4C) and time course of theGLU response (Fig. 6A).

Between-state differences in the effects of GLU wereevident for several statistical comparisons. First, the effect ofGLU, evaluated based on ANOVA values for all tests at 20–40 nA, was in awake animals (n = 84; 22.44 ± 1.44–25.73 ± 1.94 imp/s, F1,167 = 6.67, P < 0.02) about 15 times weaker

than in anesthetized conditions (n = 91; 12.12 ± 0.91–20.31 ± 1.01 imp/s, F1,185 = 105.3, P < 0.0001). Second, themean GLU-induced increase in activity in awake animals(114.66% of baseline = 100%) was 4–5 times weaker than inanesthetized animals (167.76% of baseline = 100%). Finally,this difference was also equally profound in ANOVA values forthe effect of GLU on discharge rate (see Fig. 6; F82,2572 = 2.04,P < 0.001 vs. F85,2665 = 45.98, P < 0.0001).

Since GLU had a much weaker group effect in awake vs.anesthetized conditions, we averaged only cases of statisti-cally significant increases in activity induced by GLU in bothgroups. As can be seen in Fig. 6B, despite large differences inbasal discharge rate, GLU increased activity in both cases witha similar time course. Although the effect was significant inboth groups, it was much stronger in anesthetized vs. awakeanimals (F71,2231 = 53.57 vs. F25,805 = 5.98). Therefore, it appearsthat GLU is able to activate certain SNr neurons in awakeconditions, but the number of such activations is muchsmaller than in anesthetized conditions.

2.3. GABA responses

The effects of GABAwere tested in 33 and 19 neurons in awakeand anesthetized rats, respectively; 131and 109 tests (5–60 nA)were analyzed.

As shown in Fig. 7, GABA was highly effective at inhibitingactivity of SNr neurons in both conditions (122/131 and 96/109significant decreases) at each current used. Although inindividual units the number of inhibitions slightly increased

Fig. 4 – GLU responses of SNr neurons in awake and anesthetized conditions. (A) Percent of excitations (+), inhibitions (−), andno responses (NS) following GLU applications at different ejection currents. n above each current = number of iontophoretictests. Asterisks in right graph mark values are significantly different from the awake condition (*P < 0.05; ***P < 0.001). (B, C)Absolute and relative changes in rates following GLU applications (20–40 nA) in both conditions. Regression analysis suggeststhat GLU is ineffective at influencing discharge rate in awake conditions but preferentially activates SNr neurons inanesthetized conditions.

108 B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

and their magnitude became stronger with an increase inejection current (see original examples in Fig. 5), the effect ofGABA was significant and strong at each current. The effect ofGABA in both conditions was independent of impulse activity(Fig. 7B).

Similar to GLU, the effects of GABA were stronger inanesthetized vs. awake animal. The difference was maximalat low currents (mean inhibition at 10 nA; 13.13 ± 4.83% vs.37.40 ± 7.54%; P < 0.01), weaker at moderate currents (20 nA:18.37 ± 3.03% vs. 36.10 ± 3.69%; P < 0.01), and virtuallydisappeared at high currents (40 nA: 21.66 ± 4.79% vs.23.39 ± 2.56%). These differences were also evident in theresponse time course (Fig. 7C). While in awake conditionsdischarge rate rapidly decreased at the beginning of ejection(peak at 3 s) and slowly restored to the ejection end, theinhibition was more stable and restoration more delayed inanesthetized conditions.

2.4. DA responses

The effects of DA (20–40 nA) were tested in 19 and 16 cellsduring anesthesia and waking conditions, respectively; 56and 84 tests were analyzed (see Fig. 5 for original examples).While in awake conditions, there were both weak increases(11/49) and decreases (8/11) in activity, most tests resulted inno significant changes (30/49) and the effect was absent atboth 20 and 40 nA currents (F24,49 = 0.58, P = 0.45 andF44,89 = 0.62; P = 0.43). No effect of DA iontophoresis was alsoconfirmed by results of the regression analysis (Figs. 8A andB) and the response time course (Fig. 8C; F43,1363 = 0.871,P = 0.668).

While in anesthetized animals most DA tests also did notresult in significant changes in activity, weak increasesoutnumbered decreases (24 vs.11; see Figs. 8A and B). Weakbut significant excitatory effect of DA was also found using an

Fig. 5 – Rate histograms of six different SNr neurons showing changes in activity induced by GLU, GABA and DA in awake andanesthetized conditions. Marks on the ordinates = discharge rate (10 imp/s) and on abscissa time (100 s). Applications areshown as the horizontal lines below the abscissa, and numbers depict ejection currents in nA. Marks without numbers meanthe same current as during the previous injection.

109B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

ANOVA of the response time course (see Fig. 8C; F29,929 = 2.68,P < 0.01 and original examples in Fig. 5). This effect manifestedas a slight, gradual increase in activity, peaking at the end ofejection. These “excitatory” effects of DA, moreover, weremore evident in units with higher discharge rate (Fig. 8B).Although these activity rates were atypical to anesthetizedanimals as a whole, they were seen relatively often in the unitsample tested with DA.

3. Discussion

This study revealed that general anesthesia has a powerfulinfluence on impulse activity of SNr neurons and responsesof these cells to GLU, GABA and DA. While some findingsagree with previous observations, other findings wereunexpected in light of previous iontophoretic and in vitro

application studies of these cells in anesthetized animalsand brain slices.

3.1. Impulse activity

The major parameters of impulse activity of SNr cells aredistributed in neuronal population according to ln normal law(Gaussian curve), and they are closely interrelated. Cells withhigher discharge rate had higher absolute (SD) and lowerrelative discharge variability (CV), while cells with lowerdischarge rate had lower SD and higher CV. Since lndistribution of major parameters of impulse activity andsimilarly tight relationships between individual parametersreflecting rate and discharge variability were previouslyreported in different central structures (i.e., thalamus, hypo-thalamus, central gray, striatum, VTA; see Kiyatkin, 1984;Kiyatkin and Rebec, 1998, 1999; Nakahama et al., 1971; Werner

Fig. 6 – Time course of changes in discharge rate following a 20-s application of GLU in awake and anesthetized conditions. Topgraphs (A) show the average of all iontophoretic applications (n = number of tests) at 20–40 nA, and bottom graphs (B) showthe average of significant increases for each group. Filled bars show values significantly higher (P < 0.05; Fisher test) than thelast pre-ejection value.

110 B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

et al., 1963; Werner and Mountcastle, 1965), they appear to becommon for different central neurons involved in processingof incoming information. Logarithmic transformation occursfrequently in biological systems as it allows for strongmodulation of the energetic range of transmitted messageswithout loss of their information value. In fact, this regularityis well known in sensory systems and was first substantiatedmore than 100 years ago in the Weber–Fechtner law, whichdescribes perceived sensation as a logarithmic function ofphysical stimulus intensity (see Werner and Mountcastle,1965). While in each brain structure there are equally strongrelationships between individual parameters of impulseactivity (X, SD, and CV), their quantitative relationships arespecific for each structure, showing some degree of variabilitywithin the same regression lines. As shown in Fig. 2, quitesimilar relationships between X, SD, and CV were typical ofSNr cells in both conditions and dramatic changes in rate anddischarge variability in anesthetized conditions occurredwithin the same regression lines as in awake conditions.

However, there were significant differences inmean valuesand distributions of these parameters. The largest differencewas found in SD, suggesting that absolute variability ofdischarge rate of SNr neurons is much higher in awake thananesthetized conditions. While differences were quantitative-ly smaller, mean discharge rate of SNr neurons in awakeanimals was also higher than in anesthetized ones. Despiteabout two-fold difference in modal values, both fast- and

slow-firing SNr neurons were found in anesthetized condi-tions. The number of slow-firing units, however, was higher,and the distribution of discharge rate was wider (compareFigs. 1A and 2B and D). Discharge rates of individual SNr units,moreover, were almost three-foldmore variable in awake thanin anesthetized conditions following repeated evaluations(see Fig. 3). Although lower discharge rates during anesthesiamay reflect both a direct inhibiting action of chloral hydrate onneuronal excitability and potentiation of inhibiting, GABA-mediated effects, shown in this study, the degree of anesthe-sia-related inhibition was relatively weak. This resistance ofdischarge rate well agrees with autoactivity of SNr neuronsand the crucial role of internal, pacemaker mechanisms indetermining their activity. Although anesthesia typically hasan inhibiting effect on neuronal discharge rate as shown forcortex and different subcortical structures and various anes-thetic drugs (Evans andNelson, 1973; Kuwada et al., 1989; Pateland Chapin, 1990; Syka et al., 2005; Zurita et al., 1994), anopposite change–increase in discharge rate–had been reportedin the medial thalamus during urethane anesthesia vs. quietrest in awake, unrestrained rats (Kiyatkin, 1984). Under thiscondition, discharge rate of thalamic cells was more regularand the distribution of discharge rate of individual units inpopulation wider than in awake rats. Therefore, the pattern ofanesthesia-related alterations in discharge rate may dependupon the type of anesthetic and be opposite in different brainstructures.

Fig. 7 – GABA responses of SNr neurons in awake and anesthetized conditions. A and B: absolute and relative changes in ratesfollowing GABA applications at different ejection currents. Each graph shows regression lines, number of inhibitions, andcoefficient of correlation for each current. Regression analysis suggests that GABA strongly inhibits SNr neurons in bothconditions, with equally strong effects at different discharge rates. All regression lines were located well below the line of noeffect and were almost parallel to this line. C: time course of changes in discharge rate following a 20-s application of GABA inawake and anesthetized conditions. Filled symbols show values significantly lower (P < 0.05, Fisher test) than the lastpre-ejection value.

111B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

3.2. GABA-induced inhibition of SNr neurons:anesthesia-related potentiation

Consistent with dense striato-nigral, pallido-nigral and axonalcollateral GABA inputs and abundance of postsynaptic GABAreceptors (Fallon et al., 1995; Smith and Bolam, 1989), SNr cellswere highly sensitive to iontophoretic GABA, showing dose-dependent, clear-cut, phasic inhibitions in both tested condi-tions. Significant GABA-induced inhibitions were seen in alltested units, at different ejection currents, and in most tests(122/131). While in individual units the effect became strongerwith increased ejection current (mean 37.5%, 31.4%, and 23.4%of baseline for 10, 20, and 40 nA currents in awake conditions),prevalence of inhibitions was evident at each current equal orhigher than 10 nA. This effect, moreover, was independent ofdischarge rate.

Unexpectedly, the effects of GABA during chloral hydrateanesthesia were stronger than those in the awake, non-anesthetized animals. Although fast-firing cells had a ten-dency to be more susceptible to the inhibiting action of GABAand the number of such units was lower in anesthetizedanimals, between-state differences in the effects of GABAwere evident in terms of both absolute and relative responsemagnitudes. This difference was especially pronounced at lowcurrents (13.3% vs. 37.5% for 10 nA) and almost disappeared athigh currents. Anesthesia also affected the time course of theGABA response. Under awake conditions, the GABA-inducedinhibition rapidly peaked at the start of application andbecame smaller at its end (fading effect), while duringanesthesia the inhibiting effect was more tonic and hadlonger offset latencies. While the reasons for these differencesremain unclear, two factors may contribute to stronger effects

Fig. 8 – DA responses of SNr neurons in awake and anesthetized conditions. (A, B) Absolute and relative changes in ratesfollowing DA applications at different ejection currents (20 and 40 nA). Each graph shows regression lines, regressionequations, number of responses, and coefficient of correlation for each current. Regression analysis suggests that DA has noeffect on activity of SNr neurons in awake animals but induces weak excitation in a subset of these cells in anesthetizedanimals. C: time course of changes in discharge rate following a 20-s application of DA at 40 nA current in awake andanesthetized conditions. Filled symbols show values significantly higher (P < 0.05; Fisher test) than the last pre-ejection value.The effect of DA on discharge rate was absent in awake conditions, but a weak excitatory effect was seen during anesthesia.

112 B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

of GABA during anesthesia. First, any general anesthesiainhibits to a different extent both metabolism of a neuro-transmitter and its uptake (Adachi et al., 2001), thus prolong-ing the act ion of GABA on its receptors . Sinceneurotransmitter uptake is highly temperature-dependent(Xie et al., 2000), anesthesia-related brain hypothermia mayalso contribute to stronger cellular effects of GABA. Althoughwe used body warming, this procedure is unable to fullycompensate for brain hypothermia related to drug-inducedmetabolic inhibition (Kiyatkin and Brown, 2005). The secondfactormay be specific to chloral hydrate, which ismetabolizedinto trichloroethanol that is known to enhance chloridecurrents induced by GABA, thus reinforcing its inhibitoryaction (Lovinger et al., 1993). Independent of the mechanisms

responsible for the enhanced GABA sensitivity of SNr neuronsduring anesthesia, this influence appears to be an importantfactor for lowered discharge rate of these cells.

3.3. An apparent lack of GLU sensitivity of SNr neurons inawake animals

Although many previous studies demonstrated that SNrneurons are excited by iontophoretic GLU in anesthetizedanimal preparations (Hammond et al., 1978; Robledo andFeger, 1990; Schmitt et al., 1999; Waszczak and Walters, 1983,1986), these cells are much less sensitive to GLU in awake,unrestrained animals. Many of these same cells, however,showed GLU-induced excitations during chloral hydrate

113B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

anesthesia.While low sensitivity of SNr neurons to GLU agreeswith relatively sparse GLU innervation of this structure fromsubthalamic nucleus and a limited number of GLU receptors(Fallon et al., 1995), the reasons for such drastic between-statedifferences in responsiveness to GLU remain unclear.

It is unlikely that the awake preparation by itself mayexplain these differences since in other areas, such asstriatum, GLU consistently activates most neurons (Kiyatkinand Rebec, 1999). The GLU-induced excitations of striatalneurons, moreover, are strongly dependent on neuronaldischarge rate, with high-magnitude excitations in slow-firingunits and response saturation at rates of 40–60 imp/s.Therefore, lower discharge rate of SNr neurons (and possiblydiminished GLU input) in anesthetized conditions maydetermine their stronger responses to GLU. Under theseconditions, the GLU response evaluated within a neuronalpopulation was strongly dependent upon discharge rate.Excitations were seen in most slow-firing cells, while theresponse disappeared at rates about 20–25 imp/s (see Fig. 4C).The GLU-induced excitation, however, was relatively weak interms of magnitude, with slow acceleration and plateau atabout 24 imp/s, the level that corresponded to the mean basalrate of these cells in an awake preparation.

Between-state differences in discharge rate, however,cannot fully explain differences in GLU responsiveness.Although in awake conditions GLU was virtually ineffectiveto excite SNr neurons as a group, a small number of significantincreases and decreases in activity were observed during GLUiontophoresis. Most of these changes were weak in amplitudeand similar to spontaneous fluctuations in discharge ratetypical of SNr neurons in these conditions. Increases inactivity, however, occurred three times more often thandecreases, and their amplitude was twice the magnitude ofinhibitions. When these increases were averaged as a groupand compared to a similar group in anesthetized conditions,the change in activity following GLU application followed thesame temporal pattern in awake and anesthetized animals(Fig. 6B). It appears, therefore, that some neurons undercertain circumstances are sensitive to GLU in awake condi-tions, but the number of such units is low.

3.4. The action of DA on SNr neurons

SNr neurons in awake, unrestrained rats were not sensitiveto iontophoretic DA applied at different ejection currents(20–40 nA). While this finding agrees with the lack ofpostsynaptic DA receptors on SNr neurons (Yung et al.,1995), it is unclear why DA-induced excitations have beenpreviously reported on these cells in anesthetized animals(Ruffieux and Schultz, 1981; Waszczak and Walters, 1983,1986) and why weak excitations were reproduced in thepresent study during chloral hydrate anesthesia. Interest-ingly, increases in discharge rate induced by iontophoreticDA were found in about 39% of SNr neurons during chloralhydrate but only in 8% during urethane anesthesia (Ruffieuxand Schultz, 1981).

Between-state differences in unit responses to DA may bepartially related to the statistical procedures used for evalu-ating iontophoretic responses. In contrast to arbitrary criteriaof 15–30% change in discharge rate used inmost iontophoretic

studies, each iontophoretic test in our study was evaluatedstatistically and the response was recognized when meandischarge rate either significantly increased or decreased.While the use of this universal criterion is important forbetween-group comparisons, significant increases ordecreases in rate induced by iontophoretically applied drugdo not alwaysmean neuronal excitation or inhibition andmayreflect spontaneous fluctuations in discharge rate. As shownin Fig. 3, these fluctuations are more pronounced in slow-firing SNr units, they are equally distributed in upward anddownward directions, but they are almost three-fold larger inawake vs. anesthetized conditions. While this statisticalapproach for evaluating iontophoretic responses works wellwith substances that have a powerful effect on discharge rate,it shows some degree of pseudo-excitations and inhibitionswhen substances have weak effects on discharge rate. Forexample, from 240 GABA applications in both awake andanesthetized animals, 218 resulted in significant decreaseswith only one case of significant increase. Similarly, 95 of 110applications of GLU in anesthetized rats resulted in significantrate increases with no decreases at all. In contrast, 19 of 116GLU tests in awake conditions produced atypical decreases,which obviously reflect spontaneous fluctuations in dischargerate, when the excitatory action of GLU is absent. While mostDA applications did not affect discharge rate, both weakincreases and decreases were seen in anesthetized and awakeanimals. In awake animals, the ratio was almost equal (± = 19/16), but increases prevailed over decreases (24/11) in anesthe-tized animals and ANOVA revealed a significant group effect(F35,111 = 9.32, P < 0.001). This effect, however, was very weak,with a mean 6.69 ± 2.32% increase over baseline. Therefore,this marginal effect is obviously not a true DA-inducedexcitation but most likely an artifact of cationic currentapplication, which, in spite of ionic balance, may be respon-sible for a weak excitatory effect on neuronal excitability (seeBloom, 1974). More often, appearance of these weak excita-tions in fast-firing neurons is also consistent with thismechanism.

3.5. Conclusions

The results of the present study suggest that generalanesthesia is a factor affecting neuronal activity of SNrneurons and their responsiveness to natural afferent inputs.Therefore, the factor of anesthesia must be considered in theinterpretation of results and evaluation of their relevance. Thesignificant and multi-dimensional effect of anesthesia dem-onstrated in this study argues for the use of awake, unre-strained animals, whichmay provide amore natural picture ofneuronal activity and neurochemical processing. The direc-tion and degree of anesthesia-related aberrations in individualparameters cannot be easily predicted because, along withcommon features of anesthesia, various anesthetic drugs andtheir metabolites have distinct direct and indirect effects onvarious neural cells. The influence of anesthetics and generalanesthesia, moreover, may be quite different in units, inwhich activity is maintained by either excitatory afferentinputs (i.e., most cortical and striatal neurons) or by endoge-nous pacemaker mechanisms, which are modulated byexcitatory and inhibitory inputs.

114 B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

4. Experimental procedures

4.1. Animals and surgery

Data were obtained from 22 Long–Evans male rats (400 ± 50 g)supplied by Charles River Laboratories (Greensboro, NC). Allanimals were housed individually under standard laboratoryconditions (12-h light cycle beginning at 07:00) with free accessto food and water. Protocols were performed in compliancewith the Guide for the Care and Use of Laboratory Animals (NIH,Publication 865-23) and were approved by the NIDA-IRPAnimal Care and Use Committee.

The surgery procedures used have been described previ-ously (Windels and Kiyatkin, 2003). Briefly, under generalanesthesia (Equithesin 0.33 ml/100 g i.p.; dose of sodiumpentobarbital 32.5 mg/kg and chlorale hydrate 145mg/kg), ratswere implanted with a plastic, cylindrical hub, designed tomate with a microelectrode holder (Rebec et al., 1993) duringrecording. This hub was centered over a hole drilled above theSN (4.8–6.0 mm posterior and 1.4–2.6 mm lateral to thebregma). After a 3- to 4-day period of recovery and habituationto the experimental chamber, recording sessions were heldonce daily over the next 1–3 days. Recordings duringanesthesia were performed in a separate group of rats(n = 13), prepared as described above. Chloral hydrate(400mg/kg) was injected intraperitoneally, and the initialdose was supplemented by additional doses at 120 mg/kgper hour. In these experiments, body temperature wasmaintained automatically at 37.2 ± 0.2 °C with an electricheating pad and feedback rectal thermal probe.

4.2. Single-unit recording and iontophoresis

Four-barrel, microfilament-filled, glass pipettes (Omega Dot50744, Stoelting, Wood Dale, IL), pulled and broken to adiameter of 5 ± 1 μm, were used for single-unit recordingand iontophoresis. The recording barrel contained 2% Ponta-mine sky blue (BDH Chemicals Ltd, Poole, England) in 3 MNaCl, and the balance barrel contained 0.25 M NaCl. Theremaining barrels were filled with 0.25 M solutions of L-GLUmonosodium salt (in distilled water, pH 7.5), GABA (in 0.125 MNaCl, pH 4.0), and DA hydrochloride (in distilled water, pH 4.5).All drugs were obtained from Sigma (St. Louis, MO). Ourelectrode design allowed us to test within the same recordingonly two neuroactive substances. The resistance of therecording channel was 3–5 MΩ (measured at 100 Hz), and thedrug-containing barrels ranged between 10 and 35 MΩ.Retaining [(±)8–10 nA] and ejecting [(±)5–60 nA, anionic forGLU and cationic for DA and GABA] currents were appliedwitha constant current generator (Ion 100T, Dagan, Minneapolis,MN). Each multibarrel pipette was filled with fresh solutionless than 1 h before use and fixed in a microdrive assemblythat later was inserted into the skull-mounted hub. Theelectrode was then advanced 8.0 mm below the skull surfaceto the starting point of unit recording.

Neuronal discharge signals were sent to a head-mountedpreamplifier (OPA 404KP, Burr Brown, Tucson, AZ) and thenadditionally amplified and filtered (band pass: 300–3,000 Hz)with a Neurolog System (Digitimer, Hertfordshire, UK). The

filtered signal was then recorded using a Micro 1401 MK2interface (Cambridge Electronic Design, Cambridge, UK). Spikeactivity was monitored with a digital oscilloscope and audioamplifier and analyzed using a Spike2 interface (CambridgeElectronic Design). After the isolation of single-unit dischargesand identification of the spike waveform, data collection foreach neuron typically lasted 10–30 min. Our protocol typicallyincluded several 20 s applications of neurotransmittersperformed at 60 to 90 s intervals with either constant orincreasing ejection currents. During the experiment, the rats'activity was recorded using a wide-angle camera (CreativeTechnology, Milpitas, CA). All iontophoretic applications usedfor statistical analysis were performedwhen the animals wereat rest with no sign of overt movements.

4.3. Histology

After the last recording session, animals were anesthetizedand Pontamine sky blue was deposited by current injection(−20 μA for 20 min) at the last recording site. Rats wereperfused transcardially with saline solution followed by 4%paraformaldehyde solution. The brain was then removed andfrozen on dry ice. Subsequent histological location of themarked site was made on 30-μm-thick frontal sections. Insome cases, the brain was additionally processed for tyrosinehydroxylase immunohistochemistry. The Paxinos and Wat-son atlas (Paxinos et al., 1998) served as the basis forhistological analyses.

4.4. Data analysis

For each recorded unit, three parameters of impulse activity –mean rate (X), standard deviation (SD), and coefficient ofvariation (CV) –were calculated based on 20-s samples with 1-s quantification bins. These parameters were used forpopulational analysis of impulse activity under two condi-tions. Since in many brain structures (i.e., thalamus, hypo-thalamus, central gray matter, striatum) these parameters aredistributed according to the ln normal law (Kiyatkin, 1984;Kiyatkin and Rebec, 1999; Nakahama et al., 1971; Werner et al.,1963; Werner and Mountcastle, 1965), we used ln transforma-tion of these parameters to describe impulse activity of SNrneurons and we used their ln derivates for statisticalcomparison of between-group differences. Since each neuronwas tested with several applications of one or two neuro-transmitters and we calculated basal discharge rates preced-ing each iontophoretic test, these values were used tocharacterize the variability of basal impulse activity in thetwo conditions.

Each iontophoretic test was statistically evaluated, and theresponse was accepted (i.e., excitation or inhibition) if themean firing rate during iontophoresis differed significantly(P < 0.05; two-tail Student's t test) from an equivalent period ofbaseline activity immediately preceding the iontophoreticapplication. These responses were also assessed in terms ofabsolute and relative magnitude and onset and offsetlatencies. Because the duration of each neuronal recordingin freely moving rats varied from 5 to 30 min and the testingprogram for each unit was different, it was impossible toassess response thresholds and dose–response relationships

115B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

in each individual unit. Therefore, our data are reported as thenumber of both units and iontophoretic responses. Variousrelationships between impulse activity and iontophoreticresponses were assessed with Student's t tests and analysisof variance (ANOVA) that was followed by Fisher post hoc test.

Acknowledgments

We thank Drs. Roy Wise and Barry Hoffer for valuablecomments regarding the manuscript and Paul Brown foreditorial assistance. This research was supported by theIntramural Research Program of the NIH, NIDA.

R E F E R E N C E S

Adachi, Y.U., Watanabe, K., Satoh, T., Vizi, E.S., 2001. Halothanepotentiates the effect of methamphetamine and nomifensineon extracellular dopamine levels in rat striatum: amicrodialysis study. Br. J. Anesth. 86, 837–845.

Atherton, J.F., Bevan, M.D., 2005. Ionic mechanisms underlyingautonomous action potential generation in the somata anddendrites of GABAergic substantia nigra pars reticulataneurons in vitro. J. Neurosci. 25, 8272–8281.

Bloom, F.E., 1974. To spritz or not to spritz: the doubtful value ofaimless iontophoresis. Life Sci. 14, 1834–1919.

Celada, P., Paladini, C.A., Tepper, J.M., 1999. GABAergic control ofrat substantia nigra dopaminergic neurons: role of globuspallidus and substantia nigra pars reticulata. Neuroscience 89,813–825.

Collingridge, G.L., Davies, J., 1981. The influence of striatalstimulation and putative neurotransmitters on identifiedneurons in the rat substantia nigra. Brain Res. 212, 345–359.

Deniau, J.M., Kitai, S.T., Donoghue, J.P., Grofova, I., 1982. Neuronalinteractions in the substantia nigra pars reticulata throughaxon collaterals of the projection neurons. Anelectrophysiological and morphological study. Exp. Brain Res.47, 105–113.

Evans, E.F., Nelson, P.G., 1973. The responses of single neurons inthe cochlear nucleus of the cat as a function of their locationand anesthetic state. Exp. Brain Res. 17, 402–427.

Fallon, J.H., Loughlin, S.E., 1995. Substantia nigra. In: Paxinos, G.(Ed.), The Rat Nervous System. Academic Press, San-Diego,pp. 215–237.

Gauchy, C., Kemel, M.L., Desban, M., Romo, R., Glowinski, J.,Besson, M.J., 1987. The role of dopamine released from distaland proximal dendrites of nigrostriatal dopaminergic neuronsin the control of GABA transmission in the thalamic nucleusventralis medialis in the cat. Neuroscience 22, 935–946.

Hajos, M., Greenfield, S.A., 1994. Synaptic connections betweenpars compacta and pars reticulata neurones:electrophysiological evidence for functional modules withinthe substantia nigra. Brain Res. 660, 216–224.

Hammond, C., Deniau, J.M., Rizk, A., Feger, J., 1978.Electrophysiological demonstration of an excitatorysubthalamonigral pathway in the rat. Brain Res. 151,235–244.

Kiyatkin, E.A., 1984. Neuronal activity in rats medial thalamusunder conditions of urethane anesthesia, wakefulness, andrestraint stress. J. Highest Nerv. Act. 34, 560–752.

Kiyatkin, E.A., Brown, P.L., 2005. Brain and body temperaturehomeostasis during sodium pentobarbital anesthesiawith and without body warming. Physiol. Behav. 84,563–570.

Kiyatkin, E.A., Rebec, G.V., 1998. Heterogeneity of ventraltegmental area neurons: single-unit recording andiontophoresis in awake, unrestrained rats. Neuroscience 85,1285–1309.

Kiyatkin, E.A., Rebec, G.V., 1999. Modulation of striatal neuronalactivity by glutamate and GABA: iontophoresis in awake,unrestrained rats. Brain Res. 822, 88–106.

Kuwada, S., Batra, R., Stanford, T.R., 1989. Monaural and binauralresponse properties of neurons in the inferior colliculus of therabbit: effects of sodium pentobarbital. J. Neurophysiol. 61,269–282.

Lovinger, D.M., Zimmerman, S.A., Levitin, M., Jones, M.V.,Harrison, N.L., 1993. Trichloroethanol potentiates synaptictransmission mediated by gamma-aminobutyric acid-Areceptors in hippocampal neurons. J. Pharmacol. Exp. Ther.264, 1097–1103.

Michenfelder, J., 1988. Anesthesia and the Brain: Clinical,Functional, and Vascular Correlates. Churchill Levingstone,New York.

Nakahama, H., Ishii, N., Yamamoto, M., Saito, H., 1971. Stochasticproperties of spontaneous impulse activity in central singleneurons. Tohoku J. Exp. Med. 104, 373–391.

Nieoullon, A., Cheramy, A., Glowinski, J., 1977. Release ofdopamine in vitro from cat substantia nigra. Nature 266,375–377.

Patel, I.M., Chapin, J.K., 1990. Ketamine effects on somatosensorycortical single neurons and on behavior in rats. Anesth. Analg.70, 635–644.

Paxinos, G., Watson, C., 1998. The Rat Brain in StereotaxicCoordinates, Fourth Edition. Academic Press, San Diego.

Rebec, G.V., Langley, P.E., Pierce, R.C., Wang, Z., Heidenreich, B.A.,1993. A simple micromanipulator for multiple uses in freelymoving rats: electrophysiology, voltammetry, andsimultaneous intracerebral infusions. J. Neurosci. Methods 47,53–59.

Richards, C.D., Shiroyama, T., Kitai, S.T., 1997. Electrophysiologicaland immunocytochemical characterization of GABA anddopamine neurons in the substantia nigra of the rat.Neuroscience 80, 545–557.

Rick, C.E., Lacey, M.G., 1994. Rat substantia nigra pars reticulateneurons are tonically inhibited via GABA-A, but not GABA-B,receptors in vitro. Brain Res. 659, 133–137.

Robledo, P., Feger, J., 1990. Excitatory influence of rat subthalamicnucleus to substantia nigra pars reticulata and the pallidalcomplex: electrophysiological data. Brain Res. 518, 47–54.

Ruffieux, A., Schultz, W., 1981. Influence of dopamine on parsreticulata neurons of substantia nigra. J. Physiol. (Paris) 77,63–69.

Schmitt, P., Souliere, F., Dugast, C., Chouvet, G., 1999. Regulation ofsubstantia nigra pars reticulata neuronal activity by excitatoryamino acids. Naunyn-Schmiedeberg's Arch. Pharmacol. 360,402–412.

Smith, Y., Bolam, J.P., 1989. Neurons of the substantia nigrareticulata receive a dense GABA-containing input from theglobus pallidus in the rat. Brain Res. 493, 160–167.

Syka, J., Suta, D., Popelar, J., 2005. Responses to species-specificvocalizations in the auditory cortex of awake and anesthetizedguinea pigs. Hear. Res. 206, 177–184.

Waszczak, B.L., Walters, J.R., 1983. Dopamine modulation of theeffects of gamma-aminobutyric acid on substantia nigra parsreticulata neurons. Science 220, 218–221.

Waszczak, B.L., Walters, J.R., 1986. Endogenous dopamine canmodulate inhibition of substantia nigra pars reticulata neuronselicited by GABA iontophoresis or striatal stimulation.J. Neurosci. 6, 120–126.

Werner, G., Mountcastle, V.B., 1965. Neural activity inmechanoreceptive cutaneous afferents: stimulus–responserelations, Weber functions, and information transmission.J. Neurophysiol. 28, 359–397.

116 B R A I N R E S E A R C H 1 0 8 6 ( 2 0 0 6 ) 1 0 4 – 1 1 6

Werner, G., Mountcastle, V.B., 1963. The variability of centralneural activity in a sensory system, and its implications for thecentral reflection of sensory events. J. Neurophysiol. 26,958–977.

Windels, F., Kiyatkin, E.A., 2003. Modulatory action ofacetylcholine on striatal neurons: microiontophoretic study inawake, unrestrained rats. Eur. J. Neurosci. 17, 613–622.

Xie, T., McGann, U.D., Kim, S., Yuan, J., Ricaurte, G.A., 2000. Effectof temperature on dopamine transporter function andintracellular accumulation ofmethamphetamine: implications

for methamphetamine-induced dopaminergic neurotoxicity.J. Neurosci. 20, 7838–7845.

Yung, K.K., Bolam, J.P., Smith, A.D., Hersch, S.M., Ciliax, B.J., Levey,A.I., 1995. Immunocytochemical localization of D1 and D2dopamine receptors in the basal ganglia of the rat: light andelectron microscopy. Neuroscience 65, 709–730.

Zurita, P., Villa, A.E.P., De Ribaupierre, Y., De Ribaupierre, F.,Rouiller, E.M., 1994. Changes in single unit activity in the cat'sauditory thalamus and cortex associated to differentanesthetic conditions. Neurosci. Res. 19, 303–316.