Noxious Stimulation Excites Serotonergic Neurons

PAINÒ 154 (2013) 647–659

www.elsevier.com/locate/pain

Research papers Noxious stimulation excites serotonergic neurons: A comparison between the lateral paragigantocellular reticular and the raphe magnus nuclei ⇑ Rémi Gau a,b,c, Caroline Sévoz-Couche a, Michel Hamon a,b,c, Jean-François Bernard a,b,c, a Université Pierre et Marie Curie, Paris 6, Site Pitié-Salpêtrière, Paris, France b Université Paris Descartes, Paris, France c Centre de Psychiatrie et Neurosciences, INSERM UMR 894, Paris, France

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article. article info abstract

Article history: The present study was designed to record electrophysiological responses to graded noxious thermal stim- Received 13 January 2012 uli of serotonergic and nonserotonergic neurons in the lateral paragigantocellular reticular (LPGi) and the Received in revised form 28 August 2012 raphe magnus (RMg) nuclei in rats. All of the neurons recorded were juxtacellularly filled with neurobio- Accepted 21 September 2012 tin and identified with double immunofluorescent labeling for both neurobiotin and serotonin. Under halothane anesthesia (0.75%), noxious thermal stimuli P48°C activated almost all (88%) of the serotonergic neurons located within the LPGi (n = 16). The increase in firing was clear (3.4 ± 0.3 spike/s: mean of Keywords: responses above the population median) and sustained during the whole application of strong thermal Baroreflex noxious stimuli, with a high mean threshold (48.3 ± 0.3°C) and large receptive fields. Recording of sero- Electrophysiology Noxious stimuli tonergic neurons in the RMg (n = 21) demonstrated that the proportion of strongly activated (>2 spike/s) Pain modulation neurons (19% vs 59% for the LPGi) as well as the magnitude of the activation (2.1 ± 0.4 spike/s: mean of Paragigantocellular nucleus responses above the population median) to thermal noxious stimuli were significantly lower than in the Raphe magnus LPGi (P < .05). Within the boundaries of both the LPGi and the RMg (B3 group), nonserotonergic neurons Serotonin were also predominantly excited (75%) by noxious stimuli, and the resulting activation (7.9 ± 1.2 spike/s) was even greater than that of serotonergic neurons. Thermal noxious stimuli–induced activation of LPGi serotonergic cells probably plays a key role in serotonin-mediated modulations of cardiac baroreflex and transmission of nociceptive messages occurring under such intense noxious conditions. Ó 2012 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction stimuli (P48°C). Importantly, the latter stimuli increased c-Fos expression in serotonergic neurons, markedly within the LPGi, Serotonergic neurons located within the lateral paragigantocel- and moderately within the RMg, which shed new light on the lular reticular nucleus (LPGi) and the raphe magnus nucleus (RMg) physiological role of the serotonergic B3 group. Indeed, these data are well known to be involved in descending control of the trans- support the idea that most serotonergic nociceptive neurons are mission of nociceptive messages [7,11,28,31]. In particular, strong grouped within the LPGi, whereas previous electrophysiological antinociceptive effects can be induced by electrical stimulation of studies, although conflicting, showed that the serotonergic neu- the LPGi [57] or the RMg [23,65,73]. In line with these behavioral rons which respond (slightly) to noxious stimuli are located within observations, spinal neuronal response to noxious stimuli is de- the RMg [3,19,35,64,78,95,96]. creased by direct stimulation of the LPGi [42] or the RMg However, our inference about the functional implication of the [29,59,80] through serotonin receptor activation at the spinal level LPGi was based on a c-Fos technique that provides only an indirect [6,44,68]. Importantly, LPGi and RMg serotonergic neurons to- view of neuronal response [38]. Thus, it was essential to support gether constitute the B3 group [83], the major source of serotonin and extend these data with single-unit electrophysiological record- release within spinal superficial laminae [50–52,72,82]. ings. Indeed, to our knowledge, neuronal recordings have been so Gau et al. [38] demonstrated that serotonergic neurons of the far centered on the RMg, and no electrophysiological studies have LPGi (located laterally to RMg wings) are also responsible for the yet been dedicated to the responses of LPGi serotonergic neurons inhibition of cardiac baroreflex induced by strong thermal noxious to noxious stimuli. The first aim of the present study was therefore to evaluate the ⇑ Corresponding author. Address: INSERM/UPMC, UMR 894, Site Pitiè-Salpêtrière, intensity and the nature of the electrophysiological responses of 91 Boulevard de l’Hôpital, F-75013 Paris, France. LPGi serotonergic neurons to nonnoxious and noxious thermal E-mail address: [email protected] (J.-F. Bernard).

0304-3959/$36.00 Ó 2012 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.pain.2012.09.012 648 R. Gau et al. / PAINÒ 154 (2013) 647–659 stimuli of different intensities. In order to make a comparison, RMg 1.5% (w/v) neurobiotin (Vector Laboratories, Burlingame, CA, serotonergic neurons were also recorded under the same condi- USA) and 2% (w/v) NaCl. The micropipettes were inserted via the tions. Within the boundaries of B3 serotonergic region, we have cisterna magna into the dorsal surface of the brainstem at a level also recorded and identified nonserotonergic neurons whose close to the medullocerebellar angle. The micropipettes were posi- importance in descending controls of nociception and their tioned in a parasagittal plane at 45° angle with respect to horizon- involvement in the analgesic action of morphine were early recog- tal plane. Coordinates of the rostral tip were: 0.2 mm caudally to nized [30–32,56,88]. We took advantage of the electrophysiologi- the obex, 0.2–0.4 or 0.8–1.2 mm laterally to the midline to record cal juxtacellular technique to fill each recorded neuron with neurons in the RMg or the LPGi, respectively. The recording depth neurobiotin, to determine its precise location and evaluate was 4.2–5.2 mm below the surface of the brainstem, along the whether or not it had a serotonergic phenotype (double immuno- micropipette track. fluorescent labeling: neurobiotin + serotonin). After initial recording amplification (Axoclamp 2B; Axon Instru- ments, Union City, CA, USA), complementary amplification with 2. Materials and methods minimal filtration and removal of 50 Hz component without filtration (Hum Bug electronic device; Quest Scientific, Vancouver, 2.1. Animal preparation Canada), the signal was observed on an oscilloscope to allow on-line monitoring and digitized at 24 kHz frequency with a data Electrophysiological experiments were performed on 84 acquisition system (CED 1401 with Spike 2 software; Cambridge Sprague–Dawley male rats weighing 250–300 g. Animals were Electronic Design, Cambridge, UK). The action potentials were also kept under controlled environmental conditions (ambient temper- fed directly into a window discriminator and monitored with a sec- ature 21 ± 1°C, 60% relative humidity, unrestricted access to food ond oscilloscope. Storage of the full recording signal in a computer and water, alternate 12 h light/12 h dark cycles) for at least 1 week allowed further controls and analysis (online as well as off-line). after receipt from the breeding center (CER Janvier, Le Genest-St Each time a spontaneously active unit was found, its baseline Isle, France). Procedures involving animals and their care were activity was recorded for a minimal time around 5 min in the ab- conducted in conformity with the institutional guidelines in com- sence of stimulation and at a steady-state concentration of halo- pliance with the Council Directive No. 87-848 of the Ministère de thane. This sampling method ignored silent neurons but had the l’Agriculture et de la Forêt, Service Vétérinaire de la Santé et de advantage of minimizing the disruption caused by repeated nox- la Protection Animale (permissions No. 75-148 to JFB, No. 75-116 ious stimuli. to MH, and No. 75-855 to CSC). Before anesthesia induction, rats received a dose of atropine 2.3. Innocuous and noxious stimulation (0.75 mg/kg, i.p.) to decrease respiratory secretion. Anesthesia was induced with 2% halothane in a 67% N2O–33% O2 mixture. A Thermal and mechanical noxious stimuli usually lasted 20 s. A tracheal cannula was inserted by endotracheal intubation [94]. pause of at least 3 min between successive noxious stimuli allowed The animals were mechanically ventilated at a rate of 48 breaths/ the return of the unit activity to its background level during inter- min with a Palmer pump for the whole duration of the experiment. stimuli intervals. First, innocuous mechanical stimuli (touch,

The expiratory CO2 was monitored continuously with a Capnomac brushing, rubbing, light pressure) were applied to a paw and the II (Datex Instruments, Helsinki, Finland). The end-tidal CO2 was tail. Then thermal noxious stimulation consisted of immersing con- maintained at 4%. The end-tidal levels of halothane, O2, and N2O secutively each paw and occasionally the tail of the animal in a were also checked systematically during the whole experimental 50°C waterbath during 20 s. This method allowed us to estimate period. The core temperature was maintained at 37.0 ± 0.5°Cby the size of the receptive field of the recorded unit. means of a homeothermic blanket system. Finally, additional tests were made in the portion of the receptive The animals were then paralyzed with an infusion of gallamine field that gave the most intense response. The adequate reproduc- triethiodide (48 mg/kg/h; Sigma-Aldrich, St. Quentin-Fallavier, ibility of this response was controlled; then graded temperatures France) via an intraperitoneal catheter and mounted in a stereotaxic of 44, 46, 48, 50, and 52°C were systematically applied to determine frame with the incisors bar placed 18 mm below the interaural hor- the threshold temperature and the encoding properties of the unit. izontal plane (ie, 14.7 mm below the horizontal plane defined in the Importantly, a pause of 5 min was respected after 50 and 52°C stereotaxic atlas of Paxinos and Watson [75]). This represents an noxious stimuli. Furthermore, the 52°C temperature was used as approximate 30° head tilt from the reference position in the atlas sparingly as possible to avoid sensitization/desensitization phe- of Paxinos and Watson [75]. The dorsal surface of the brainstem nomena. For most neurons, a mechanical noxious stimulation was then exposed by a small incision of the atlanto-occipital liga- (pinch) was applied once with calibrated forceps (16–32 N/cm2), ment and the dura matter underneath to allow electrode descent. at least to the most sensitive portion of the receptive field. After surgery, the halothane level was reduced to 0.7–0.8% in a mixture of 50% N2O–50% O2. A 15–30-min stable anesthetic condi- 2.4. Juxtacellular filling tion was allowed before starting recording. This level of anesthesia, which did not excessively depress neuronal responses to noxious Once examination of the responses of an adequately isolated stimuli (see discussion in Bester et al. [12]), met the IASP ethical neuron was completed (as described above), neurobiotin was in- recommendations for animal experiments [97]: it was sufficient jected by juxtacellular iontophoresis with continuous control of to reach the minimum alveolar concentration of inhaled anesthetic neuronal firing to ensure that the same unit was recorded during agent required to prevent a direct response to a noxious stimulus the whole procedure [76]. Using the bridge circuit of the recording in 50% of animals [20]. In previous experiments, it was also suffi- amplifier (Axoclamp 2B), the tracer was injected with 200 ms ON– cient to prevent electrocortical signs of arousal during strong nox- 200 ms OFF pulses of DC current at gradually increasing intensity ious stimulation [43]. (1–12 nA, anode in the pipette). After a delay of a few seconds to several minutes, the electrical background noise increased before 2.2. Recordings the occurrence of a spike firing driven by current pulses, which indicated that the microelectrode tip was in juxtacellular position. Single extracellular recordings were made with glass micropi- A clear-cut increase in firing during the 200 ms ON period of the pette electrodes (12–20 MO) filled with an aqueous solution of current pulse indicated that the neuron had been filled efficiently R. Gau et al. / PAINÒ 154 (2013) 647–659 649 by neurobiotin [76]. Once the critical change in firing pattern had then converted into a permanent and intense black labeling as fol- occurred, the intensity of pulses was then rapidly adjusted (it lows: Sections were incubated for 1 h in PBS with only the biotin– was usually reduced to 2–3 nA) to maintain a sufficient level of fir- peroxidase component (1 drop per 10 mL of solution) of the ABC ing of the neuron and prevent cellular damage. Because the ampli- Elite kit (Vector Laboratories). The sections were then rinsed with tude and the discharge frequency of spikes often varied during the PBS for 20 min and incubated in a Tris buffer solution (0.12 M, pH injection period, it was sometimes necessary either to adjust the 7.4) containing 0.05% (w/v) of 3,3-diaminobenzidine (DAB; Sigma- current intensity or to move the microelectrode by a few microm- Aldrich) and 0.2% (w/v) of ammonium nickel sulfate (Sigma- eters up or down. In this study, the juxtacellular current was deliv- Aldrich) for 2 min. Finally, increasing doses of H2O2 were added ered over an average time of 4 min (range: 20 s to 30 min). In most every 5 min to sections floating in the DAB–nickel solution to cases, only one neuron was recorded and juxtacellularly injected obtain H2O2 concentration reaching successively 0.001%, 0.003%, per rat. 0.007%, 0.015%, 0.040% and 0.085% (v/v) of the H2O2 parent Importantly, in our hands, only the neurons whose firing fol- solution (30% v/v). Serial sections were finally mounted on lowed clearly the current injected through the micropipette were gelatin-covered slides and coverslipped. labeled by neurobiotin. Furthermore, only one neuron was labeled Digitized photomicrographs were made with a tri-CCD color in the recording site. digital camera (KY-F50, 3 768 576 pixels; JVC Professional Eur- ope, Paris, France) for bright-field microscopy. For epifluorescence, 2.5. Immunohistochemistry and visualization of neurobiotin-filled we used a mono-CCD high-sensitivity grayscale digital camera neurons (ORCA-285/C4742–95, 1 1344 1024 pixels; Hamamatsu Pho- tonics, Massy, France) connected to a microscope (BX21; Olympus Animals were kept under anesthesia and ventilation for a usual France, Rungis, France), which sent signal outputs to a Macintosh duration of 4–6 h after completion of the juxtacellular injection. computer. Using Openlab software (Improvision, Conventry, UK), Then animals were overdosed with 5% halothane and perfused images at different focal planes were captured, digitized with a transcardially. The perfusion was made with a warm (37°C) 24-bit color scale, or artificially colored for making epifluorescence heparinized saline solution for 2 min, followed by a 0.12 M images. An operator allowed the combination, pixel by pixel, of phosphate-buffered (pH 7.4) saline (PBS) solution containing 4% images in different focal planes. These operations resulted in the paraformaldehyde, 0.1% glutaraldehyde, and 0.05% picric acid for production of one image by incorporating the brightest (epifluores- 20 min. It was ended by a 20% (w/v) sucrose solution in PBS for cence) or the darkest (bright field) value of each corresponding pix- 10 min. Finally the brain was removed. el (ie, with the same x, y coordinates) in each focal plane (ie, with The next day, the brain was cut in coronal serial sections 80 lm variable z coordinate throughout the thickness of the section). thick on a freezing microtome and collected in PBS. The sections Low-magnification images were used to determine the location were then incubated overnight at room temperature in PBS con- of each recorded labeled neuron with respect to anatomical land- taining 0.4% Triton X-100, 1% normal goat serum, and the primary marks (section ventral border and midline; B3 serotonergic group; rabbit anti-serotonin antibody (1:40,000, Ref. PC228L, batch facial, superior and inferior olivary nuclei; pyramidal and medial D21789-1; Calbiochem, La Jolla, CA, USA). This polyclonal antibody lemniscus tracts). Images were exported to Adobe Photoshop in or- was raised against serotonin conjugated to bovine serum albumin der to adjust brightness and contrast. Finally, the images were im- by glutaraldehyde; the distribution of serotonin cells labeled with ported into FreeHand to insert labels and anatomical landmarks. this antibody matched completely that described by Steinbusch Labeled neurons were finally mapped together onto coronal sec- [83]. The sections were then rinsed for 30 min with PBS and incu- tions of brain atlas [75] at the appropriate level. bated for 2 h in PBS containing 0.4% Triton X-100, 1% normal goat Because every neuron recorded here was immunohistochemi- serum, the secondary Alexa fluor 594 goat anti-rabbit antibody cally tested for serotonin, very little uncertainty remains about (1:150; Invitrogen, Cergy Pontoise, France) to label serotonin, and the location and the serotonergic or nonserotonergic phenotype. streptavidin conjugated to Alexa fluor 488 (1:150; Invitrogen) to The few doubtful cases were easily cleared up through confocal label neurobiotin. After a second rinse for 30 min with PBS, observation. sections were transitorily mounted on Superfrost slides and coverslipped with only PBS. 2.6. Electrophysiological analysis and identification of neurons The sections containing a neurobiotin filled neuron were identified under epifluorescent illumination. Low- and medium-magnifi- Single-unit activity was analyzed by Spike 2 software. For each cation (4to20 objectives) digitized photomicrographs of the neuron, spike waveforms were extracted during spontaneous firing neurons and the sections of interest were made immediately. The and averaged to obtain a mean spike waveforms. Then we mea- latter sections were then removed from the transitory mounting sured the mean spike duration from the onset of the first (negative) in PBS and mounted temporarily (4–6 h) with fluoromount for con- potential to the offset of the second (positive) potential of the focal imaging. The remaining sections were put back in PBS. Confo- mean spike. cal fluorescent images of the filled neuron were generated with a To recognize serotonergic neurons, Mason [64] developed an Leica TCS-400 laser scanning confocal microscope (40 to 60 electrophysiological algorithm that is based on the low frequency oil-immersion objective). Stacks of serial images were taken for and the regularity of their spontaneous activity. This method uses each fluorophore throughout the entire Z-axis of the labeled neu- the interspike interval (ISI, in ms), the mean ISI (ISI), the standard ron. This single-pass format was utilized to allow each tracer exci- deviation of ISI (SDISI), and the coefficient of variation of ISI tation by specific lasers. Emitted spectra were collected separately (CVISI ¼ SDISI=ISI), which were all measured in our study during to minimize overlap. Alexa fluor 488 and 594 were excited with the first recording period of spontaneous firing (5-min duration).

488 nm (argon) and 543 nm (HeNe) lasers, respectively. Pinhole Then the value of discriminant function Y ðISI; CVISIÞ¼ setting and voxel depth were 1 Airy unit and 1–2 lm, respectively, ð½146=ISIþ0:98CVISI 1Þ was calculated for each neuron. Finally, for all images. Z-average projection images were made from stacks each neuron was classified according to its coordinates, ISI (abscis- of images by Image J software (National Institutes of Health, sa) and CVISI (ordinate), with regard to a straight line representing Bethesda, MD, USA). the discriminant function Y (ISI; CVISIÞ¼0, in a logarithmic scales After confocal imaging, the sections of interest were dis- diagram. Following the criteria defined by Mason [64], a neuron mounted and put back in PBS. Neurobiotin fluorescent signal was had serotonergic electrophysiological features when Y 650 R. Gau et al. / PAINÒ 154 (2013) 647–659

(ISI; CVISIÞ < 0—that is, the point (ISI; CVISI) representing this neu- cal test used is indicated only when it is different from the t test). ron was below or on the right of the curve Y ðISI; CVISIÞ¼0. When Signiﬁcance was taken as P < .05.

Y ðISI; CVISIÞ > 0—that is, the point (ISI; CVISI) was above or on the left of the curve—the corresponding neuron was categorized as 3. Results having nonserotonergic electrophysiological features. To determine whether 20-s noxious heat-evoked changes in In this study, we recorded, characterized, and juxtacellularly in- discharge were significantly different from spontaneous changes, jected 77 neurons found to be located strictly within the bound- the spontaneous activity period was randomly divided into 20-s aries of the B3 group (Fig. 1) visualized directly by the set of bins. Then we quantified the variability of spontaneous discharge serotonin-labeled cells (Fig. 2D1). These 77 identified neurons as the standard deviation of the differences between 2 consecutive comprised 37 serotonergic (16 in LPGi and 21 in RMg) and 40 non- 20-s bins, as recommended by Leung and Mason [56] and Gao and serotonergic (21 in LPGi and 19 in RMg) neurons (Fig. 1). Our anal- Mason [35]. The magnitude of unit response to a given stimulation ysis covered most of the rostrocaudal extent of B3 group (Bregma was calculated as the difference in unit firing rates 20 s before and 10 to 11.8 mm, Fig. 1), except its caudal and rostral extremities. after the stimulus onset. Evoked increase or decrease in discharge, Importantly, in the present study, the term LPGi referred to the which were superior to 2 standard deviations of background dis- serotonergic portion of the LPGi, which was markedly smaller than charge (20-s bins), were considered as excitatory or inhibitory re- the LPGi nucleus defined in the atlas of Paxinos and Watson [75].In sponses, respectively. The threshold temperature was defined as this electrophysiological study, we demonstrated for the first time the first temperature of those gradually applied that evoked a sig- that LPGi serotonergic cells had peculiar nociceptive properties, nificant excitatory or inhibitory response. which were significantly different from those of RMg serotonergic To illustrate LPGi and RMg neuronal responses, we calculated cells. for each of these populations 10 interdecile intervals named here ‘‘deciles’’. Each ‘‘decile’’ corresponded to 10% of the population re- 3.1. LPGi serotonergic neurons sponse classified from the lower to the higher magnitude. In the LPGi population (n = 16 responses, from r1 to r16), one ‘‘decile’’ From a slow (1.5 ± 0.2 spike/s, range: 0.4–2.9 spike/s, n = 16) and

(d) represented 1.6 responses. The calculation was made as fol- regular spontaneous discharge (CVISI = 0.46 ± 0.04), almost all LPGi lows: d1 = [(r1 1) + (r2 0.6)]/1.6; d2 = [(r2 0.4) + (r3 1) + serotonergic neurons (88%, 14 of the 16 recorded neurons with this (r4 0.2)]/1.6; d3 = [(r4 0.8) + (r5 0.8)]/1.6; d4 = [(r5 0.2) + phenotype) responded markedly to strong noxious thermal stimu- (r6 1) + (r7 0.4)]/1.6; d10 = [(r15 0.6 + r16 1)/1.6]. In the lations (P50°C) applied to a part of their receptor field, which was RMg population (n = 21 responses, from r1 to r21), one ‘‘decile’’ generally large (all parts of the body tested, ie, the 4 paws, often (d) represented 2.1 responses. The calculation was made as with a predominant paw that evoked stronger response). The usual follows: d1 = [(r1 1) + (r2 1) + (r3 0.1)]/2.1; d2 = [(r3 0.9) + response consisted of a tonic increase in the discharge during the (r4 1) + (r5 0.2)]/2.1; d3 = [(r5 0.8) + (r6 1) + (r7 0.3)]/2.1; whole stimulation period with a mean increase of 3.4 ± 0.3 spike/s d10 = [(r19 0.1) + (r20 1) + (r21 1)]/2.1. Note that the sum of above the background discharge (mean from the 8 neurons above coefficients attributed to successive rn1,rn,rn+1, for the calculation the population median). Fig. 2A illustrates the typical increase of of one ‘‘decile’’, was equal to 1.6 and 2.1 for the LPGi and the RMg, firing of a well-isolated single unit during 20 s dipping of the con- respectively. The sum of coefficients attributed to one given re- tralateral hindpaw (Fig. 2B) in a 52°C noxious waterbath, followed sponse (rn) was equal to 1. For ‘‘decile’’ calculation, the excitatory by a short-lived postdischarge. These neurons responded also sim- and inhibitory responses had positive and negative value, respec- ilarly to noxious mechanical stimulation (pinch), but none of them tively. The representation in ‘‘decile’’ allowed an adequate compar- responded to light mechanical or innocuous thermal stimulation. ison of the statistical distribution of response magnitudes in LPGi Fig. 2C illustrates the encoding of increasing temperatures by the and RMg nuclei. Furthermore, ‘‘decile’’ values were close to the neuron in Fig. 2A, the mean threshold for the whole recorded pop- individual experimental values. However, one limitation of this ulation being 48.3 ± 0.3°C(n = 14). Only one LPGi serotonergic neu- representation was the impossibility of providing worthwhile indi- ron was unresponsive (6%, 1 out of 16), and 1 was inhibited (6%, 1 vidual error bars. Thus, statistical comparisons considered the out of 16) by noxious stimulation. Fig. 2D depicts the location of whole population, or at a minimum, half of each population. the typical neuron presented above within the LPGi serotonergic To illustrate LPGi and RMg neuronal responses, we calculated group, as well as its double labeling for both neurobiotin and sero- the mean response of populations or subpopulations of serotoner- tonin, analyzed by confocal microscopy. gic and nonserotonergic neurons. For this calculation, we syn- chronized the responses of the different neurons on the 3.2. Comparison of serotonergic neurons in RMg vs LPGi nuclei stimulus onset (t0); then we added, in every 2-s interval, the number of spikes from each response before and after the t0. Fi- nally, each number was divided by the total number of responses. Fig. 3 illustrates the moderate activation from a slow spontane- It must be emphasized that, in this calculation, negative numbers ous discharge of a typical RMg unit during 52°C noxious stimula- were not used in case of inhibitory response. Indeed, in this case, tion, with poor encoding properties and its serotonergic the number of spikes per time unit decreases to a minimum of phenotype determined by epifluorescent microscopy. zero but cannot be negative. It is only later, when the mean spon- In the RMg, serotonergic neurons had a similarly slow taneous discharge was subtracted, that a negative value could be (1.3 ± 0.1 spike/s, n = 21) but significantly more regular (CVISI: obtained. 0.37 ± 0.03, P < .05) spontaneous discharge than their LPGi counter- parts. In addition, they had large receptive fields and high mean acti- 2.7. Statistical analyses vation threshold (48.5 ± 0.5°C, n = 13), similar to those of the LPGi. As illustrated in Fig. 4A, the responses to temperature P50°C Responses to noxious heat stimuli referred to the most intense were significantly distributed in a lower-magnitude scale in the response to a thermal stimulation, usually 52°C stimulation. Values RMg serotonergic population than in the same phenotype popula- are expressed as mean ± SEM. Statistics were made by t test, v2 tion in the LPGi (n1 = 21, n2 = 16, P = .017, Mann-Whitney U test). test, 2-way analysis of variance test with repeated measure (2W- The difference between the 2 populations resulted, at least in part, 2 ANOVA-R), and nonparametric Mann-Whitney U test (the statisti- from the significantly lower proportion (P < .02, v test) of strongly R. Gau et al. / PAINÒ 154 (2013) 647–659 651

was significantly lower (P < .05) than the mean response of LPGi neurons (1.0 ± 0.3 spike/s, Fig. 4B1). Fig. 4C shows that, within upper ‘‘deciles’’ 6–10, the mean response of RMg neurons (2.1 ± 0.4 spike/s, Fig. 4C2) was significantly lower (P < .05) than the mean response of LPGi neurons (3.4 ± 0.3 spike/s, Fig. 4C1). Thus, the analysis illustrated in Fig. 4A–C shows that mean responses of RMg neurons to noxious heat were lower that those of LPGi neurons. Importantly, the statistical comparison provided similar results for ‘‘deciles’’ 1–5 that mixed inhibited and activated cells as for ‘‘deciles’’ 6–10 that comprised only activated cells’ response. This analysis also pointed out the sustained character of the mean responses of RMg and LPGi serotonergic neurons (Fig. 4B1, C1, C2). The mean response of the whole population of RMg-activated neurons (1.8 ± 0.4 spike/s; n = 13) was not significantly different from that of LPGi neurons (2.6 ± 0.3 spike/s; n = 14) (P = .081, Mann-Whitney test). Importantly, this comparison does not take into account the difference in the proportion of RMg-vs LPGi-activated neurons (61% and 88% of the populations, respectively). Fig. 4D shows that both RMg and LPGi serotonergic neurons exhib-

ited signiﬁcant nociceptive encoding properties (n1 = 13 and n2 =9, respectively, F3,84 = 35.75, P < .0001, 2W-ANOVA-R test for 46, 48, 50, and 52°C). In the population of serotonergic neurons inhibited by noxious stimuli, the discharge decrease was similarly moderate for RMg (50 ± 8%, n = 6) and LPGi (57%, n = 1) neurons.

3.3. Comparison of serotonergic vs nonserotonergic neurons

Individualized comparison within each nucleus between serotonergic and nonserotonergic neurons demonstrated basically similar differences. Thus, for the sake of simplicity and clarity, we pooled on the one hand LPGi and RMg serotonergic neurons (n = 37), and on the other hand LPGi and RMg nonserotonergic neurons (n = 40). Whether the recorded neurons had a serotonergic or a nonserotonergic phenotype, their large receptive field as well as the proportions of neurons activated (73% vs 75%, respectively), inhibited (18% vs 25%) or unresponsive (8% vs 0%) to noxious ther- Fig. 1. Location of neurons recorded and juxtacellularly filled in the B3 region mal stimuli were similar. However, serotonergic neurons had a sig- including the LPGi and the RMg. (A1–A3) Schematic drawing of coronal sections nificantly (P < .001) lower and more regular spontaneous discharge through the B3 region from rostral to caudal. Each level includes 600- m l (1.4 ± 0.1 spike/s; CVISI: 0.41 ± 0.02, n = 37) than the nonserotoner- rostrocaudal extent. Star, serotonergic (5HT) neuron strongly activated P2 spike/s gic neurons (5.5 ± 1.2 spike/s; CVISI: 0.78 ± 0.08, n = 40). (red), moderately activated <2 spike/s (black), unresponsive (white filling) or inhibited (grey filling + dash) by noxious stimuli. Circle, nonserotonergic (non-5HT) Furthermore, in comparison with the response of individual neuron activated (black) and inhibited (gray filling + dash) by noxious stimuli. Gray serotonergic neurons shown in Figs. 2 and 3, Fig. 5 illustrates the area, region containing the great majority of serotonergic neurons. 7, nucleus of the larger tonic increase of the firing of a typical nonserotonergic unit seventh nerve; B3, third group of serotonergic neurons; b, mean level of the during 50°C noxious stimulation. The latter response had a sharp reconstructed section in relation to the bregma [75]; GiA, gigantocellular reticular onset and increased so as to encode noxious thermal stimuli in a nucleus, alpha part; Gi, gigantocellular reticular nucleus; LPGi, lateral paragigantocellular reticular nucleus; ml, medial lemniscus; py, pyramidal tract; RMg, raphe wide range of noxious temperatures. The absence of serotonin magnus nucleus; RPa, raphe pallidus nucleus; SO, superior olivary nucleus. Units 1, labeling superimposed with neurobiotin labeling demonstrated 2, and 3: neurons illustrated in Figs. 2, 3 and 5, respectively. Scale bar, 1 mm. the nonserotonergic phenotype of the recorded unit (Fig. 5D). Fi- nally, Fig. 6A illustrates clearly the fact that the mean increase of excited units (response >2 spike/s, red stars in Fig. 1) in the RMg firing of all serotonergic neurons activated by 52°C noxious stimu- (19%, 4 of 21 recorded neurons) than in the LPGi (56%, 9 of 16) lation (2.2 ± 0.2 spike/s over basal firing rate, n = 27) was signifi- (Fig. 4A). However, no significant differences were observed be- cantly lower (P < .001) than that of nonserotonergic neurons tween RMg (n = 21) and LPGi (n = 16) in the proportions of excited (7.9 ± 1.2 spike/s, n = 30). (61% vs 88%, respectively; P = .066, v2 test) and inhibited (29% vs Thorough comparison of serotonergic vs nonserotonergic neu- 6%; P = .087) neurons. The proportions of unresponsive neurons rons illustrated in Fig. 6 shows that the mean response of the first were similarly low in the RMg and the LPGi (10% and 6%, population was markedly lower but more sustained than that of respectively). the second one. Importantly, none of the neurons in either popula- For further comparison between RMg and LPGi serotonergic tions responded to stimulus <44°C. Clear-cut activation occurred at neurons, we averaged responses of ‘‘deciles’’ 1–5 (ie, half of the pop- higher temperatures (P48°C), and the activation of serotonergic ulation below the median) and responses of ‘‘deciles’’ 6–10 (ie, half neurons was significantly lower than that of the nonserotonergic of the population above the median) for the RMg (Fig. 4B2, C2; neurons at both 50 and 52°C(n1 = 22, n2 = 17, P < .01) (Fig. 6B). n1 = 10.5, n2 = 10.5) vs those for the LPGi (Fig. 4B1, C1; n1 =8, On the other hand, the mean temperature threshold was not signif- n2 = 8). Fig. 4B shows that, within lower ‘‘deciles’’ 1–5, the almost icantly different for serotonergic (48.4 ± 0.3°C, n = 27) and nonse- null mean response of RMg neurons (0.1 ± 0.2 spike/s, Fig. 4B2) rotonergic (47.6 ± 0.4°C, n = 30) neurons. 652 R. Gau et al. / PAINÒ 154 (2013) 647–659

Fig. 2. Typical example of a serotonergic LPGi neuron activated by noxious stimuli and juxtacellularly filled with neurobiotin (Fig. 1, unit 1). (A1 and A2) Activation during 52°C noxious stimulation (horizontal bar, 20-s duration) shown both with spikes (A1) and 1-s bin size histogram (A2). (B) Receptive field including higher (contralateral hind paw, black) and lower (other paws, gray) sensitivity cutaneous regions. (C1–C5) Response to graded thermal stimuli ranging from innocuous to strongly noxious (44–52°C), with 48°C threshold; horizontal bar, 20-s duration. (D1) Low-magnification epifluorescent microphotograph of the recorded LPGi neuron double-labeled for neurobiotin and serotonin (arrow) within the serotonergic group of cells (red). (D2–D4) High-magnification confocal imaging of the same neuron for serotonin in red (D2), for neurobiotin in green (D3), and the superimposition of both labeling (D4, light green). Abbreviations as in Fig. 1. Scale bar, D1, 500 lm; D2–D4, 20 lm.

In the population of neurons inhibited by noxious stimuli, the (3.64 ± 0.12 ms, n = 37) than that of nonserotonergic units discharge decrease was less pronounced (P < .01) for serotonergic (2.08 ± 0.13 ms, n = 40).

(51 ± 7%, n = 7) than for nonserotonergic (77 ± 5%, n = 10) Second, the ISI=CVISI representation (Fig. 6D) shows that all the neurons. recorded serotonergic neurons were adequately located below or

on the right of the curve YðISI; CVISIÞ¼0. Thus, the neurons identi- 3.4. Electrophysiological differentiation of serotonergic vs ﬁed as serotonergic in the present study had electrophysiological nonserotonergic neurons features of their spontaneous discharge similar to those deﬁned by

Mason [64]. On the other hand, the ISI=CVISI representation shows First, as illustrated in Fig. 6C, mean spike recorded in serotoner- that the majority of nonserotonergic neurons (61%, 23 of 38 recorded gic units had a signiﬁcantly (P < .001) longer duration units) were adequately located above or on the left of the curve. R. Gau et al. / PAINÒ 154 (2013) 647–659 653

Fig. 3. Typical example of a serotonergic RMg neuron activated by noxious stimuli and juxtacellularly filled with neurobiotin (Fig. 1, unit 2). (A1 and A2) Activation during 52°C noxious stimulation (horizontal bar, 20-s duration) shown both with spikes (A1) and 1-s bin size histogram (A2). (B1–B5) Response to graded thermal stimuli (44–52°C) ranging from innocuous to strongly noxious, with 50°C threshold; horizontal bar, 20-s duration. (C) Large receptive field including the 4 paws (gray). (D1–D3) Epifluorescent microphotographs of the recorded RMg neuron double-labeled for neurobiotin in green (D1), serotonin in red (D2). Superimposition of both labeling resulted in yellow color (D3). Scale bars, 20 lm.

However, about one third of the nonserotonergic units (39%, 15 of 4. Discussion 38) were located below the curve, in the serotonergic space (Fig. 6D). Thus, in our study, only 61% of nonserotonergic neurons This study demonstrated for the ﬁrst time that the great major- were adequately classed by this algorithm, as compared to 91% of ity of the serotonergic neurons (88%) located within the LPGi re- such neurons in the study by Gao and Mason [35]. Accordingly, in sponded to noxious heat. They were clearly and tonically our hands, the discriminant curve of Mason [64] categorized ade- activated (3.4 ± 0.3 spike/s: mean of responses above the popula- quately all the neurons above the curve as nonserotonergic, whereas tion median) by strong thermal noxious stimuli from a high only 71% (37 out of 52) of the neurons below the curve had actually threshold (48.3 ± 0.3°C). the expected serotonergic phenotype. 654 R. Gau et al. / PAINÒ 154 (2013) 647–659

Fig. 4. Comparison of nociceptive responsiveness of serotonergic (5HT) neurons in lateral paragigantocellular (LPGi) and raphe magnus (RMg) nuclei. (A) Distribution of responses to 52°C noxious stimuli within 10 classes corresponding to the 10 interdecile intervals, named ‘‘deciles’’. The 5HT-LPGi (solid curve, n = 16) and 5HT-RMg (dashed gray curve, n = 21) responses are distributed from the ‘‘decile’’ 1 (lower responses <0 spike/s, ie, an inhibition) to the ‘‘decile’’ 10 (higher responses >4 spike/s, ie, a strong activation). Each point (black or gray/empty) corresponds to one ‘‘decile’’. Abscissa, ‘‘decile’’ ordinate, response frequency (spike/s). (B) Mean response in lower ‘‘deciles’’ 1–5 of 5HT-LPGi (B1, black, n = 8) and 5HT-RMg (B2, gray, n = 10.5) neurons (2-s bin size histogram + SEM). (C) Mean response in higher ‘‘deciles’’ 6–10 of 5HT-LPGi (C1, black, n = 8) and 5HT-RMg (C2, gray, n = 10.5) neurons. Responses are aligned to the onset of 20-s duration stimulus (horizontal bar). (D) Mean stimulus–response curves to graded thermal stimuli for 5HT-LPGi (solid line, n = 9) and 5HT-RMg (dashed line, n = 13) activated neurons. Abscissa, stimulus temperature; ordinate, mean response frequency.

For the purpose of comparison, serotonergic neurons were also of LPGi (56%) than RMg (29%) serotonergic neurons was strongly recorded in the RMg, which constitutes, together with the LPGi, the activated (>2 spike/s) by noxious stimuli P48°C is reminiscent of B3 group [83]. Surprisingly, although the proportion of clearly acti- our previous findings that c-Fos expression in serotonergic neurons vated neurons was lower than in the LPGi, the responsiveness of was enhanced to a much larger extent in the LPGi than the RMg by RMg neurons was higher than expected. In line with previous re- noxious stimuli [38]. ports [35,56], most of nonserotonergic neurons (75%) recorded in the B3 region were also excited by noxious stimuli. Indeed, their 4.2. Comparison with previous electrophysiological studies increase of firing over the baseline (7.9 ± 1.2 spike/s) was even greater than that of serotonergic neurons. However, our study The only suitable comparison deals with recording of RMg sero- mainly focused on serotonergic neurons, whose neurochemical tonergic neurons often identified by indirect means, but previous phenotype was unambiguously determined. relevant studies gave conflicting results. Thus, Wessendorf and Anderson [95] and Chiang and Gao [19] showed that, in ure- 4.1. Comparison with previous c-fos studies thane-anesthetized rats, a majority of RMg serotonergic neurons were activated either by pinch or by noxious heat. In contrast, To our knowledge, no electrophysiological study had systemat- more recently, Potrebic et al. [78], Mason [64], and Winkler et al. ically examined the response of LPGi serotonergic neurons to nox- [96], using halothane or isoflurane anesthetic regimens, found that ious stimuli. Interestingly, our finding that LPGi neurons were most RMg serotonergic neurons were unresponsive to noxious activated by strong noxious stimuli is in general agreement with stimuli. previous c-Fos [18,55,71,91] and 2-deoxyglucose [61,77] studies. Of such studies, that by Gao and Mason [35] is of particular In particular, the observation that a significantly higher proportion interest because these authors examined the response to noxious R. Gau et al. / PAINÒ 154 (2013) 647–659 655

Fig. 5. Typical example of a nonserotonergic LPGi neuron activated by noxious stimuli and juxtacellularly filled with neurobiotin (Fig. 1, unit 3). (A1 and A2) Activation during 50°C noxious stimulation (horizontal bar, 20-s duration) shown both with spikes (A1) and 1-s bin size histogram (A2). (B) Receptive field including higher (ipsilateral forepaw, black) and lower (other paws, gray) sensitivity cutaneous regions. (C1–C5) Response to graded thermal stimuli (44–52°C) ranging from innocuous to strongly noxious, with 46°C threshold; horizontal bar, 20-s duration. (D1–D3) Epifluorescent microphotographs of the recorded LPGi neuron single labeled for neurobiotin in green (D2, arrow), but unlabeled for serotonin (D1, arrow). Superimposition of both plans yielded only the neurobiotin green color (D3). (D4) Microphotograph of the same neuron after neurobiotin labeling with DAB–nickel procedure. Scale bars, 50 l m. heat of a large set of immunohistochemically identified serotoner- reasons could explain this difference. First, we used longer stimuli gic neurons. Indeed, when deducting the 3-s response onset, the duration, 20-s vs 6-s in the study of Gao and Mason [35]. Second, moderate mean response of all activated RMg serotonergic neu- we tested the 4 paws vs a portion of the tail in the previous study; rons observed in our study, 1.9 spike/s (initially 1.8 ± 0.4 spike/s), it might be inappropriate to generalize the response of a cell by is in good agreement with the value roughly estimated from testing only a part of its receptive field [24]. For the data herein re- Fig. 4A of Gao and Mason [35]: 1.5 spike/s (initially ported, if we had considered only the first 6 s of neuronal response 0.79 ± 0.16 spike/s). However, we observed a higher proportion of driven from a single paw, this would have decreased the propor- serotonergic neurons activated by thermal noxious stimuli (61%) tion of neurons classified as responsive from 61% to 29%. Third, than in this previous report (17%; Gao and Mason [35]). Several the portions of stimulated skin differed in the 2 studies: we 656 R. Gau et al. / PAINÒ 154 (2013) 647–659

Fig. 6. Comparison of B3 serotonergic (5HT) vs nonserotonergic (non-5HT) neurons activated by noxious thermal stimulation. (A) Mean response of 5HT (A1, black, n = 27) and non-5HT (A2, gray, n = 30) neurons activated by noxious stimulation (P50°C), shown with histograms aligned to the 20-s duration stimulus onset (horizontal bar). (B) Mean stimulus–response curves for activated 5HT (n = 22) and non-5HT (n = 17) neurons. Abscissa, stimulus temperature. Ordinate, mean response frequency. (C1) Respective averaged spike waveforms of 5HT and non-5HT units. Arrows, measurement points. (C2) Histograms of spike duration for 5HT cells in lateral paragigantocellular (LPGi) and raphe magnus (RMg) nuclei (black bars, n = 16 and n = 21; respectively) and non-5HT cells in LPGi and RMg nuclei (empty bars, n = 21 and n = 19, respectively). D, Characteristics of spontaneous discharge of 5HT units and non-5HT units in LPGi and RMg nuclei; abscissa, mean interval interspike (ISI); ordinate, coefﬁcient of variation of

ISI (CVISI). The solid line is the discriminant function YðISI; CVISIÞ¼0 that deﬁnes a theoretical boundary between serotonergic and nonserotonergic cells [35]. Two neurons with too low spontaneous discharge do not appear in this graph. ⁄P < .05; ⁄⁄P < .01; ⁄⁄⁄P < .001; vertical bar, SEM. immersed a large portion of one paw in hot water, whereas they conditions (0.7% halothane in 50% N2O–50% O2) were somewhat applied heat to a limited 2.5-cm patch of tail. Last, our anesthetic different from those of Gao and Mason [35], who used 1% R. Gau et al. / PAINÒ 154 (2013) 647–659 657

halothane in 100% O2. Our lower halothane level should be less The LPGi receives numerous nociceptive projections directly depressive [12,20,54]; however, N2O might interfere because of from deep laminae (IV–VII) of the spinal cord and indirectly from its analgesic properties [74] (see, however, [40,41,60]). Interest- the gigantocellular and the lateral reticular nuclei ingly, halothane depresses the serotonergic neurons firing via the [1,2,14,34,58,70,81], which also receive themselves dense projec- activation of a specific potassium channel [92]. tions from deep laminae (IV–VII) of the spinal cord [17,66,67]. Altogether, the current literature and data reported herein indi- The LPGi receives also major afferent projections from the dorsal cate a relative consensus about the moderate response of RMg periaqueductal gray matter (dPAG), the colliculi, the hypothala- serotonergic neurons to noxious stimuli. By comparison, LPGi sero- mus, and the prefrontal cortex, and thus could be recruited tonergic neurons appeared to be more concerned with nociception. through the activation of the spino-parabrachio-hypothalamic- dPAG nociceptive circuit [9,15,25,39,70]. 4.3. Specificity of the response of B3 serotonergic neurons to noxious stimuli 4.6. Specific role for LPGi serotonergic neurons in nociceptive processing? Auerbach et al. [3] reported that some RMg serotonergic neurons, identified with indirect criteria, responded to noxious stimuli The LPGi serotonergic neurons have been involved in descend- in the unanesthetized cat. Nonetheless, these authors concluded ing control of nociceptive messages [42,51,52,57,82]. The fact that that these responses could be linked to an arousal reaction rather LPGi serotonergic neurons are clearly and specifically activated by than to nociception per se. In the present study, because the cho- intense noxious stimuli (>48°C) now suggests their implication in sen anesthesia prevents any arousal reaction [43], the responses antinociceptive feedback control in response to unbearable pain, of LPGi neurons were very probably specific of noxious stimuli. which triggers a defense reaction [38]. In line with this hypothesis, It might also be argued that B3 neurons were activated indi- the LPGi receives major afferent projection from the dPAG rectly as a result of increase in blood pressure triggered by strong [15,25,70], which is involved in analgesia associated with an in- thermal noxious stimuli used in our study. However, this hypoth- tense defense reaction [16,26,27,53]. The LPGi neurons contribute esis seems unlikely because the great majority of the B3 neurons also to baroreflex inhibition triggered by noxious stimuli and dPAG are not sensitive to blood pressure increase [36,37] and because stimulation [10,21,38] through activation of a dense serotonergic no increase of c-Fos expression was described in the B3 region in projection to the nucleus tractus solitarius [4,86]. Thus, LPGi sero- response to blood pressure changes [22], whereas a marked up- tonergic neurons would be key actors of both analgesia and cardio- regulation of c-Fos was observed in this region after noxious stim- vascular activation that characterize an active reaction of defense ulation [38]. induced either by strong noxious stimuli or dPAG activation. In contrast, although RMg serotonergic neurons are similarly in- 4.4. Responsiveness of B3 serotonergic vs nonserotonergic neurons volved in descending control of nociception, only a lower proportion of them is clearly activated by very intense noxious stimuli. The present study demonstrated that almost all nonserotoner- Thus, their main role in pain control could be different. Indeed, gic neurons within the B3 region were strongly and specifically these neurons receive a predominant afferent projection from the activated (ON, pronociceptive cells) or inhibited (OFF, antinocicep- ventral PAG (vPAG), the role of which contrasts with that of dPAG tive cells) by intense noxious stimuli. According to Fields’s theory [15,25,46]: the stimulation of vPAG induces immobility [31,45], these neurons, which project densely to superficial lami- [5,16,53,69] and pure analgesia [11,26,27]. Furthermore, contrast- nae of the spinal cord [51,52,82], would intervene on noxious stim- ing with the cardiovascular activating action of the dPAG-LPGi ulation-evoked tail (or limb) withdrawal by modulating the system, the vPAG-RMg serotonergic neurons induce mainly transmission of nociceptive messages. slowing-down modulation of cardiovascular function [16,47,48]. Importantly, we also found that a significant number of seroto- Finally, RMg serotonergic neurons might intervene mostly in nergic neurons in B3 group, especially in the LPGi, clearly re- circumstances other than acute nociception, such as intense fear sponded to intense noxious stimuli. Their response consisted of or inescapable pain, which induce passive coping behavior with an increase of firing by 1–6 spike/s, which could seem moderate immobility, decreased cardiac activity, and strong analgesia. in comparison with the response of nonserotoninergic nociceptive neurons in brainstem and spinal cord [8,12,56,90]. However, it is Conflict of interest statement noteworthy that serotonergic neurons never or only rarely fire at a rate higher than 6–8 spike/s [3,13,33,49,62,63,89,95]. On the The authors report no conflict of interest. other hand, convergent data indicated that noxious stimuli do increase serotonin release within the dorsal horn [85,87,93]. Thus, it seems reasonable to hypothesize that noxious stimulation– Acknowledgments induced 1–6 spike/s increase in the firing of B3 serotonergic neurons can have a physiological effect mediated by serotonin This research was supported by grants from INSERM, Université release within the dorsal horn of the spinal cord. Accordingly, acti- Pierre et Marie Curie, Ministère de la Recherche, and the Société vated serotonergic neurons, especially in the LPGi, might decrease Française d’Etude et de Traitement de la Douleur (fellowship to the transmission of nociceptive messages (see [11,68]) in case of RG). intense and sustained noxious stimuli. However, it has also been reported that the B3 serotonergic neurons could have a pronoci- References ceptive action in neuropathic pain [79,84]. [1] Abols IA, Basbaum AI. Afferent connections of the rostral medulla of the cat: a neural substrate for midbrain–medullary interactions in the modulation of 4.5. Nociceptive afferent circuit to LPGi area containing serotonergic pain. J Comp Neurol 1981;201:285–97. neurons [2] Andrezik JA, Chan-Palay V, Palay SL. The nucleus paragigantocellularis lateralis in the rat. Demonstration of afferents by the retrograde transport of horseradish peroxidase. Anat Embryol (Berl) 1981;161:373–90. Importantly, the serotonergic cluster of LPGi (named LPGi in the [3] Auerbach S, Fornal C, Jacobs BL. Response of serotonin-containing neurons in present study) was restricted to a smaller rostroventral portion of nucleus raphe magnus to morphine, noxious stimuli, and periaqueductal gray the LPGi defined in Paxinos and Watson [75] (Fig. 1). stimulation in freely moving cats. Exp Neurol 1985;88:609–28. 658 R. Gau et al. / PAINÒ 154 (2013) 647–659

[4] Babic T, Ciriello J. Medullary and spinal cord projections from cardiovascular [33] Fornal C, Auerbach S, Jacobs BL. Activity of serotonin-containing neurons in responsive sites in the rostral ventromedial medulla. J Comp Neurol nucleus raphe magnus in freely moving cats. Exp Neurol 1985;88:590–608. 2004;469:391–412. [34] Gallager DW, Pert A. Afferents to brain stem nuclei (brain stem raphe, nucleus [5] Bandler R, Shipley MT. Columnar organization in the midbrain periaqueductal reticularis pontis caudalis and nucleus gigantocellularis) in the rat as gray: modules for emotional expression? Trends Neurosci 1994;17:379–89. demonstrated by microiontophoretically applied horseradish peroxidase. [6] Barbaro NM, Hammond DL, Fields HL. Effects of intrathecally administered Brain Res 1978;144:257–75. methysergide and yohimbine on microstimulation-produced antinociception [35] Gao K, Mason P. Serotonergic raphe magnus cells that respond to noxious tail in the rat. Brain Res 1985;343:223–9. heat are not ON or OFF cells. J Neurophysiol 2000;84:1719–25. [7] Basbaum AI, Fields HL. Endogenous pain control mechanisms: review and [36] Gao K, Mason P. The discharge of a subset of serotonergic raphe magnus cells is hypothesis. Ann Neurol 1978;4:451–62. influenced by baroreceptor input. Brain Res 2001;900:306–13. [8] Bernard JF, Besson JM. The spino(trigemino)pontoamygdaloid pathway: [37] Gao K, Mason P. Physiological and anatomic evidence for functional subclasses electrophysiological evidence for an involvement in pain processes. J of serotonergic raphe magnus cells. J Comp Neurol 2001;439:426–39. Neurophysiol 1990;63:473–90. [38] Gau R, Sévoz-Couche C, Laguzzi R, Hamon M, Bernard JF. Inhibition of cardiac [9] Bernard JF, Bester H, Besson JM. Involvement of the spino-parabrachio- baroreflex by noxious thermal stimuli: a key role for lateral amygdaloid and -hypothalamic pathways in the autonomic and affective paragigantocellular serotonergic cells. PAINÒ 2009;146:315–24. emotional aspects of pain. Prog Brain Res 1996;107:243–55. [39] Gauriau C, Bernard JF. Pain pathways and parabrachial circuits in the rat. Exp [10] Bernard JF, Netzer F, Gau R, Hamon M, Laguzzi R, Sévoz-Couche C. Critical role Physiol 2002;87:251–8. of B3 serotonergic cells in baroreflex inhibition during the defense reaction [40] Goto T, Marota JJ, Crosby G. Nitrous oxide induces preemptive analgesia in the triggered by dorsal periaqueductal gray stimulation. J Comp Neurol rat that is antagonized by halothane. Anesthesiology 1994;80:409–16. 2008;506:108–21. [41] Goto T, Marota JJ, Crosby G. Volatile anaesthetics antagonize nitrous oxide and [11] Besson JM, Chaouch A. Descending serotoninergic systems. Pain Headache morphine-induced analgesia in the rat. Br J Anaesth 1996;76:702–6. 1987;9:64–100. [42] Gray BG, Dostrovsky JO. Inhibition of feline spinal cord dorsal horn neurons [12] Bester H, Chapman V, Besson JM, Bernard JF. Physiological properties of the following electrical stimulation of nucleus paragigantocellularis lateralis. A lamina I spinoparabrachial neurons in the rat. J Neurophysiol comparison with nucleus raphe magnus. Brain Res 1985;348:261–73. 2000;83:2239–59. [43] Guilbaud G, Peschanski M, Gautron M, Binder D. Neurons responding to

[13] Bjorvatn B, Fornal CA, Martín FJ, Metzler CW, Jacobs BL. The 5-HT1A receptor noxious stimulation in VB complex and caudal adjacent regions in the Ò antagonist p-MPPI blocks 5-HT1A autoreceptors and increases dorsal raphe thalamus of the rat. PAIN 1980;8:303–18. unit activity in awake cats. Eur J Pharmacol 1998;356:167–78. [44] Hammond DL, Yaksh TL. Antagonism of stimulation-produced antinociception [14] Braz JM, Enquist LW, Basbaum AI. Inputs to serotonergic neurons revealed by intrathecal administration of methysergide or phentolamine. Brain Res by conditional viral transneuronal tracing. J Comp Neurol 2009;514: 1984;298:329–37. 145–60. [45] Heinricher MM, Tavares I, Leith JL, Lumb BM. Descending control of [15] Cameron AA, Khan IA, Westlund KN, Willis WD. The efferent projections of the nociception: specificity, recruitment and plasticity. Brain Res Rev periaqueductal gray in the rat: a Phaseolus vulgaris–leucoagglutinin study. II. 2009;60:214–25. Descending projections. J Comp Neurol 1995;351:585–601. [46] Hermann DM, Luppi PH, Peyron C, Hinckel P, Jouvet M. Afferent projections to [16] Carrive P. The periaqueductal gray and defensive behavior: functional the rat nuclei raphe magnus, raphe pallidus and reticularis gigantocellularis representation and neuronal organization. Behav Brain Res 1993;58:27–47. pars alpha demonstrated by iontophoretic application of choleratoxin (subunit [17] Chaouch A, Menétrey D, Binder D, Besson JM. Neurons at the origin of the b). J Chem Neuroanat 1997;13:1–21. medial component of the bulbopontine spinoreticular tract in the rat: an [47] Inui K, Murase S, Nosaka S. Facilitation of the arterial baroreflex by the anatomical study using horseradish peroxidase retrograde transport. J Comp ventrolateral part of the midbrain periaqueductal grey matter in rats. J Physiol Neurol 1983;214:309–20. 1994;477:89–101. [18] Chen T, Dong YX, Li YQ. Fos expression in serotonergic neurons in the rat [48] Inui K, Nomura J, Murase S, Nosaka S. Facilitation of the arterial baroreflex by brainstem following noxious stimuli: an immunohistochemical double- the preoptic area in anaesthetized rats. J Physiol 1995;488:521–31. labelling study. J Anat 2003;203:579–88. [49] Jacobs BL, Fornal CA. Activity of brain serotonergic neurons in the behaving [19] Chiang CY, Gao B. The modification by systemic morphine of the responses of animal. Pharmacol Rev 1991;43:563–78. serotonergic and non-serotonergic neurons in nucleus raphe magnus to [50] Johansson O, Hökfelt T, Pernow B, Jeffcoate SL, White N, Steinbusch HW, heating the tail. PAINÒ 1986;26:245–57. Verhofstad AA, Emson PC, Spindel E. Immunohistochemical support for three [20] Cole DJ, Kalichman MW, Shapiro HM, Drummond JC. The nonlinear potency of putative transmitters in one neuron: coexistence of 5-hydroxytryptamine, sub-MAC concentrations of nitrous oxide in decreasing the anesthetic substance P– and thyrotropin releasing hormone-like immunoreactivity in requirement of enflurane, halothane, and isoflurane in rats. Anesthesiology medullary neurons projecting to the spinal cord. Neuroscience 1990;73:93–9. 1981;6:1857–81. [21] Comet MA, Sevoz-Couche C, Hanoun N, Hamon M, Laguzzi R. 5-HT-mediated [51] Jones SL, Light AR. Termination patterns of serotoninergic medullary inhibition of cardiac baroreceptor reflex response during defense reaction in raphespinal fibers in the rat lumbar spinal cord: an anterograde the rat. Am J Physiol Heart Circ Physiol 2004;287:H1641–9. immunohistochemical study. J Comp Neurol 1990;297:267–82. [22] Dampney RA, Horiuchi J. Functional organisation of central cardiovascular [52] Jones SL, Light AR. Serotoninergic medullary raphespinal projection to the pathways: studies using c-fos gene expression. Prog Neurobiol lumbar spinal cord in the rat: a retrograde immunohistochemical study. J 2003;71:359–84. Comp Neurol 1992;322:599–610. [23] Dickenson AH, Oliveras JL, Besson JM. Role of the nucleus raphe magnus in [53] Keay KA, Bandler R. Parallel circuits mediating distinct emotional coping opiate analgesia as studied by the microinjection technique in the rat. Brain reactions to different types of stress. Neurosci Biobehav Rev 2001;25:669–78. Res 1979;170:95–111. [54] Kitahata LM, Ghazi-Saidi K, Yamashita M, Kosaka Y, Bonikos C, Taub A. The [24] Ellrich J, Ulucan C, Schnell C. Is the response pattern of on- and off-cells in the depressant effect of halothane and sodium thiopental on the spontaneous and rostral ventromedial medulla to noxious stimulation independent of evoked activity of dorsal horn cells: lamina specificity, time course and dose stimulation site? Exp Brain Res 2001;136:394–9. dependence. J Pharmacol Exp Ther 1995;195:515–21. [25] Fardin V, Oliveras JL, Besson JM. Projections from the periaqueductal gray [55] Lantéri-Minet M, Weil-Fugazza J, de Pommery J, Menétrey D. Hindbrain matter to the B3 cellular area (nucleus raphe magnus and nucleus reticularis structures involved in pain processing as revealed by the expression of c-Fos paragigantocellularis) as revealed by the retrograde transport of horseradish and other immediate early gene proteins. Neuroscience 1994;58:287–98. peroxidase in the rat. J Comp Neurol 1984;223:483–500. [56] Leung CG, Mason P. Physiological survey of medullary raphe and [26] Fardin V, Oliveras JL, Besson JM. A reinvestigation of the analgesic effects magnocellular reticular neurons in the anesthetized rat. J Neurophysiol induced by stimulation of the periaqueductal gray matter in the rat. I. The 1998;80:1630–46. production of behavioral side effects together with analgesia. Brain Res [57] Lovick TA. Analgesia and the cardiovascular changes evoked by stimulating 1984;306:105–23. neurones in the ventrolateral medulla in rats. PAINÒ 1986;25:259–68. [27] Fardin V, Oliveras JL, Besson JM. A reinvestigation of the analgesic effects [58] Lovick TA. Projections from brainstem nuclei to the nucleus induced by stimulation of the periaqueductal gray matter in the rat. II. paragigantocellularis lateralis in the cat. J Auton Nerv Syst 1986;16:1–11. Differential characteristics of the analgesia induced by ventral and dorsal PAG [59] Lovick TA, Wolstencroft JH. Inhibitory effects of nucleus raphe magnus on stimulation. Brain Res 1984;306:125–39. neuronal responses in the spinal trigeminal nucleus to nociceptive compared [28] Fields HL, Basbaum AI. Brainstem control of spinal pain-transmission neurons. with non-nociceptive inputs. PAINÒ 1979;7:135–45. Annu Rev Physiol 1978;40:217–48. [60] Mahmoudi NW, Cole DJ, Shapiro HM. Insufficient anesthetic potency of nitrous [29] Fields HL, Basbaum AI, Clanton CH, Anderson SD. Nucleus raphe magnus oxide in the rat. Anesthesiology 1989;70:345–9. inhibition of spinal cord dorsal horn neurons. Brain Res 1977;126:441–53. [61] Mao J, Mayer DJ, Price DD. Patterns of increased brain activity indicative of [30] Fields HL, Bry J, Hentall I, Zorman G. The activity of neurons in the rostral pain in a rat model of peripheral mononeuropathy. J Neurosci medulla of the rat during withdrawal from noxious heat. J Neurosci 1993;13:2689–702. 1983;3:2545–52. [62] Martin-Cora FJ, Fornal CA, Jacobs BL. Single-unit responses of serotonergic [31] Fields HL, Heinricher MM, Mason P. Neurotransmitters in nociceptive medullary raphe neurons to cardiovascular challenges in freely moving cats. modulatory circuits. Annu Rev Neurosci 1991;14:219–45. Eur J Neurosci 2005;22:3195–204. [32] Fields HL, Vanegas H, Hentall ID, Zorman G. Evidence that disinhibition of [63] Martin-Cora FJ, Fornal CA, Metzler CW, Jacobs BL. Single-unit responses of brain stem neurones contributes to morphine analgesia. Nature serotonergic medullary and pontine raphe neurons to environmental cooling 1983;306:684–6. in freely moving cats. Neuroscience 2000;98:301–9. R. Gau et al. / PAINÒ 154 (2013) 647–659 659

[64] Mason P. Physiological identification of pontomedullary serotonergic neurons [81] Shokunbi MT, Hrycyshyn AW, Flumerfelt BA. Spinal projections to the lateral in the rat. J Neurophysiol 1997;77:1087–98. reticular nucleus in the rat: a retrograde labelling study using horseradish [65] Mayer DJ, Liebeskind JC. Pain reduction by focal electrical stimulation of the peroxidase. J Comp Neurol 1985;239:216–26. brain: an anatomical and behavioral analysis. Brain Res 1974;68:73–93. [82] Skagerberg G, Björklund A. Topographic principles in the spinal projections of [66] Menétrey D, Chaouch A, Besson JM. Location and properties of dorsal horn serotonergic and non-serotonergic brainstem neurons in the rat. Neuroscience neurons at origin of spinoreticular tract in lumbar enlargement of the rat. J 1985;15:445–80. Neurophysiol 1980;44:862–77. [83] Steinbusch HW. Distribution of serotonin-immunoreactivity in the central [67] Menétrey D, Roudier F, Besson JM. Spinal neurons reaching the lateral reticular nervous system of the rat-cell bodies and terminals. Neuroscience nucleus as studied in the rat by retrograde transport of horseradish 1981;6:557–618. peroxidase. J Comp Neurol 1983;220:439–52. [84] Suzuki R, Rygh LJ, Dickenson AH. Bad news from the brain: descending 5-HT [68] Millan MJ. Descending control of pain. Prog Neurobiol 2002;66:355–474. pathways that control spinal pain processing. Trends Pharmacol Sci [69] Morgan MM, Whitney PK. Immobility accompanies the antinociception 2004;25:613–7. mediated by the rostral ventromedial medulla of the rat. Brain Res [85] Taguchi K, Suzuki Y. The response of the 5-hydroxyindole oxidation current to 2000;72:276–81. noxious stimuli in the spinal cord of anesthetized rats: modification by [70] Normandin JJ, Murphy AZ. Nucleus paragigantocellularis afferents in male and morphine. Brain Res 1992;583:150–4. female rats: organization, gonadal steroid receptor expression, and activation [86] Thor KB, Helke CJ. Serotonin- and substance P-containing projections to the during sexual behavior. J Comp Neurol 2008;508:771–94. nucleus tractus solitarii of the rat. J Comp Neurol 1987;265:275–93. [71] Oh SH, Imbe H, Iwai-Liao Y. TMJ inflammation increases Fos expression in the [87] Tyce GM, Yaksh TL. Monoamine release from cat spinal cord by somatic nucleus raphe magnus induced by subsequent formalin injection of the stimuli: an intrinsic modulatory system. J Physiol 1981;314:513–29. masseter or hindpaw of rats. Okajimas Folia Anat Jpn 2006;83:43–52. [88] Vanegas H, Barbaro NM, Fields HL. Tail-flick related activity in medullospinal [72] Oliveras JL, Bourgoin S, Hery F, Besson JM, Hamon M. The topographical neurons. Brain Res 1984;321:135–41. distribution of serotoninergic terminals in the spinal cord of the cat: [89] Veasey SC, Fornal CA, Metzler CW, Jacobs BL. Response of serotonergic caudal biochemical mapping by the combined use of microdissection and raphe neurons in relation to specific motor activities in freely moving cats. J microassay procedures. Brain Res 1977;138:393–406. Neurosci 1995;15:5346–59. [73] Oliveras JL, Redjemi F, Guilbaud G, Besson JM. Analgesia induced by electrical [90] Villanueva L, Bing Z, Bouhassira D, Le Bars D. Encoding of electrical, thermal, stimulation of the inferior centralis nucleus of the raphe in the cat. PAINÒ and mechanical noxious stimuli by subnucleus reticularis dorsalis neurons in 1975;1:139–45. the rat medulla. J Neurophysiol 1989;61:391–402. [74] Parbrook GD. The levels of nitrous oxide analgesia. Br J Anaesth [91] Wang H, Li KY, Wu GC, Cao XD. C-fos expression in spinal cord and brainstem 1967;39:974–82. following noxious stimulation and electroacupuncture plus noxious [75] Paxinos P, Watson C. The rat brain in stereotaxic coordinates. 5th stimulation. Acupunct Electrother Res 1995;20:163–72. ed. London: Academic Press; 2005. [92] Washburn CP, Sirois JE, Talley EM, Guyenet PG, Bayliss DA. Serotonergic raphe [76] Pinault D. A novel single-cell staining procedure performed in vivo under neurons express TASK channel transcripts and a TASK-like pH- and halothane- electrophysiological control: morpho-functional features of juxtacellularly sensitive K+ conductance. J Neurosci 2002;22:1256–65. labeled thalamic cells and other central neurons with biocytin or neurobiotin. J [93] Weil-Fugazza J, Godefroy F, Le Bars D. Increase in 5-HT synthesis in the dorsal Neurosci Methods 1996;65:113–36. part of the spinal cord, induced by a nociceptive stimulus: blockade by [77] Porro CA, Cavazzuti M, Galetti A, Sassatelli L. Functional activity mapping of morphine. Brain Res 1984;297:247–64. the rat brainstem during formalin-induced noxious stimulation. Neuroscience [94] Weksler B, Ng B, Lenert J, Burt M. A simplified method for endotracheal 1991;41:667–80. intubation in the rat. J Appl Physiol 1994;76:1823–5. [78] Potrebic SB, Fields HL, Mason P. Serotonin immunoreactivity is contained in [95] Wessendorf MW, Anderson EG. Single unit studies of identified bulbospinal one physiological cell class in the rat rostral ventromedial medulla. J Neurosci serotonergic units. Brain Res 1983;279:93–103. 1994;14:1655–65. [96] Winkler CW, Hermes SM, Chavkin CI, Drake CT, Morrison SF, Aicher SA. Kappa [79] Ren K, Dubner R. Descending modulation in persistent pain: an update. PAINÒ opioid receptor (KOR) and GAD67 immunoreactivity are found in OFF and 2002;100:1–6. NEUTRAL cells in the rostral ventromedial medulla. J Neurophysiol [80] Rivot JP, Chaouch A, Besson JM. Nucleus raphe magnus modulation of response 2006;96:3465–73. of rat dorsal horn neurons to unmyelinated fiber inputs: partial involvement of [97] Zimmermann M. Ethical guidelines for investigations of experimental pain in serotonergic pathways. J Neurophysiol 1980;44:1039–57. conscious animals. PAINÒ 1983;16:109–10.