B R I E F COM M U N I C AT I O N S

NATURE NEUROSCIENCE VOLUME 6 | NUMBER 11 | NOVEMBER 2003 1139

Midbrain serotonergic neuronsare central pH chemoreceptorsChristopher A Severson1, Wengang Wang1, Vincent A Pieribone2,3,Carolin I Dohle4 & George B Richerson1,2,5

Serotonergic neurons in the medulla are central respiratorychemoreceptors. Here we show that serotonergic neurons in the midbrain of rats are also highly chemosensitive to smallchanges in CO2/pH and are closely associated with largepenetrating arteries. We propose that midbrain raphé neuronsare sensors of blood CO2 that maintain pH homeostasis byinducing arousal, anxiety and changes in cerebrovascular tonein response to respiratory acidosis.

Serotonergic raphé neurons are involved in many different brain func-tions, including arousal, emotions, motor control, breathing and auto-nomic output. Despite a tendency to study these neurons in relation toonly a single brain function or disease, their highly divergent projectionsand the homogeneity of their cellular properties1 suggest that there maybe a shared function of serotonergic neurons. This possibility hasrecently led to proposals that serotonergic neurons share a role in facili-tation of motor output2, or homeostasis of neural tissue3. Likewise, wehave proposed that serotonergic neurons in the medulla share a role ascentral respiratory chemoreceptors4,5. This subset of serotonergic neu-rons is closely associated with large branches of the basilar artery4, andthey are highly pH chemosensitive5. Here we examined whether a role incentral chemoreception might be a property shared by all serotonergicneurons in the brainstem, a possibility consistent with the finding thatthere are neurons within both the midbrain6 and medullary7 raphé thatare stimulated by hypercapnia in vivo.

Perforated patch clamp recordings were made from neurons in ratmidbrain slices while PCO2 was changed between 9% and 3% (pH 7.18to 7.58). Neurons were defined as chemosensitive using previously estab-lished criteria (for experimental details, see Supplementary Methodsonline; the Yale Animal Care and Use Committee approved all experi-ments).Acidosis caused an increase in firing rate in a subset of neurons inthe region of the dorsal raphé (Fig. 1a). Of 100 recordings stable enoughto classify the response to acidosis, 16 were acidosis-stimulated, nonewere acidosis-inhibited and 84 did not respond to acidosis.

We then used tissue culture to study midbrain raphé neurons. Thishas been a highly effective method for studying chemosensitivity ofmedullary raphé neurons5, permitting more accurate quantificationof the response to pH changes, and enabling recordings after matura-tion of the response to acidosis. A subset of cultured midbrain

neurons increased their firing rate in response to acidosis (Fig. 1b). Itshould be noted that the criteria used here are designed to classifyneurons as chemosensitive only if they have a large and consistentresponse to physiologically relevant changes in pH. Of 78 recordingsstable enough to define their response to acidosis, 21 neurons wereacidosis-stimulated, with an increase in firing rate to 405 ± 70%(mean ± s.e.m.) of control in response to a decrease in pH from 7.4 to7.2 (see Supplementary Note online). Ten neurons were acidosis-inhibited, and 47 were unresponsive to acidosis.

Serotonergic neurons characteristically display highly regular firing,wide action potentials and a prominent after-hyperpolarization1. Thispattern was seen in 16 of 16 acidosis-stimulated neurons in slices and 18of 21 in culture. Twenty-eight neurons recorded in culture were recov-ered after immunohistochemistry for tryptophan hydroxylase (TPH). Of12 neurons that were immunoreactive for TPH, 9 (75%) were acidosis-stimulated (Fig. 1c,d). The other 3 TPH-immunoreactive neurons alsohad a small subthreshold response. Thus, most midbrain serotonergicneurons are intrinsically chemosensitive.

It has previously been shown that neurons within the raphé pontisand nucleus linealis rostralis enwrap large penetrating arteries8.However, the use of Golgi staining prevented demonstration thatthese neurons were serotonergic, and it was thought that this relation-ship only occurred within these two serotonergic subnuclei. We nowshow that this anatomical specialization is a ubiquitous feature ofserotonergic neurons. Large arteries that branched directly off thebasilar artery often defined the borders of the median raphé and dor-sal raphé, and serotonergic neurons were closely associated with thesepenetrating arteries (Fig. 2). We then confirmed that dorsal raphéneurons next to arteries were chemosensitive by making recordings inslices. Neurons that were acidosis-stimulated and were immediatelyadjacent to large arteries (n = 9) all had the characteristic electrophy-siological signature of serotonergic neurons. A subset of these (n = 3)were recovered and were TPH-immunoreactive (Fig. 3).

There are few physiological parameters that are as tightly regulatedas systemic pH. This is accomplished in part by changes in breathing,

Figure 1 Midbrain serotonergic neurons are pH chemosensitive. (a) Membrane potential (Vm) of a neuron from the dorsal raphé in a ratmidbrain slice. This neuron fired action potentials with a highly regular rateand a waveform typical of a serotonergic neuron. It was stimulated by adecrease in pH. (b) Firing rate of an acidosis-stimulated neuron culturedfrom the medial midbrain and simultaneous recording of bath pH. (c) Firingrate of another cultured midbrain neuron that was also acidosis-stimulated.(d) The neuron in c was immunoreactive for TPH (arrow). Scale bar, 50 µm.

Departments of 1Neurology and 2Cellular & Molecular Physiology, 333 Cedar Street, Yale University, New Haven, Connecticut 06510, USA. 3The John B L PierceLaboratory, 290 Congress Street, New Haven, Connecticut 06510, USA. 4Department of Neurophysiology, Universitätsstrasse 150, Ruhr-Universität, 44801Bochum, Germany. 5The Veteran’s Affairs Medical Center, 950 Campbell Avenue, West Haven, Connecticut 06516, USA. Correspondence should be addressed toG.B.R. (george.richerson@yale.edu).

Published online 28 September 2003; doi:10.1038/nn1130

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

ren

euro

scie

nce

B R I E F COM M U N I C AT I O N S

but there are other responses to hypercapnic acidosis that can be just ascritical for survival, including arousal from sleep9, an intense feeling ofanxiety10 and changes in cerebrovascular tone. We propose that amajor role of serotonergic neurons throughout the brainstem is tomonitor the acid/base status of blood and to initiate multipleresponses aimed at restoring pH to normal, including changes inbehavior, breathing and autonomic control. It is not yet clear how thisrole for serotonergic neurons is related to their roles in brain home-ostasis3 and motor control2, but it is possible that they are intimatelylinked. For example, increased motor output may be important forcombating suffocation, and feedforward stimulation of medullaryraphé neurons by descending motor fibers could mediate the anticipa-tory increase in breathing at the onset of exercise11.

The hypothesis presented here may help explain why the serotoninsystem is involved in three diseases that seem otherwise unrelated: sud-den infant death syndrome (SIDS), panic disorder and migraine. Thereare defects in the medullary serotonin system in SIDS12, which maycause blunting of the hypercapnic ventilatory response4. A defect inmidbrain serotonergic neurons could contribute to this syndrome byblunting the arousal response to suffocation during sleep9. The sero-tonin system is also central to panic disorder13, which could be due toinappropriate activation of a ‘false suffocation alarm’10 as well as toabnormally sensitive central chemoreceptors14. Our results suggest thatserotonergic neurons are the locus of this link among breathing,chemoreception and panic disorder. Migraine headaches, whichinvolve dysfunction of cerebral blood vessels, are treated with manydrugs that act via serotonin receptors and may be initiated by activation

of the dorsal raphé15. Thus, these three disorders involve abnormalitiesof brain functions (breathing, arousal, anxiety and cerebral blood flow)that may seem unrelated, but are all critically important for pH regula-tion. We propose that serotonergic neurons in the medulla and mid-brain share a function as central pH chemoreceptors.

Note: Supplementary information is available on the Nature Neuroscience website.

ACKNOWLEDGMENTSThis work was supported by the US National Institutes of Health (C.A.S., W.W.,V.A.P. & G.B.R.), the Department of Veterans Affairs (G.B.R.) and the GermanMerit Foundation (C.I.D.).

COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.

Received 20 June; accepted 5 September 2003Published online at http://www.nature.com/natureneuroscience/

1. Jacobs, B.L. & Azmitia, E.C. Physiol. Rev. 72, 165–229 (1992).2. Jacobs, B.L. & Fornal, C.A. Curr. Opin. Neurobiol. 7, 820–825 (1997).3. Azmitia, E.C. Neuropsychopharmacology 21, S33–S45 (1999).4. Bradley, S. R. et al. Nat. Neurosci. 5, 401–402 (2002).5. Wang, W., Zaykin, A.V., Tiwari, J.K., Bradley, S.R. & Richerson, G.B. J. Neurophysiol.

85, 2224–2235 (2001).6. Veasey, S.C., Fornal, C.A., Metzler, C.W. & Jacobs, B.L. Neuroscience 79, 161–169

(1997).7. Veasey, S.C., Fornal, C.A., Metzler, C.W. & Jacobs, B.L. J. Neurosci. 15, 5346–5359

(1995).8. Scheibel, M.E., Tomiyasu, U. & Scheibel, A.B. Exp. Neurol. 47, 316–329 (1975).9. Berry, R.B. & Gleeson, K. Sleep 20, 654–675 (1997).10. Klein, D.F. Archiv. Gen. Psychiatry 50, 306–317 (1993).11. Eldridge, F.L. & Waldrop, T.G. in Exercise: Pulmonary Physiology and Pathophysiology

(eds. Whipp, B.J. & Wasserman, K.) 309–359 (Marcel Dekker, NewYork, 1991).12. Kinney, H.C., Filiano, J.J. & White, W.F. J. Neuropathol. Exp. Neurol. 60, 228–247

(2001).13. Grove, G., Coplan, J.D. & Hollander, E. J. Neuropsychiatry Clin. Neurosci. 9,

198–207 (1997).14. Gorman, J.M., Liebowitz, M.R., Fyer, A.J. & Stein, J. Am. J. Psych. 146, 148–161

(1989).15. Weiller, C. et al. Nat. Med. 1, 658–660 (1995).

1140 VOLUME 6 | NUMBER 11 | NOVEMBER 2003 NATURE NEUROSCIENCE

Figure 2 Midbrain serotonergic neurons are closely associated with largearteries. Shown are confocal images of transverse midbrain slices afterinjecting arteries with fluorescent albumin (red) and performingimmunohistochemistry for TPH (green). (a) Serotonergic neurons in themedian raphé were concentrated between pairs of large branches of thebasilar artery that penetrated the ventral surface of the midbrain. (b) Medianraphé at high power. Serotonergic neurons frequently had processes that wereintimately associated with arterial walls (clear space between the neuronsand the red arterial lumen). As this artery enters the plane of section near thebottom, neuronal processes (yellow) can be seen wrapping around the artery.(c) Caudal portion of the dorsal raphé. The dorsal raphé was outlined by thetwo paired arteries as they diverged from the midline, and tracked their distalbranches along the surface of the fourth ventricle. (d–f) Rostral portion of thedorsal raphé. As the paired arteries diverged further from the midline, thedorsal raphé took on this new shape. Serotonergic neurons were clustered atthe points of bifurcation of arteries, and tracked the branches laterally. Notethat arteries were usually symmetric across the midline, but some were cut offby the slicing procedure. Scale bar (a,c–f) 200 µm, (b) 50 µm.

Figure 3 Acidosis-stimulated neurons next to arteries were serotonergic. (a) Recording of a dorsal raphé neuron in a brain slice. (b) Biocytin injectionof the neuron in a. This neuron (green) was next to an artery (red) that wasimmunostained with an antibody for smooth-muscle α-actin. (c) The recordedneuron (arrow) was also immunoreactive for TPH. Scale bar, 50 µm.

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

ren

euro

scie

nce