The development of the medullary serotonergic systemin early human life☆
Hannah C. Kinney a,⁎, Richard A. Belliveau a, Felicia L. Trachtenberg b,Luciana A. Rava a, David S. Paterson a
a Department of Pathology, Children's Hospital and Harvard Medical School, Boston, MA, United Statesb New England Research Institutes, Watertown, MA, United States
Received 27 February 2006; received in revised form 13 October 2006; accepted 8 November 2006
☆ This work was supported by grants from the First Candle/SIDS Alliance,CJ Murphy Foundation (HCK), National Institute of Child Health andHuman Development (R37-HD20991 [HCK] and PO1-HD36379 [HCK]),Scottish Cot Death Trust (DSP), CJ Foundation (DSP), and Children'sHospital Mental Retardation Core Grant (P30-HD18655).⁎ Corresponding author. Department of Pathology, Enders 1112, Chil-
dren's Hospital, 300 Longwood Avenue, Boston, MA 02115, United States.Tel.: +1 617 919 4508; fax: +1 617 730 0243.
The serotonin (5-HT)-containing neurons of the brain arelocated in rostral and caudal domains in the brainstem, eachdomain with its own connectivity, functions, and embryonicorigins. The rostral domain in the midbrain and upper ponsprojects “rostrally” to virtually all forebrain regions, andparticipates in the mediation of arousal, cognition, mood,motor activity, and cerebral blood flow (Törk and Hornung,1990; Azmitia and Whitaker-Azmitia, 1991; Hornung,2003). The caudal domain in the medulla and caudal ponsprojects “caudally” to other brainstem sites, cerebellum, and
82 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
spinal cord, and modulates respiration, chemosensitivity,cardiovascular function, thermoregulation, upper airwayreflexes, motor activity, pain, and arousal (see below).Recognizing the relationship of the firing of raphé (midline)5-HT neurons to the level of arousal, various investigatorshave suggested different global roles for the caudal 5-HTsystem in mediating different vital functions relative to theindividual's state (McGinty and Harper, 1976; Trulson andJacobs, 1979; Azmitia, 1999; Bradley et al., 2002; Jacobset al., 1990; Jacobs and Fornal, 1999; Lovick, 1997; Mason,2001; Morrision, 2001; Severson et al., 2003). Buildingupon these concepts, we postulate that 5-HT neurons atraphé and extra-raphé and ventral positions within themedulla comprise a system (i.e., the medullary 5-HT system)that modulates homeostatic function in response to metabolicdisruptions in the internal milieu in a state-dependent man-ner. In the medullary 5-HT system, we propose that the raphécomponents with the 5-HT cell bodies are the raphé obscurusand raphé pallidus, the extra-raphé components include theparagigantocellularis lateralis, gigantocellularis, intermedi-ate reticular zone, and subtrigeminal nucleus, and the ventralcomponent includes the arcuate nucleus.
In the following study, we characterized the topographyof the medullary 5-HT system in early life as a first steptowards its future analysis in pediatric brainstem disorders.These disorders include the sudden infant death syndrome(Kinney et al., 2001; Kinney, 2005), Rett syndrome(Paterson et al., 2005), autism (Cook and Leveinthal,1996), and fetal alcohol syndrome (Zhou et al., 2002).Given the important role of 5-HT as a growth factor in earlycell division, migration, and differentiation in the brain(Lauder, 1990; Azmitia, 2001; Buznikov et al., 2001; Luoet al., 2003), 5-HT may be especially critical in the path-ogenesis in these disorders. At the outset of this study, wewere especially interested in determining the relationship ofthe arcuate nucleus along the ventral medullary surface tothis system, as we previously reported that 5-HT neuronswere embedded within it (Paterson et al., 2006), 5-HTreceptors are present within it (Zec et al., 1996; Patersonet al., 2006), and 5-HT receptor binding in the arcuatenucleus (as well as other medullary nuclei that are known tocontain 5-HT neurons) is abnormal in SIDS infantscompared to autopsy controls (Kinney, 2005). In thisstudy, we focused upon the medullary (caudal) 5-HT systembecause of our laboratory's focus upon its role in putativedisorders of autonomic and respiratory control, e.g., SIDS(Kinney, 2005) and Rett syndrome (Paterson et al., 2005).We first defined the topography of the medullary 5-HTsystem in the human infant, and then we established itsdevelopmental profile from the end of the embryonic period(i.e., 8 gestational weeks) through infancy (i.e., the firstpostnatal year). By “topography”, we mean the relativenumber, density, morphological subtype, and regionalposition of the 5-HT neurons. We used immunocytochem-ical techniques with the PH8 antibody that recognizes thesynthetic enzyme for 5-HT, tryptophan hydroxylase
(TPOH), combined with 2- and 3-dimensional, computer-based quantitation and graphics.
2. Materials and methods
2.1. Clinicopathologic database
Medullae from human embryos, fetuses, neonates, infants,and adults were analyzed. Four medullae of human embryoswere obtained with permission from the Laboratory for theStudy of Embryology, University of Washington, Seattle,WA, and were received in 4% paraformaldehyde. The re-maining medullae were obtained from the autopsy services ofthe Departments of Pathology, Brigham and Women's Hos-pital, Boston, and Children's Hospital Boston, MA. Inclusionin the study was based upon the absence of brainstem path-ology by histopathologic criteria. Parental permission andpermission of the Human Protection Committee of eachhospital were obtained in each case. Determination of thegestational age of the therapeutic abortuses (b22 gestationalweeks) was based upon the standard method of foot lengthmeasurement. The age of each fetal and infant case is given asthe postconceptional age (gestational age plus postnatal age).The adult cases were used for comparison as indices ofmaturation.
2.2. Definition of the boundaries of the raphé and extra-raphé neurons in the human medulla
Different investigators historically have defined theboundaries of the subnuclei of the caudal raphé complex indifferent ways (Blessing, 1997; Dahlstrom and Fuxe, 1964;Halliday et al., 1988; Olszewski andBaxter, 1954; Paxinos andHuang, 1995; Törk and Hornung, 1990). In this study, we usedthe operational definitions in the human brainstem proposedby Törk and Hornung (1990). Accordingly, the followingdefinitions were applied: 1) raphé obscurus, extending in therostrocaudal plane through the entire length of the medulla,from the level of the abducens nucleus in the caudal pons(where it coexists with the raphé magnus) to the first cervicalsegment of the spinal cord, and situated between the mediallongitudinal fasciculus and the dorsal aspect of the inferiorolive; 2) raphé pallidus, at or near the midline at thepontomedullary junction, including the medial surface of thepyramids, and ventral to the raphé obscurus. We used thehuman brainstem atlas of Olszewski and Baxter (1954) todefine all other nuclei, except the intermediate reticular zone;for its identification, we used themore recent human brainstematlas of Paxinos and Huang (1995). Of note, we have notincluded the historically-labeled raphé magnus in ourdefinition of the medullary 5-HTsystem, although this nucleusis considered part of the caudal raphé domain (Törk andHornung, 1990). Törk and Hornung (1990) defined the raphémagnus as located in the caudal pons (level of the genu of thefacial nucleus) and at the boundary of the medulla and pons,with the caudal extent at the rostral pole of the inferior olive.
83H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
Olszewski and Baxter (1954) did not define cytoarchitectoni-cally the human raphé magnus; subsequent investigators,however, labeled it, based primarily upon positional homologyof midline 5-HT neurons at the border of the rostral medullaand caudal pons to experimental animals (Blessing, 1997;Anden et al., 1965; Olson et al., 1973; Törk and Hornung,1990). We found that the boundaries of the raphé magnus inthe human infant medulla were ambiguous and difficult todistinguish from the raphé pallidus, and thus, would have beenarbitrarily defined by us; consequently, we did not define it as aseparate nucleus from the raphé pallidus in our char-acterization of the caudal raphé. In addition, we did notstudy 5-HT neurons in the pons, i.e., neurons in the operationaldefinition of the human raphé magnus. As in previous studiesof the chemical anatomy of the developing human brainstem inour laboratory, we divided the entire midbrain, pons, andmedulla into 15 levels based upon discrete nuclear landmarks(Kinney et al., 2001), using the human brainstem atlas ofOlszewski and Baxter (1954) for reference.
2.3. Serotonergic immunocytochemistry
2.3.1. Tissue preparationThe unfixed, frozen medullae were embedded in O.C.T.,
serially cut on a Leitz motorized cryostat at 20 μm thickness,and post-fixed in 4% paraformaldehyde for 24 h. A sampleof sections was post-fixed in 10% formalin. Every tenthsection was immunostained with PH8 antibody for 5-HTlocalization; every twentieth section was stained with cresylviolet for anatomic confirmation of cytoarchitecture. Select-ed medullae were fixed in 4% paraformaldehyde and immu-nostained with a specific antibody to tryptophan hydroxylase(TPOH) (Sigma) for comparison purposes with PH8 immu-nostaining. The immunostaining for the PH8 antibody wasessentially identical if the sections were post-fixed in 4%paraformaldehyde or 10% formalin, as was the immunos-taining using either the PH8 antibody or TPOH antibody (seebelow).
2.3.2. ImmunocytochemistryPH8 antibody (Chemicon, Australia) was used for
localization of 5-HT cell bodies, fibers, and terminals infixed human tissue sections. The PH8 antibody is a murinemonoclonal antibody that binds a common epitope of TPOH,tyrosine hydroxylase, and phenylalanine hydroxylase (Cot-ton et al., 1988). The enzyme TPOH converts 5-hydroxy-tryptophan to 5-HT, and is the key marker of the 5-HTneuronal phenotype. Tyrosine hydroxylase, on the otherhand, converts tyrosine to dihydroxyphenylalanine (L-dopa),a precursor of the catecholamine neurotransmitters dopa-mine, noradrenaline, and adrenaline; phenylalanine hydrox-ylase is also involved in catecholaminergic synthesis in thebrain (Elsworth and Roth, 1997). Thus, tyrosine and phe-nylalanine are markers for catecholaminergic neurons. Infresh tissue, the PH8 antibody binds to TPOH and tyrosinehydroxylase so that it is a marker for 5-HT neurons (Baker
et al., 1991; Haan et al., 1987; Halliday et al., 1990; Törk andHornung, 1990), and catecholaminergic neurons (Harriset al., 1986), respectively. In human tissue fixed in formalin(diluted formaldehyde) or para-formaldehyde (formaldehydederivative), however, the PH8 antibody only binds to TPOHdue to an apparent formaldehyde-induced change in theantigenic determinant of tyrosine hydroxylase (Baker et al.,1991; Haan et al., 1987; Halliday et al., 1990; Törk andHornung, 1990).
Free floating sections fixed in 4% paraformaldehyde werewashed using PBS prior to commencing the staining proce-dure. The samples were treated 3×15 mins in 50% alcohol,and for 20 mins in 50% alcohol and 3% hydrogen peroxide.The tissue was blocked in 10% goat serum for 60 mins inPBS buffer to block endogenous hydrogen peroxide. PH8primary antibody (1:10,000) was applied for 1–3 days tostain 5-HT cells at 4 °C. The tissue was washed (3×15 mins)with PBS, the ABC bridge (ABC Elite, Vector) was addedfor 45–60 mins, and DAB was applied. The free floatingtissue sections were mounted onto gelatinized slides andcover-slipped with Permount. The study tissue sections wererun with known positive control sections and with negativecontrol sections without primary antibody. Positive controlsdemonstrated the expected pattern of immunostaining basedupon observations in the human adult brainstem (Törk andHornung, 1990; Blessing 1997), while specific immunos-taining was absent from negative controls (data not shown).Pre-immune absorption studies were performed by themanufacturer (Chemicon) of the PH8 antibody. Followingimmunostaining with PH8 antibody selected sections fromthe cases were counter-stained with hematoxylin to helpdefine cytoarchitectonic boundaries of the medullary nuclei.A comparison of PH8 immunostaining with specific TPOHimmunostaining, using a commercially available TPOHantibody (Sigma) in human infant brainstem sections dem-onstrated essentially the same patterns of positive labeling,except the nonspecific background was less with PH8 an-tibody. Because of the decreased background, and thepotential to compare our findings in the developing humanbrainstem with those published for the adult brainstem usingPH8 (Törk and Hornung, 1990; Blessing, 1997), we selectedthe PH8 antibody for use in this study.
2.4. Quantitation of 5-HT neurons in the medulla in earlyhuman life
The number, density, and position of 5-HT neurons,including its four morphological phenotypes (see below),were determined in a sample of serial sections through therostrocaudal length of the medulla at selected levels in orderto make comparisons with 5-HT receptor and serotonintransporter binding data obtained at the same levels in thesame cases, or cases matched by age (Zec et al., 1996; Kinneyet al., 2001; Paterson et al., 2004). We quantified 5-HTneurons in 2 sections from each of the following levels of themedulla standardized by us (Kinney et al., 2001). Here we
Fig. 1. Serotonergic neurons in the human infant medulla are distributed from the rostral to caudal level. The specific level of representative cross-sections isindicated by the intersecting lines in the 3-dimensional reconstruction viewed from the ventral surface. Each symbol in the computer-generated plots represents asingle 5-HT neuron. The 5-HT neurons are distributed in midline (raphé) (blue symbols), lateral (extra-raphé) (green symbols), and ventral (red symbols)positions, and are located within the boundaries of nuclei defined by classic cytoarchitectonic methods. The 5-HT neurons in the medulla are intermingled with,or adjacent to the major cranial nerve and other tegmental nuclei involved directly in respiratory and autonomic control, e.g., the nTS, DMX, HG, and PGCL.These latter nuclei are presented as shaded in one level only for illustration, although they all extend the rostrocaudal plane of the medulla. ARC, arcuate nucleus;DMX, dorsal motor nucleus X; GC, gigantocellularis lateralis; HG, hypoglossal nucleus; IRZ, intermediate reticular zone; NTS, nucleus of the solitary tract;PGCL, paragiantocellularis lateralis; Rob, raphé obscurus; RPa, raphé pallidus.
84 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
provide these levels from the rostral to caudal planes with theplate numbers identifying the levels in Olszewski and Baxter(1954), the landmark nuclei that define the level, and thecytoarchitectonically defined nuclei that contain 5-HTneurons, as determined by us (see below): (Plate XVI):Very rostral medulla at the junction with the pons, level ofrostral principal inferior olive, raphé magnus, intermediatereticular zone, gigantocellularis, paragigantocellularis later-alis, and arcuate nucleus; (Plate XIV): Rostral medulla, level
Fig. 2. There are four morphological subtypes of 5-HT neurons in the human infabar=10 μm.
of the (nucleus) n. praepositus (rostral to the hypoglossal n.),raphé obscurus, raphé pallidus, intermediate reticular zone,gigantocellularis, paragigantocellularis lateralis, and arcuatenucleus; (Plate XII):Mid-medulla, level of the n. Roller (justventral to the hypoglossal n.), raphé obscurus, raphé pallidus,intermediate reticular zone, and arcuate nucleus; and (PlateX): Caudal medulla, level of the area postrema, intermediatereticular zone, lateral reticular nucleus, and arcuate nucleus.The human brainstem atlas of Paxinos and Huang (1995) was
nt medulla: A. pyramidal; B. multipolar; C. fusiform; and D. granule. Scale
Fig. 3. The 5-HT topography of the raphé obscurus is demonstrated withPH8 immunostaining. A. With PH8 immunostaining, 5-HT neurons arelocated in almost a single file in bilateral, paramedian “plates”. ×10. Scalebar=80 μm. B. Processes from 5-HT neurons in the paramedian distributioncrossed in the midline to the ipsilateral plate of 5-HT neurons. ×20.
Fig. 4. Serotonergic neurons in raphé pallidus straddle the median sulcus andextend laterally in the boundaries of the historically labeled conterminalisalong the dorsal surface of the pyramids. These 5-HT neurons in theconterminalis lie predominately with their long axes parallel to the dorsalsurface of the pyramid (long arrows). Serotonergic neurons are also locatedin the arcuate nucleus (short arrows). ×10. Scale bar=80 μm.
85H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
used for reference in the identification of the intermediatereticular zone.
Serotonergic neurons were counted in 20 μm sectionsusing computer-based methods with Neurolucida software(Microbrightfield, Incorporated, Williston, VT). In addition,their individual position was plotted in 2-dimensions in themid and rostral standardized levels in order to help visualizegraphically the sequential development of the topography.Infant and adult sections were counterstained with hematox-ylin and eosin to help define nuclear boundaries. For eachsection, the perimeter of the medulla was traced (×2), andlabeled neurons were marked (×10) in the raphé, extra-raphé,and ventral medullary surface. Here we defined the “raphé” asthe 5-HT neurons within the raphé obscurus and raphépallidus. The extra-raphé included 5-HT neurons in theparagigantocellularis lateralis, gigantocellularis, intermediatereticular zone, and subtrigeminal nucleus. The ventral 5-HT
neurons were located in the arcuate nucleus. Immunolabeledcell bodies were counted only if they were morphologicallyidentifiable as neurons, axon and dendrite(s) were visible,and the neuronal cytoplasm had a dense distribution of reac-tion product that excluded the nucleus visible in the section.The position of the neurons were mapped in 2-dimensionswith computer-based (Neurolucida) graphics, with the colorcode of raphé neurons, blue; extra-raphé neurons, green; andventral surface neurons, red. We identified four 5-HT celltypes: granular, multipolar, fusiform, and pyramidal 5-HTneurons. Granular neurons were small and round-to-oval witha large nucleus and one–two thick, cytoplasmic processes.Fusiform neurons were medium in size with spindle-shapedcytoplasm, and two processes, one from each pole. Pyramidalneurons are large with a triangular-shaped cytoplasm with athick process from each of the three points; and multipolarneurons are large with oval-shaped cytoplasm and multipleprocesses extending outward.
2.5. Photomicrograph production
Immunostained sections were visualized with an OlympusBX51microscope (OlympusAmerica Inc.,Melville, NY)withimage capture using an Optronics Microfire S99808 cameraand Microfire 1.0 and Neurolucida 5.0 software (Microbright-field, Williston, VT). Images of 5-HT neurons in the medullawere captured as TIFF files and imported into Photoshop 6.0(Adobe Systems, San Jose, CA) where they were scaledrelative to each other and appropriate labels and scale barswere added.
2.6. Statistical analysis
The number, density, and position of serotonergic neu-rons, in toto and by individual morphological subtypes, werecompared at the 4 defined levels through the medulla among
Fig. 5. Serotonergic neurons extend ventrally and laterally from the regiondefined cytoarchitectonically as the raphé pallidus to line the median surfaceof the pyramids. A. ×4. Scale bar=200 μm. B. ×10. Scale bar=80 μm.
Fig. 6. A. Serotonergic neurons in the paragigantocellularis lateralis are large(arrows) and adjacent to a cluster of smaller 5-HT neurons (arrowheads) justat the ventrolateral surface. ×4. Scale bar=200 μm. B. Serotonergic fibersamid the cluster of 5-HTsmall neurons extend to the medullary surface at thelateral border of the paragigantocellularis lateralis. ×10. Scale bar=80 μm.C. The large neurons are multipolar and pyramidal in shape in the moremedial component. ×10. Scale bar=80 μm.
86 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
the cases from 8 gestational weeks through 108 gestationalweeks, i.e., approximately 16 postnatal months. From 8 weeksthrough early fetal life, i.e., to midgestation (19 weeks), themedullary 5-HT system is described qualitatively due to theimmaturity of its topography and the inability to distinguishthe defined 5-HT cell populations, i.e. raphé, extra-raphé andventral neurons. In a subset of ten cases from midgestationthrough infancy, linear regression of postconceptional age onthe total number of neurons, aswell as by regional position andmorphological 5-HT cell subtype, was performed. Addition-ally, linear regressions of postconceptional age on total celldensity and by regional positions were performed. Analysis ofvariance was used to determine whether the total number ofneurons and the total cell density varied by level, and whetherthis trend varied by age group. Tukey tests were used todetermine further significant differences in the analyses ofvariance. In this subset of 10 fetal and infant cases, the numberand density of 5-HT neurons in the raphé, extra-raphé, andventral surface were quantified as serially-sectioned medullaebecause we were able in these cases to obtain the medullae in
excellent condition, and to fix them directly in 4%paraformaldehyde at the time of autopsy. Serotonergicneuronal positions were mapped in the specimens from casesyounger than 22 weeks, including in embryos, but complete
Fig. 7. Clusters of 5-HT neurons are located in the arcuate nucleus. A. At thejunction of the pyramid with the principal inferior olive, the exit site ofcranial nerve XII, there is a cluster of 5-HT neurons embedded within theexternal fibers along the rim (A; ×10; Scale bar=80 μm.), in which external(5-HT) fibers are present, and the processes of the arcuate 5-HT neurons abutthe medullary surface (B; ×20; Scale bar=40 μm.). C. Processes of the 5-HTneurons in the arcuate nucleus abut the ventral surface, and in someinstances, appear to penetrate the glial limitans into the subarachnoid space(arrows). ×20. Scale bar=40 μm.
Fig. 8. A. Dense 5-HT fibers are located in the external fiber pathway (longarrows), and travel along the medial and ventral rim of the pyramid. ×4.Scale bar=200 μm. B. The source of the 5-HT fibers in the external arcuatepathway includes processes from 5-HT neurons along the ventral rim andembedded within the arcuate nucleus. ×10. Scale bar=80 μm.
87H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
specimens throughout the entire rostrocaudal length were notavailable, and thus, we were not able to quantify 5-HT cellnumber and density systematically. These problems reflect theinherent difficulties in obtaining such human specimens fromvery early pregnancies.
3. Results
3.1. Autopsy database
The human medullae of a total of 30 cases were analyzedwith serotonergic immunocytochemistry, including 4 em-bryos at 7–8 postconceptional weeks, 5 early-gestationalfetuses (range: 9–13 postconceptional weeks), 4 mid-gestational fetuses (19–20 postconceptional weeks), 2 late-
gestational fetuses (25–35 postconceptional weeks), 4 term-neonates (37–42 postconceptional weeks), 6 infants in earlyinfancy (44–58 postconceptional weeks, or approximately1–4.5 postnatal months), and 3 infants in late infancy (≥ 64postconceptional weeks, or approximately 10–17 postnatalmonths). Two adult medullae were analyzed for comparison
Fig. 9. 2-D images of a mid- and rostal levels which illustrate thedevelopmental topography of 5-HT neurons in the human medulla at mid-and rostral levels from the embryonic period through infancy. Each dotrepresents a single 5-HT neuron. The 2-D plots are to scale with each other(scale bar=1 cm). The blue dots represent 5-HT neurons in the raphé, thegreen dots, 5-HT neurons in the extra-raphé, and red dots, 5-HT neurons atthe ventral surface.
Fig. 10. The midline raphé (raphé obscurus primordia) in the human embryoat 8 weeks. A. Diagram illustrating the appearance of the medulla at this age.At this age, the inferior olive has not formed, nor have the pyramids.Abbreviations: r.o.p., raphé obscurus primordial; tr.s., tract of the solitarynucleus; r.l., rhombic lip; corp. rest., restiform body; tr.sp.n, V, tract of thespinal trigeminal nucleus. B. The appearance of the primorida of the raphéobscurus, as demonstrated by PH8 immunostaining for 5-HT neurons. The5-HT neurons form to paramedian plates on either side of the midline. ×10.Scale bar=80 μm.
88 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
as an index of maturity. Although the cases died of a varietyof causes, all brainstems were free of pathology byconventional histological criteria. The postmortem intervalin all cases was ≤24 h. There was no obvious adverse effectof the postmortem interval upon PH8 immunostaining byvisual inspection.
3.2. Overall organization of the infant medullary 5-HTsystem
Irrespective of nuclear labels based upon classic cytoarch-itectonic methods, we found that 5-HT neurons identified byPH8 immunostaining in the human medulla were distributedin the midline (raphé) and lateral (extra-raphé) positions, andalong the ventral and ventrolateral surface. The midline 5-HTneurons were located in the historically-labeled raphéobscurus and raphé pallidus; the extra-raphé 5-HT neuronsin the paragigantocellularis lateralis, gigantocellularis, inter-mediate reticular zone, and lateral reticular nucleus; and theventral 5-HT neurons in the arcuate nucleus (Fig. 1). These 5-HT neurons in the medulla were distributed in raphé, extra-
raphé, and ventral positions which placed them adjacent to, orintermingled with, the neurons in the lower cranial nervenuclei and reticular formation that directly mediate respira-tion, upper airway reflexes, and autonomic function. Thesepopulations include the nucleus of the solitary tract, nucleusambiguus, hypoglossal nucleus, and reticular subnuclei in therostral and caudal ventral medulla that contain neuronalpopulations critical for blood pressure regulation and forpresumably for respiratory rhythm generation (Fig. 1).
3.2.1. Morphological cell types of 5-HT neuronsWith PH8 immunostaining, we identified four 5-HT cell
types in the human infant medulla (Fig. 2), as defined above.The major 5-HT cell type in the medullary 5-HT system wasfusiform, accounting for approximately 50% of the 5-HTneurons in the raphé, with the pyramidal and granule cellsaccounting each for approximately 25%, and the multipolarcells, approximately 2%. This same distribution was alsofound in the extra-raphé component. The majority of the 5-HTcells along the ventral medullary surface was fusiform, andoriented parallel to the surface. The four 5-HT morphologicalsubtypes were intermixed with each other within the raphéand extra-raphé.
3.2.2. The raphé component of the medullary 5-HT systemIn the caudal raphé, 5-HT neurons were located in the
dorsal and paired raphé obscurus, and the ventral andunpaired raphé pallidus (Figs. 1,3–5). In the raphé obscurus,the 5-HT neurons formed thin parallel “plates” that extendedfrom the medial longitudinal fasciculus dorsally to the raphépallidus ventrally in transverse sections (Figs. 1,3), andthrough the entire medulla in the rostrocaudal plane (Fig. 1).The 5-HT neurons within the parallel plates extendedprocesses transversely into the adjacent reticular formation,as well as across the midline to the contralateral plate (Fig. 3).
89H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
Within the unpaired raphé pallidus, scattered 5-HT neuronsformed a cap over the median sulcus between the pyramids atall levels, and were distributed within the medial lemniscus(Figs. 4,5). The 5-HT neurons extended ventrally into thearcuate nucleus along the medial surface of the pyramids,particularly at rostral levels (Fig. 5). Scattered, single, fusiform5-HT neurons extended laterally from the raphé pallidus intothe conterminalis nucleus which is located along the dorsalsurface of the pyramids bilaterally (Fig. 4). These contermi-nalis 5-HT cell bodies were situated with their long axesparallel to the dorsal surface (Fig. 5).
3.2.3. Extra-raphé component of the medullary 5-HT systemSerotonergic cell bodies in the lateral medullary tegmentum
clustered in distinct subnuclei of the reticular formationdefined by cytoarchitectonic criteria (Olszewski and Baxter,1954; Paxinos and Huang, 1995). These subnuclei were thegigantocellularis, paragigantocellularis lateralis, intermediatereticular zone, subtrigeminal nucleus, and lateral reticularnucleus (Fig. 1). The majority of 5-HT neurons in the extra-raphé were concentrated in the paragigantocellularis lateralis(Fig. 1). Serotonergic neurons in the paragigantocellularislateralis were typically oriented in the mediolateral axis, but
Fig. 11. The development of the 5-HT topography of the raphé obscurus at representhe midline. With increasing age, the overall density decreases. A. 35 postconcep(neonate). D. Adult. ×4. Scale bar=200 μm.
also more “haphazardly”, with heavily immunostainedprocesses extending laterally to the ventrolateral surface ofmedulla (Fig. 6). The processes of these neurons were delicateand formed an intricate background “mesh” (Fig. 6). In certaincases, a cluster of small 5-HT neurons was identified in the farlateral region of the paragigantocellularis lateralis bilaterally,close to the ventrolateral medullary surface, and just dorsal tothe inferior olive (Fig. 6). These small 5-HT neurons also hadprocesses which appeared to contact the ventrolateral surfacedirectly. Serotonergic neurons in the intermediate reticularzone were oriented obliquely in transverse sections, andformed a convex arc of cells in the reticular formation, andoverlapped the region of the paragigantocellularis lateralis(Fig. 1).
3.2.4. The ventral component of the medullary 5-HT systemScattered, single, fusiform 5-HT neurons were located in
the medial, ventral, and ventrolateral arcuate nucleus at thesurface of the medulla throughout its rostrocaudal length(Figs. 1,7). These 5-HT neurons extended ventrally andlaterally outward from the raphé pallidus (Fig. 5). In addition,there were clusters of 5-HT neurons along the ventral surface,particularly at the junction of the pyramid with the inferior
tative ages. The 5-HT neurons form two paramedian plates on either side oftional weeks. B. 44 postconceptional ages. C. 58 postconceptional weeks
Fig. 12. A hemisection of the medulla at 11 postconceptional weeks. Theraphé forms two paramedian plates of 5-HT neurons (Ra). The extra-raphé(E-R) is composed of 5-HT neurons in the regions that will develop into theparagigantocellularis lateralis (ventrolateral tegmentum) and gigantocellu-laris (ventromedial tegmentum). The inferior olive has not yet formed. Theregion of the pyramid has been artifactually lost in tissue processing. ×2.Scale bar=400 μm.
90 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
olive at the site of exit of cranial nerve XII. These surface 5-HT neurons were located preferentially, but not exclusively,at the rostral medullary levels (Fig. 1). The processes of the 5-HT neurons in the arcuate nucleus abutted the medullarysurface, and appeared in some instances to penetrate the gliallimitans and contact the subarachnoid space (Fig. 7).Serotonergic neurons were estimated to comprise less than5% of the total neurons in the arcuate nucleus.
3.2.5. Serotonergic neurons outside the boundaries of theraphé, extra-raphé, and ventral components of the medul-lary 5-HT system
Serotonergic neurons were scattered throughout themedullary tegmentum dorsal to the gigantocellularis andparagigantocellularis in seemingly random and inconsistentpatterns among the cases (Fig. 1), with single 5-HT neuronsoccasionally observed in the nucleus of the solitary tract.However, no 5-HT neurons were observed in the cranialnerve nuclei.
3.2.6. Serotonergic fibers in the external arcuate pathwayalong the ventral and ventrolateral medullary surface
Dense 5-HT fibers were located in the external arcuatefiber pathway, and extended along the medial and ventral rimof the pyramid (Fig. 8). These fibers were identified asarising from the 5-HT neurons within the arcuate nucleus atthe ventral and ventrolateral rim of the medullary surface(Fig. 8). They also appeared to project, at least in small part,from neurons in the raphé pallidus, and possibly from theraphé obscurus as well (Fig. 5). These fibers appeared to endjust dorsolateral to the bulge of the developing (convoluted)principal inferior olive, and ventral to the inferior cerebellarpeduncle.
3.2.7. Three-dimensional, computer-based reconstructionsof the distribution of 5-HT neurons in the infant medulla
In the ventral view, the 5-HT neurons formed three distinct“columns” in the rostrocaudal plane. The lateral columns oneither side of the midline were formed by 5-HT neurons of theparagigantocellularis lateralis at the very ventrolateral surfaceat the rostral pole, and were continuous with 5-HT neuronswithin the lateral reticular nucleus and n. subtrigeminalis(Fig. 1). The 5-HT neurons in the midline column werepresent in the raphé (Fig. 1). In the sagittal plane, the dorsalposition of the 5-HT neurons in the extra-raphé (lateralcolumn) to the inferior olive was readily apparent, with“displacement” of these neurons from the immediate ventraland ventromedial position by this large structure. In contrast,the raphé 5-HT neurons extended more ventrally.
3.3. Developmental changes in the topography of medullary5-HT neurons across early human life
The sequential development of the topography of thehuman medulla is illustrated in Fig. 9 in computer-generatedplots from the embryonic period through infancy at compa-
rable mid- and rostral levels, and at the same magnification.This figure provides a reference for the cellular andtopographic changes observed by us across early humandevelopment (Figs. 10–20).
3.3.1. Embryonic development: 7–8 gestational weeksFour specimens were examined at 7–8 gestational weeks,
the earliest time-point available. Given that at this age there isno cytoarchitectonic differentiation of individual nuclei and/orneuronal subtypes to serve as anatomic landmarks, the sectionswere identified as at the level of the medulla by comparison topublished transverse sections of staged, vertebrate embryonicmedullae for the human (Harkmark, 1954; Rakic and Sidman,1982), rat (Lidov and Molliver, 1982), and chick (Harkmark,1954) (Fig. 10). There was no differentiation of alpha-motorneurons, one of the earliest neuronal subtypes to differentiatecytologically (data not shown), in the primordia of the primarymotor nuclei, e.g., hypoglossal nucleus. At this age, theinferior olive primordia were also not formed (Figs. 9,10). Yet,5-HT neurons were already present in the raphé (midline) andextra-raphé positions in cross-sections of the medulla atdifferent levels (Figs. 9,10). The 5-HT neurons weremorphologically small, round, with scant cytoplasm and fewor no processes, and were classified as granular. The other
Fig. 13. The development of the 5-HT neurons in the paragigantocellularis lateralis at representative ages, as demonstrated by PH8 immunostaining. At 12postconceptional weeks (A), the 5-HT neurons are poorly differentiated. At 35 gestational weeks (B), multipolar, fusiform, and pyramidal morphologicalsubtypes are present, and oriented predominately in the lateromedial axis. A. 12 postconceptional weeks. B. 35 postconceptional weeks. C. 44 postconceptionalweeks (neonate). D. 58 postconceptional weeks (infant). ×20. Scale bar=40 μm.
91H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
three 5-HT subtypes were not identified, i.e., multipolar,pyramidal, or fusiform 5-HT neurons. In the raphé, thesemorphologically undifferentiated 5-HT neurons formed two,distinct plates in a paramedian distribution just off the midline(Fig. 10); rarely, a single cell was in the midline itself, andthick, 5-HT processes crossed it. These paramedian 5-HT cellplates in the embryo were essentially identical in location andorientation to those in the fetus, newborn, and infant (Fig. 11).Intermingled with the 5-HT neurons within the plates weresmall, undifferentiated, presumable neuroblasts that werelikely either destined not to be 5-HT neurons, or to attain the 5-HT phenotype at a later stage.
3.3.2. Early fetal development: 9–13 gestational weeksFive fetal medullae were examined at this time-period.
The major change in the overall configuration of the medullafrom the late embryonic period was the presence of theprimordia of the principal inferior olive in all cases, com-posed of a round collection of small neuroblasts in bilateralclusters without convolutions. The two paramedian plates of5-HT neurons remained prominent (Fig. 11). Granular 5-HTneurons were also present in the ventrolateral and ven-
tromedial tegmentum, as illustrated at 11 gestational weeks(Fig. 12), and were in the regions of the major future sub-divisions of the reticular formation that contain 5-HTneurons, notably the gigantocellularis and paragigantocellu-laris lateralis (Figs. 1,13). Serotonergic neurons were noted inthe presumed ventrolateral migratory pathway of post-mitoticneurons from the rhombic lip, such as the inferior olive andarcuate nucleus (Fig. 14) (Rakic and Sidman, 1982). Suchneurons typically had a distinct leading and trailing process,suggesting that they were migrating from the lateral tomidline locations (Fig. 14). Serotonergic neurons were notfound in the rhombic lip itself which, at this age and subse-quent ages, was composed of undifferentiated, small, round,germinal neurons (data not shown). Just ventral to the basalplate, granular 5-HT neurons appeared to stream downwardfrom the basal plate itself into the midline and paramedianregion of the developing raphé obscurus (Fig. 15). Along theventral surface overlying the pyramids, undifferentiated,small, neuroblasts were also present, suggesting that theseneurons were destined for a non-5-HT phenotype, and/orsubsequent 5-HT differentiation at a later stage; theseprimitive neurons were surrounded by delicate 5-HT fibers.
Fig. 14. Serotonergic neurons, as demonstrated by PH8 immunostaining, are identified in the ventrolateral tegmentum at 11 postconceptional weeks (A), andthereafter predominately at the ventral surface in the arcuate nucleus overlying the pyramid (B–D). These neurons have leading and trailing tails, particularly at21, 35, and 44 postconceptional weeks, suggesting the possibility that they are migrating from the lateral to medial positions along the ventrolateral migratorypathway of the rhombic lip, as reinforced by the quantitative data showing a shift in 5-HT neuron number in the extra-raphé to raphé and ventral regions. Theprocesses form part of the external arcuate pathway (C, D). A. 11 postconceptional weeks. B. 21 postconceptional weeks. C. 35 postconceptional weeks(neonate). D. 44 postconceptional weeks (infant). Abbreviation: PYR, pyramid; VLT, ventrolateral tegmentum. ×20. Scale bar=40 μm.
92 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
3.3.3. Mid-fetal development: 15–22 gestational weeksTwo fetal medullae at this age were examined. By 15
gestational weeks, the configuration of the adult-like medullawas taking shape (Fig. 9), with differentiation of alpha-motorneurons in the hypoglossal nucleus, and the formation of theprincipal inferior olive into a large, C-shaped hook, albeit stillcomposed of undifferentiated neuroblasts. By midgestation(20 gestational weeks), the distribution of 5-HT neuronsapproximately “set” in an adult-like pattern by visualinspection (Fig. 9), with subtle quantitative changes in theratios of the raphé, extra-raphé, and ventral neurons relativeto each other through the first postnatal year (Figs. 16–20). Atthis time-period, the four 5-HT morphological subtypes werereadily distinguished. Serotonergic neurons with leading andtrailing processes were noted predominately overlying thepyramids at the ventral surface (Fig. 14).
3.3.4. Late fetal development: 25–35 gestational weeksTwo fetal medullae at this age were examined. By mid-
gestation, 5-HT neurons continued to be concentrated in theraphé obscurus (Fig. 11) and raphé pallidus in the midline,and, at 35 gestational weeks, a sharp distinction between the
raphé obscurus and raphé pallidus was appreciable. At thisage, the 5-HT neurons defining the raphé pallidus in therostral medulla formed a cap over the median sulcus betweenthe pyramids. In the raphé obscurus, the 5-HT neurons werealigned dorsoventrally, and sent processes transversely intothe adjacent reticular formation and across the midline to theother side of the nucleus (Fig. 11). Scattered, single, nowfusiform 5-HT neurons extended laterally from the midlineraphé pallidus into the conterminalis nucleus located alongthe dorsal surface of the pyramid, supporting the possibilitythat this nucleus is a lateral extension of the raphé pallidus.Alternatively, they were 5-HT neurons from the ventralsurface that were “entrapped” dorsal to the pyramids, as theventrally positioned corticospinal tract increased progres-sively in volume. These conterminalis 5-HT cell bodies weresituated predominately with their long axes parallel to thedorsal surface of the pyramid.
By midgestation and thereafter in late gestation, theneonatal period, and infancy, scattered, single 5-HT neuronswere located in the medial and ventral arcuate nucleus at thesurface of the medulla throughout its rostrocaudal length(Fig. 9). These surface neurons were predominately fusiform
Fig. 15. Serotonergic neurons and processes abut the germinal basal plate inthe midline of the floor of the fourth ventricle (vent), and appear to bestreaming out from this zone, as illustrated in a fetus at 11 postconceptionalweeks. Serotonergic fibers are noted overlying the luminal surface of thebasal plate neurons (arrows). PH8 immunostaining counterstained withhematoxylin. ×40. Scale bar=20 μm.
Fig. 16. The proportion of the total number of 5-HT neurons of eachmorphological subtype versus (vs) age.
Fig. 17. The ratio of granular to fusiform neurons versus (vs) age. Withincreasing age, there is a significant decrease in the ratio (p=0.048;slope=0.003 cells per week, 95% CI=[0.006, 0.000]).
93H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
and were oriented parallel to the ventral surface, again withleading and trailing processes. This subtype was consistentwith continued lateral to medial migration, seen first in theembryonic period (see above), and consistent with thequantitative data of a shift in the ratio of extra-raphé to raphéneurons (see below) (Figs. 18–20). They were locatedpreferentially at the rostral medullary levels, as opposed tothe mid- and caudal levels. At 35 gestational weeks, the 5-HT fibers in the external arcuate pathway extended along theventral and ventrolateral surface, and ended just dorsal to thebulge of the developing (convoluted) principal inferior olive,and ventral to the inferior cerebellar peduncle. Serotonergicfibers were also noted in this age-range and thereafter toproject to the ventral surface (Fig. 7).
By midgestation, 5-HT cell bodies in the lateraltegmentum clustered in distinct regions of the reticularformation that have been subparcellated on the basis ofcytoarchitectonic criteria in standard atlases. These regionswere the gigantocellularis (Figs. 9,13), paragigantocellularislateralis (Fig. 9), and intermediate reticular zone (Fig. 9), andeach region contained a mixture of morphological 5-HTsubtypes. Serotonergic neurons in the paragigantocellularislateralis were typically oriented in a mediolateral direction,
heavily-stained dendrites extending great distances (Fig. 13),both medially to the adjacent gigantocellularis, and laterallyto the surface. In this time-frame, the principal inferior olivewas well-convoluted, and the cytoarchitectonic differentia-tion of the cranial nerve nuclei was distinct.
3.3.5. Term birth: 37–42 gestational weeksFour medullae in the neonatal period were examined. By
birth, the adult-like configuration of the topography of 5-HT neurons appeared established by qualitative observa-tion, but subtle (quantitative changes) were still occurring(Figs. 9,16–20). The fusiform neurons along the ventralsurface continued to have leading and trailing processes.
3.3.6. Early infancy: 44–58 postconceptional weeks (firstsix postnatal months)
Three medullae in the early infant period were examined.The quantitative changes in the 5-HT topography at this agecontinued (Figs. 9,16–20).
Fig. 18. Changes in total 5-HT neuronal density (A) and density for the raphé (B) and extra-raphé (C) regions with increasing age, as well as the change in totalarea of the cross-section (D). The changes are significant for all measurements.
94 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
3.3.7. Late infancy: N64 postconceptional weeks (second sixpostnatal months)
Three medullae in the late infant period were examined.Again, differences in the overall 5-HT topography werequantitative since around midgestation (Figs. 9,16–20).
3.3.8. AdulthoodTwo medullae from adults were examined as the index of
maturity for comparison with the developmental cases. Wefound the adult topography of medullary 5-HT neurons to begenerally similar to that presented in the classic anatomicdescriptions of the human rostral and caudal raphé domainsby Törk and Hornung (1990) and Blessing (1997), includingwith 3-dimensional reconstructions (Törk and Hornung,1990). Like these previous investigators, we found 5-HTneurons in the raphé obscurus, raphé pallidus, gigantocellu-laris, and paragigantocellularis lateralis. In the raphéobscurus, the density of 5-HT neurons was qualitativelyreduced (Fig. 11). We also found 5-HT neurons in theintermediate reticular zone, an anatomic designation notused by Törk and Hornung (1990) and Blessing (1997), but
by Paxinos and Huang (1995). Unlike previous investiga-tors, however, we describe and place emphasis upon 5-HTneurons in the adult conterminalis nucleus and arcuatenucleus (Figs. 7,8), similar in cytoarchitecture and positionto those in the infant medulla, but qualitatively less innumber. In the midline of the raphé obscurus (i.e., intra-rapheles [Olszewski and Baxter (1954)], large, multipolarneurons were present that were not serotonergic, in contrastto the paramedian plates of 5-HT neurons, as found in theinfant medulla. Serotonergic fibers were present along theventral and ventrolateral surface in the external arcuatepathway (Figs. 7,8), and extended as far laterally as the entryborder to the inferior cerebellar peduncle, as in the infantmedulla.
3.4. Quantitative changes in cellular parameters in themedullary 5-HT system across early human life
3.4.1. Morphological subtypesBy 20 gestational weeks (midgestation), the ratio of the
5-HT morphological subtypes relative to the total number
Fig. 19. Changes in total 5-HT neuronal number (A) and number for the raphé (B), extra-raphé (C), and ventral (D) regions with increasing age. The changes aresignificant for the 5-HT raphé (B) and ventral (D) neurons.
95H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
of neurons (raphé and extra-raphé combined) were rela-tively established, with approximately 45% of the neuronsfusiform, followed by pyramidal (∼25%), granular(∼25%), and multipolar neurons (∼2%) (Fig. 16). Withincreasing age, however, the percent granular cell signi-ficantly decreased ( p=0.043, slope=−0.1% per week),and the percent fusiform neurons marginally increased( p=0.084, slope, 0.07% per week), although these latterneurons remained the major subtype postnatally (∼50%),as well as prenatally (Fig. 16). At all ages, the multipolarneurons comprised only a small subset (∼2%), and theirpercent of the total number of neurons did not changewith age ( p=0.87) (Fig. 16). The ratios of the four dif-ferent 5-HT cell types to each other versus postconcep-tional age were examined: the only ratio that significantlychanged with age was granular to fusiform neurons whichdecreased with age ( p=0.048, slope=−0.3% neurons perweek) (Fig. 17). Given that the total number of neuronsstayed the same, and the granular neurons were decreasingfaster than the fusiform neurons were increasing (Fig. 17),
it appeared that the fusiform cell increase was probablyvalid.
3.4.2. Developmental changes in the density and number ofthe 5-HT neurons
In order to determine if 5-HT cell density varied by age,we summed the total numbers of neurons and cross-sectionalareas over the four levels sampled (Levels 4, 5, 7, and 8), andthen calculated the density as the number of neurons dividedby the total area. Not surprisingly, the total 5-HT cell density,as well as the densities of 5-HT neurons in the raphé andextra-raphé decreased with age, as the total cross-sectionalarea increased with age (Fig. 18). In the analysis of 5-HT cellnumber, the cell counts were summed over the four levelsthat we sampled to produce a total count (raphé and extra-raphé combined), and the total number of neurons, as well asthe number of the neurons in the raphé and extra-raphé, wasplotted versus increasing postconceptional age (Fig. 19). Ofnote, although the regression of total 5-HT neurons by agewas not significant, and the data appear to be a flat line
Fig. 20. Changes in ratio of the raphé (A), extra raphé (B), and ventral (C) 5-HT neurons to the total number of 5-HT neurons with increasing age, as well as theratio to of the raphé to extra-raphé neurons (B). The changes are significant for all measurements.
96 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
beginning at 33 weeks, the younger case at 22 weeks has amuch lower total number. This case may be randomly low, oralternatively, the total number of cells is indeed lower atmidgestation and becomes “fixed” sometime thereafter, i.e.,between 22 and 33 weeks. To establish the latter possibility,more complete cases for serial sectioning at the lower age areneeded, but were not available to us in the course of thestudy. The combined total number of 5-HT neurons did notchange significantly from midgestation through the end ofthe first postnatal year, nor did the total number of extra-raphé neurons significantly change (Fig. 19). The number of5-HT raphé neurons, however, increased with age( p= 0.021), as did the number of ventral neurons( p=0.001; slope=approximately 1 cell/week) (Fig. 19).The ratio of the 5-HT raphé, extra-raphé, and ventral neuronswere also examined relative to the combined total number(Fig. 20). For this ratio, we calculated the sum of the specific5-HT subpopulations to the sum of the combined totalnumber. The ratio of raphé and ventral neurons relative to thetotal neurons significantly increased with age (p=0.037 andp=0.002, respectively), whereas the ratio of extra-raphéneurons decreased significantly (p=0.009) (Fig. 20). In
addition, the ratio of raphé to extra-raphé neurons increasedsignificantly with age ( p=0.009) (Fig. 20).
4. Discussion
In this study, we delineated the development of thetopographic organization of the medullary 5-HT system inthe human fetus and infant as a first step towards its futureanalysis in pediatric brainstem disorders. We conceptualize aneural system in the medulla that is comprised of 5-HTneurons located in raphé, extra-raphé, and ventral positionsthat modulate homeostatic control in response to differentstressors according to the level of arousal (Kinney et al., 2001;Kinney, 2005). Multiple physiological studies demonstratethat the raphé and extra-raphé nuclei which contain 5-HTneurons, or the 5-HT neurons themselves, influence one ormore homeostatic function, i.e., chemosensitivity (Bradley etal., 2002; Dreshaj et al., 1998; Wang et al., 2001; Penatti et al.,2006), respiration (Bou-Flores et al., 2000; Burnet et al., 2001;Pena and Ramirez, 2002; Gunther et al., 2006; Tryba et al.,2006), upper airway reflexes (Hilaire et al, 1993; Sood et al.,2003; Brandes et al., 2006), blood pressure (Gao and Mason,
97H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
2001a; Henderson et al., 2002; Sevoz-Couche et al., 2006),pain (Mason, 2001), thermoregulation (Berner et al., 1999;Morrison, 1999; Ootsuka and Blessing, 2006), and arousal(Ursin, 2002). In addition, neuroanatomic studies indicate thatthe 5-HT containing nuclei, or specifically the 5-HT neuronswithin them, project to the different raphé and extra-raphécomponents within the system (Zagon, 1993), as well as to the“effector” or “final common pathway” sites in the brainstemand spinal cord, which directly mediate homeostatic functions(Jansen et al., 1995; Loewy and McKellar, 1981). Physiolog-ical and anatomic studies showing that 5-HT neuronsspecifically alter firing, or are otherwise “activated” (i.e.,upregulate c-fos), in response to hypercarbia, hypoxia, thermalstress, and hypoglycemia (Douglas et al., 2001; Erickson andMillhorn, 1994; Maekawa et al., 2000; Wang et al., 2001). Inthis study, we identified certain topographic features of thehuman medullary 5-HTsystem in early development that havenot been previously described or emphasized, and which arediscussed in light in regards to expanding information aboutthe caudal domain of brainstem 5-HT neurons from experi-mental studies.
4.1. The topographic relationship among 5-HT neurons and“homeostatic effector and sensing” regions in the humaninfant medulla
In this study, we found that the 5-HT neurons in themedulla are distributed in raphé, extra-raphé, and ventralpositions which place them adjacent to, or intermingledwith, the “effector” neurons in the lower cranial nerve nucleiand reticular formation that directly mediate respiration,upper airway reflexes, and autonomic function. Theseeffector populations include the nucleus of the solitarytract (viscerosensory input), nucleus ambiguus and peri-ambiguus region (ventral respiratory group and control ofupper airway musculature in the larynx and pharynx),hypoglossal nucleus (tongue control and upper airwaypatency, including during sleep), and reticular subnuclei inthe rostral and caudal ventral medulla that contain neuronalpopulations critical for blood pressure regulation. Thedemonstration by us that radioligand binding to 5-HTreceptor subtypes (Zec et al., 1996; Paterson et al., 2004,2006) and/or the 5-HT transporter (Paterson et al, 2004) areheavily concentrated in these “effector” nuclei in the humanmedulla support the idea that the 5-HT neurons project tothese nuclei in the human. In addition, 5-HT neurons arepresent in the paragigantocellularis lateralis which wepropose contains the human homologue of the rat pre-Bötzinger complex based upon the co-mingling of the 5-HTneurons and their processes with a cluster of neurons thatexpress NK1 receptors, an “anatomic marker” for thiscomplex (Gray et al., 2001; Paterson et al., 2003). Thisrostral ventrolateral region is considered critical forrespiratory rhythm generation (Gray et al., 1999; Smith etal., 1991). Endogenous activation of 5-HT2A receptors isrequired for the generation of respiratory rhythm in the
preBötzinger complex (Pena and Ramirez, 2002; Trybaet al., 2006).
4.1.1. The external arcuate fibers along the ventralmedullary surface
In this study, we found that the external arcuate pathway,which has been historically thought to represent axonalprojections from the arcuate nucleus, contains dense 5-HTfibers. This observation suggests that this pathway arises, atleast in part, from the 5-HT neurons that are heavilyconcentrated in the raphé obscurus and/or raphé pallidus,and not solely from arcuate neurons. Nevertheless, we foundsingle, fusiform 5-HT neurons in the arcuate nucleus thatextended long processes from either pole along the ventralrim in the external arcuate pathway. Thus, this pathway islikely a composite tract comprised of fibers from 5-HTneurons in the raphé obscurus, raphé pallidus, and arcuatenucleus, as well as potentially non-5-HT fibers whose originand target are yet to be defined. Historical observations withhistological stains in human medullae suggested that theexternal arcuate pathway projected to the cerebellum via themidline (fibers of Piccolomini) and/or inferior cerebellarpeduncle, and thus, that the arcuate nucleus was a cerebellar-relay nucleus that formed in part from downward displace-ment of neurons from the (rhombic lip-derived) basis pontis(Essick, 1912; Rasmussen and Peyton, 1946). Our dataindicate that the external arcuate fibers contain 5-HT axonalprojections from the caudal raphé and the ventral surface thatmay project to one another, as well as to the cerebellum.Although we were unable to trace these 5-HT fibers as far asthe inferior cerebellar peduncle in our material, animalstudies now demonstrate specific projections from medullaryraphé 5-HT neurons to the cerebellum, with a variety of 5-HTreceptor subtypes in different cerebellar cortical regions(Dieudonne, 2001).
4.1.2. Serotonergic neurons in the arcuate nucleus as acomponent of the medullary 5-HT system
In this study, we confirmed the previously reported findingby us of single or clustered 5-HT neurons embedded withinthe arcuate nucleus along the ventral medullary surface thatcomprise approximately 5% of the arcuate neuronal popula-tion (Paterson et al., 2006). The ventral surface of the medullaof animals has long been recognized as a site of chemosensi-tivity to carbon dioxide, culminating in the identification ofchemosensitive 5-HT and glutamatergic neurons in thisregion (Bradley et al., 2002;Mulkey et al., 2004; Richerson etal., 2005; Paterson et al., 2006).We have suggested that the 5-HT neurons embedded within the arcuate nucleus arehomologous to 5-HT chemosensitive neurons in animalsbased upon homologies in position and neurotransmitter-phenotype (Filiano et al., 1990; Paterson et al., 2006). Thispossibility is further supported by our observation that 5-HTprocesses in the arcuate nucleus directly abut the ventralsurface and, in some instances, penetrate the glial limitansand extend into the subarachnoid space where they are in
98 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
position to “sense” CO2 levels in the cerebrospinal fluid. Inthis study, we also found that there are three homologouscolumns along the ventromedial and ventrolateral surface ofthe human infant medulla, as demonstrated also by 3-dimen-sional mapping of 5-HT neurons. Based upon extrapolationsfrom the rodent anatomic and physiological data, thetopographic homology of ventral 5-HT cells between ratsand human infants is compelling evidence that the 5-HTneurons in the ventromedial and ventrolateral sites are thechemosensitive respiratory fields in humans. The demon-stration of connections among the arcuate nucleus, raphé, andother regions involved in cardiorespiratory control in thehuman fetus with DiI tracing methods underscores thispossibility (Zec et al., 1997). The 5-HT neurons in the lateralcolumns are specifically located in the lateral (surface) regionof the paragigantocellularis lateralis in both species. Thepossibility that the lateral surface regions of the humanparagigantocellularis lateralis are chemosensitive in humansis supported by clinical observations of CO2 insensitivityduring sleep in adult patients with unilateral, focal strokesinvolving this region, as demonstrated by sophisticatedneuroimaging techniques and sleep studies in these patients(Morrell et al., 2001).
4.2. The development of the human medullary 5-HT system
The development of this system, beginning in the em-bryonic period, is known to result from the interaction ofmultiple growth factors and transcriptional regulators. Thesesignaling factors interact with one another in a cascade-likefashion, leading to the specification, differentiation, andsurvival of 5-HT precursor neurons into mature neurons(Hendricks et al., 1999; Cheng et al., 2003; Cordes, 2005).Serotonergic neurons in the raphé are among the earliestneurons born in the mammalian brain Lidov and Molliver,1982; Rubenstein, 1998), and 5-HT itself appears very earlyin widespread terminal regions of the developing brain, wellbefore it is involved in synaptic transmission (Lauder, 1990).Indeed 5-HT has been shown to act as a growth factor itselfbeginning in early embryogenesis (Buznikov et al., 2001);subsequently, 5-HT contributes to the regulation of neuro-genesis, cell migration, synaptogenesis, and dendritic or-ganization in its terminal fields (Lauder, 1990; Luo et al.,2003). We found that 5-HT neurons in the caudal domainappear as early as 7 gestational weeks in the human embryo.Sundstrom et al. (1993) reported that 5-HT neurons firstappear in the brainstem even earlier, by 5 gestational weeks,the earliest time-point examined by them. At 7 gestationalweeks, we found raphé 5-HT neurons form two plates in aparamedian distribution which we recognized as theprimordia of the raphé obscurus. At this time-point, 5-HTneurons are also widely localized in the ventrolateral andventromedial tegmentum, presumably representing theprimordia of the paragigantocellularis lateralis and gigan-tocellularis by 13 gestational weeks, these neurons clusterinto the early differention into these subnuclei which are
known in the adult to contain the extra-raphé 5-HT neuronalstomata. The development of 5-HT neurons has been ana-lyzed in the human fetus by others (Dahlstrom and Fuxe,1964; Nobin and Bojorklund, 1973; Olson et al., 1993; Shenet al., 1989; Sundstrom et al., 1993; Yew and Chan, 1999),but it has not been systematically studied in the humannewborn or infant, as in this study. Moreover, no previousstudies have focused upon 5-HT neurons at the ventralsurface and in relationship to the external arcuate pathway.
4.3. Developmental shifts in relative 5-HT cell positionsfrom the embryonic period through infancy
In this study, we found that midgestation (around20 weeks), the adult-like raphé and extra-raphé configurationof the medullary 5-HT system appears qualitatively to be “inplace”. Thus, by birth (around 40 weeks), the time-pointwhen the fetus switches to homeostasis independent of thematernal–placental unit, this adult-like configuration hasbeen in place for the last half of gestation. Nevertheless,subtle (quantitative) topographic changes continue to occurfrom midgestation, past birth, and into the first postnatal yearwhich we did not appreciate by visual inspection alone. Thepattern of these ongoing, age-related changes occur in therelative ratios of extra-raphé, raphé, and ventral 5-HTneurons without a concomitant change in the total numberof 5-HT neurons, at least after 33 gestational weeks. Thesedata suggest a shift in 5-HT neurons from extra-raphé to raphéand ventral positions. These data suggest the possibility that5-HT neurons in the extra-raphé region migrate ventrallyalong the medullary surface and medially to the midlinethrough the last half of gestation, past birth, and into infancy.Indeed, this migration may extend beyond infancy, but wehave no information in this regard, given that our analysisstopped at the end of infancy. An alternative explanation isthat neurons destined to develop the 5-HT phenotype expressTPOH at different time-points once they are in their finalpositions. In this scenario, the expression of TPOHexpression is first completed in the extra-raphé, and then inthe raphé and ventral cells.
The 5-HT neurons in the primoridia of the rostral andcaudal raphé in the rodent embryo are generally thought toarise in and migrate ventrally from the basal plate(Rubenstein, 1998). In support of this idea in human raphédevelopment, we found single 5-HT neurons in very closeproximity to the basal plate, as if their precursors migratedventrally from it. Yet, we also observed 5-HT neurons singlyor in small clusters especially along ventral migratorypathway of known rhombic lip-derived neurons in thehuman medulla (Essick, 1912; Rakic and Sidman, 1982),including the arcuate nucleus at the ventral surface,suggesting the possibility that medullary 5-HT neuronslikewise derive, at least in part, from this germinal zone. Thisidea is strengthened by the observation that certain 5-HTneurons in the migratory pathway have leading and trailingprocesses, suggesting the possibility that they are migrating
99H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
from lateral to medial and midline positions. Classic ablationand histological studies indicate that neurons (unspecifiedneurotransmitter phenotype) in the chick caudal raphé andhuman arcuate nucleus derive from the rhombic lip (Essick,1912; Harkmark, 1954; Rakic and Sidman, 1982; Lidov andMolliver, 1982). We did not detect PH8-positive immunos-taining in the rhombic lip itself, suggesting that the 5-HT-specific phenotype does not appear until the destinedprecursor neurons migrate distant from it and/or reach theirfinal addresses. If 5-HT neurons do in fact originate in therhombic lip, it is also very possible that they continue tomigrate postnatally, given the precedent that the neurons ofthe external granular layer in the cerebellum originate in therhombic lip (Rakic and Sidman, 1982), and continue both todivide along the cerebellar cortical surface and migrateinwards to form the internal granular layer in the postnatalperiod until the end of infancy. The idea that 5-HT neurons inthe medulla arise, at least in part, in the rhombic lip, requiresexperimental verification.
4.4. Developmental changes in the morphological 5-HT cellsubtypes
We defined four morphological subtypes of 5-HT neuronsin the medullary 5-HT system in the human fetus and infant,i.e., granular, multipolar, fusiform, and pyramidal. We firstdistinguished these subtypes from each other in the earlysecond trimester. Experimental studies in the raphémagnus inadult rats demonstrate anatomic and physiological subsets of5-HT neurons, notably multipolar, fusiform, and triangular,i.e., identical to pyramidal neurons defined by us, withapproximately the same relative proportions in the adult rat asin the human infant (Gao and Mason, 2001b). The rat studiessuggest further that the subtypes subserve separate and variedphysiological functions (Gao and Mason, 2001b). In thehuman medulla, the functional significance of the subtypes isunknown. With maturation, including into the postnatalperiod, there is a shift in the ratio of subtypes, with a sig-nificant decrease in the granular neurons, and a marginalincrease in the fusiform neurons. Given that the granularsubtype is presumably a less differentiated and complex cellthan the fusiform, we speculate that granular neurons differ-entiate further into fusiform neurons across early humandevelopment in order to serve a “mature” function in themedullary 5-HT system. Interestingly, multipolar neurons,the most complex of the four subtypes, are present early atmidgestation as the smallest subpopulation (around 2%), anddo not increase with fetal, newborn, and/or infant maturation,even as the physiology of the system presumably becomesmore complex in the transition from intra- to extrauterine life.
4.5. Potential limitations of the study
A potential limitation of the study is that the PH8 anti-body is not entirely specific for 5-HT neurons. We controlledfor this possibility, however, by a variety of conditions. The
PH8 antibody has been used extensively to identify spe-cifically 5-HT neurons in fixed human adult humanbrainstem (Blessing, 1997; Halliday et al., 1988; Törk andHornung, 1990). In testing the immunocytochemical proce-dures for this study, we found that PH8 immunostaining inthe developing human brainstem (after midgestation) gaveessentially the same pattern of staining as that reported in theadult human brainstem by Törk and Hornung (1990) andBlessing (1997). Moreover, the PH8 antibody in our casesstained raphé neurons, known to be serotonergic from animalstudies using specific TPOH antibodies, and did not stainneurons in known catecholaminergic sites in the human oranimal brainstem, e.g., substantia nigra, locus coeruleus. Inaddition, in a sample of sections from the same study cases,we found an identical pattern of immunostaining between thePH8 antibody and a commercial antibody specific for TPOH(Sigma), the latter which became available towards the endof our analysis with the PH8 antibody. Thus, we selected thePH8 antibody for use in human developmental brain samplesfixed in formaldehyde or paraformaldehyde in this study, duein part to the decreased background staining with the PH8antibody compared to the commercially available TPOHantibody. An additional benefit is that the use of the PH8antibody allowed us to compare the topography of 5-HTneurons in the medulla in early life with that described in theclassical adult human studies using this same antibody (Törkand Hornung, 1990; Blessing, 1997).
A second potential limitation of the study is the smallsample size used for 5-HT cell quantitation. The difficulty inobtaining human fetal tissue today accounts in large part forthe small sample, compounded by the difficulty of obtaininghuman infant brainstem specimens that are without neuro-pathologic changes. Nevertheless, despite the small samplesize, we found robust and statistically significant age-relatedchanges in the topography of the human medullary 5-HTsystem in early life. A third potential limitation is our strategyfor the quantitation of immunolabeled 5-HT neurons. Wechose not to use stereological techniques because of theirwell-recognized inherent difficulty in immunocytochemicalpreparations, due primarily to the nonuniformity of immu-nostaining through the thickness of individual tissue sections;such irregular staining prevents valid cell counting at thedeeper levels of the section. In addition, we counted 5-HTneurons in anatomically ill-defined regions (i.e., raphé, extra-raphé, and ventral) that do not correspond to distinct nuclei,as characterized by cytoarchitectonic criteria, in either thecross-sectional or rostrocaudal axis. Our strategy in cellcounting was to analyze four of the standardized levels wehave defined in the human brainstem (Kinney et al., 2001),and that we have used extensively in our analysis ofneurotransmitter receptor and transporter binding in tissuesections with autoradiography (Kinney et al., 2001). In thisway, we are able to make comparisons between quantitativecell and binding parameters at different medullary levelsamong cases. Of note, we are not reporting the absolute totalnumber of neurons in an anatomic region per case, but rather,
100 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
the relative numbers of neurons among cases, as establishedby a uniform procedure of quantitation in a rigorously definedsample of sections/case.
5. Conclusions
In conclusion, we found that the first appearance of 5-HTneurons occurs during the embryonic period (first trimester),suggesting that the critical developmental period for themedullary 5-HT system begins close to conception. Conse-quently, 5-HT development may be disrupted by unwittingexposures to certain teratogens (e.g., nicotine in maternalcigarette smoke and alcohol) early in the first trimester(Kinney, 2005), as well as 5-HT related antidepressants(Lauder, 1990). The topography of the human medullary 5-HT system is not fully developed at birth, but rather,continues to change in subtle ways after birth and throughinfancy. Because the development of the topography of thehuman medullary 5-HT system does not end at birth, butcontinues to undergo subtle refinement, we suggest that thesystem is vulnerable to developmental insults both before andafter birth, especially through the end of infancy. Simulta-neous with these topographic changes are dramatic neuro-chemical changes in markers of 5-HT synaptic formation andfunction, i.e., 5-HT receptor subtypes and serotonin trans-porter binding, both pre- and postnatally (Zec et al., 1996;Paterson et al., 2004). Thus, the critical period in thedevelopment of the medullary 5-HT system extends over aprotracted time-frame from the embryonic period at leastthrough the end of the first postnatal year, with different 5-HT-related markers changing with different spatiotemporalprofiles. Consequently, different insults at single or multiplepre- and/or postnatal time-points will likely have differentialeffects upon the structural and/or neurochemical develop-ment of the system, and ultimately upon different functionaloutcomes, not only in early life but beyond. These baselinedata provide a foundation for the future analysis of 5-HT-related brainstem disorders in early life.
Acknowledgments
We are grateful for the helpful comments of Dr. Eugene E.Nattie in manuscript preparation, and the technical assistanceof Ms. Vanessa L. Knoedler. We appreciate the assistance ofDrs. Henry Krous and Brian Blackbourne, and the medicalexaminer system of San Diego, CA, in the collection ofinfant brainstem specimens. We are grateful for the help ofthe Laboratory for the Study of Embryology (NIH 0083),University of Washington, Seattle, WA, in the collection ofembryonic tissue collection.
References
Anden, N.E., Dahlstrom, A., Fuxe, K., et al., 1965. Mapping out of cate-cholamine and 5-hydroxytryptamine neurons innervating the telenceph-alon and diencephalon. Life Sci. 13, 1275–1279.
Azmitia, E.C., 2001. Modern views on an ancient chemical: serotonin effectson cell proliferation, maturation, and apoptosis. Brain Res. Bull. 6,413–424.
Azmitia, E.C., Whitaker-Azmitia, P.M., 1991. Awakening the sleeping giant:anatomy and plasticity of the brain serotonergic system. J. Clin. Psych.52, 4–16 (Suppl.).
Baker, K.G., Halliday, G.M., Halasz, P., et al., 1991. Cytoarchitecture ofserotonin-synthesizing neurons in the pontine tegmentum of the humanbrain. Synapse 7, 301–320.
Berner, N.J., Grahn, D.A., Heller, H.C., 1999. 8-OH-DPAT-sensitiveneurons in the nucleus raphé magnus modulate thermoregulatory outputin rats. Brain Res. 831, 155–164.
Blessing, W.B., 1997. The Lower Brainstem and Bodily Homeostasis.Oxford Univ. Press, New York.
Bou-Flores, C., Lajard, A.M., Monteau, R., De Maeyer, E., Seif, I., Lanoir,J., Hilaire, G., 2000. Abnormal phrenic motorneuron activity andmorphology in neonatal monoamine oxidase A-deficient transgenicmice: possible role of a serotonin excess. J. Neurosci. 20, 4646–4656.
Bradley, S.R., Pieribone, V.A., Wang, W., Severson, C.A., Jacobs, R.A.,Richerson, G.B., 2002. Chemosensitive serotonergic neurons are closelyassociated with large medullary arteries. Nat. Neurosci. 5, 410–412.
Brandes, I.F., Zuperku, E.J., Stucke, A.O., Jakovcevic, D., Hopp, F.A.,Stuth, E.A., 2006. Serotonergic modulation of inspiratory hypoglossalmotorneurons in decerebrate dogs. J. Neurophysiol. 95, 3449–3459.
Burnet, H., Bevengut, M., Chakri, F., Bou-Flores, C., Coulon, P., Gaytan, S.,Pasaro, R., Hilaire, G., 2001. Altered respiratory activity and respiratoryregulations in adult monoamine oxidase A-deficient mice. J. Neurosci.15, 5212–5221.
Buznikov, G.A., Lambert, H.W., Lauder, J.M., 2001. Serotonin andserotonin-like substances as regulators in early embryogenesis andmorphogenesis. Cell Tissue Res. 305, 177–186.
Cheng, L., Chen, C.L., Luo, P., Tan, M., Qui, M., Johnson, R., Ma, Q., 2003.Lmx1b, Pet-1, and Nkx2.2 coordinately specify serotonergic neuro-transmitter phenotype. J. Neurosci. 23, 9961–9967.
Cook, E., Leveinthal, B.L., 1996. The serotonergic system in autism. Curr.Opin. Pediatr. 8, 348–354.
Cordes, S.P., 2005. Molecular genetics of the early hindbrain serotonergicneurons. Clin. Genet. 68, 487–494.
Cotton, R.G.H., McAdam, W., Jennings, I., Morgan, F.J., 1988. Amonoclonal antibody to aromatic amino acid hydroxylases. Identifica-tion of the epitope. Biochem. J. 255, 193–196.
Dahlstrom, A., Fuxe, K., 1964. Evidence for the existence of monoamine-containing neurons in the central nervous system. I. Demonstration ofmonoamines in cell bodies of brain neurons. Acta Physiol. Scand. 232,1–55 (suppl).
Dieudonne, S., 2001. Serotonergic modulation in the cerebellar cortex:cellular, synaptic, and molecular basis. Neuroscientist 7, 207–219.
Dreshaj, I.A., Haxhiu, M.A., Martin, R.J., 1998. Role of the medullary raphénuclei in the respiratory responses to CO2. Respir. Physiol. 111, 15–23.
Elsworth, J.D., Roth, R.H., 1997. Dopamine synthesis, uptake, metabolism,and receptors: relevance to gene therapy of Parkinson's disease. Exp.Neurol. 144, 4–9.
Erickson, J.T., Millhorn, D.E., 1994. Hypoxia and electrical stimulation ofthe carotid sinus nerve induce fos-like immunoreactivity withincatecholaminergic and serotonergic neurons of the rat brainstem.J. Comp. Neurol. 348, 161–182.
Essick, C.R., 1912. The development of the nucleus pontis and nucleusarcuatus in man. Am. J. Anat. 13, 25–54.
Filiano, J.J., Choi, J.C., Kinney, H.C., 1990. Candidate cell populations forrespiratory chemosensitive fields in the human infant medulla. J. Comp.Neurol. 293, 448–495.
101H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
Gao, K., Mason, P., 2001a. The discharge of a subset of serotonergicraphé magnus cells is influenced by baroreceptor input. Brain Res. 900,306–313.
Gao, K., Mason, P., 2001b. Physiological and anatomic evidence forfunctional subclasses of serotonergic raphé magnus cells. J. Comp.Neurol. 439, 426–439.
Gray, P.A., Rekling, J.C., Bocchiaro, C.M., Feldman, J.L., 1999. Modulationof respiratory frequency by peptidergic input to rhythmogenic neurons inthe preBotzinger complex. Science 286, 1566–1568.
Haan, E.A., Jennings, J.C., Cuello, H., Nakata, H., Fujisawa, H., Chow, C.W.,Kushinsky, R., Brittingham, J., Cotton, R.G., 1987. Identification ofserotonergic neurons in human brain by a monoclonal antibody binding toall three aromatic acid hydroxylases. Brain Res. 426, 19–27.
Halliday, G.M., Li, Y.W., Joh, T.H., Cotton, R.G.H., Howe, P.R.C., Geffen,L.B., Blessing, W.W., 1988. Distribution of monoamine-synthesizingneurons in the human medulla oblongata. J. Comp. Neurol. 273,301–317.
Halliday, G.M., Blumbergs, P.C., Cotton, R.G.H., Blessing, W.W., Geffen,L.B., 1990. Loss of brainstem serotonin and substance P-containingneurons in Parkinson's disease. Brain Res. 510, 104–107.
Harkmark, W., 1954. Cell migrations from the rhombic lip to the inferiorolive, the nucleus raphé and the pons; a morphological and experimentalinvestigation on chick embryos. J. Comp. Neurol. 100, 115–209.
Harris, T., Muller, B., Cotton, R.G.H., Voltattorni, C., Bell, C., 1986.Dopaminergic and noradrenergic sympathetic nerves of the dog havedifferent dopa decarboxylase activities. Neurosci. Letts. 65, 155–160.
Henderson, L.A., Yu, P.L., Frysinger, R.C., Galons, J.-P., Bandler, R.,Harper, R.M., 2002. Neural responses to intravenous serotonin revealedby functional magnetic resonance imaging. J. Appl. Physiol. 92,331–342.
Hendricks, T., Frances, N., Dyodorov, D., Deneris, E.S., 1999. The ESTdomain factor Pet-1 is an early and precise marker of central serotoninneurons and interacts with a conserved element in serotonergic genes.J. Neurosci. 19, 10348–10356.
Hilaire, G., Morin, D., Lajard, A.M., Monteau, R., 1993. Changes inserotonin metabolism may elicit obstructive apnoea in the newborn rat.J. Physiol. 446, 367–381.
Hornung, J.P., 2003. The human raphé nuclei and the serotonergic system.J. Chem. Neuroanat. 26, 331–343.
Jacobs, B.L., Wilkinson, L.O., Fornal, C.A., 1990. The role of brainserotonin. A neurophysiologic perspective. Neuropsychopharmacology3, 473–479.
Jansen, A.S., Wessendorf, M.W., Loewy, A.D., 1995. Transneuronallabeling of CNS neuropeptide and monoamine neurons after pseudora-bies virus injections into the stellate ganglion. Brain Res. 683, 1–24.
Kinney, H.C., Filiano, J.J., White, W.F., 2001. Medullary serotonergicnetwork deficiency in the sudden infant death syndrome: review of a 15-year study of a single dataset. J. Neuropathol. Exp. Neurol. 60, 228–247.
Kinney, H.C., 2005. Abnormalities of the medullary serotonergic system inthe sudden infant death syndrome: a review. Pediatr. Dev. Pathol. 8,507–524.
Lauder, J.M., 1990. Ontogeny of the serotonergic system in the rat: serotoninas a developmental signal. Ann. N.Y. Acad. Sci. 600, 297–313.
Lidov, H., Molliver, M., 1982. Immunohistochemical study of thedevelopment of serotonergic neurons in the rat CNS. Brain Res. Bull.9, 559–604.
Loewy, A.D., McKellar, S., 1981. Serotonergic projections from the ventralmedulla to the intermediolateral cell column in the rat. Brain Res. 211,146–152.
Lovick, T.A., 1997. The medullary raphé nuclei: a system for integration andgain control in autonomic and somatomotor responsiveness? Exp.Physiol. 82, 31–41.
Maekawa, F., Oyoda, Y., Torii, N., Miwa, I., Thompson, R.C., Foster, D.L.,Tsukahara, S., Tsukamura, H., Maeda, K., 2000. Localization ofglucokinase-like immunocytochemistry in the rat lower brainstem: forpossible location of brain glucose-sensing mechanisms. Endocrinology18, 375–384.
Mason, P., 2001. Contributions of the medullary raphe and ventromedialreticular region to pain modulation and other homeostatic functions.Annu. Rev. Neurosci. 24, 738–777.
McGinty, D., Harper, R.M., 1976. Dorsal raphé neurons: depression of firingduring sleep in cats. Brain Res. 101, 569–575.
Morrell, M.J., Heywood, P., Moosavi, S.H., Stevens, J., Guz, A., 2001.Central chemosensitivity and breathing asleep in unilateral medullarylesion patients: comparisons to animal data. Respir. Physiol. 129,269–277.
Morrision, S.F., 2001. Differential regulation of brown adipose andsplanchnic sympathetic outflows in rat: roles of raphe and rostralventrolateral medulla neurons. Clin. Exp. Pharmacol. Physiol. 128,138–143.
Morrison, S.F., 1999. RVLM and raphe differentially regulate sympatheticoutflows to splanchnic and brown adipose tissue. Am. J. Physiol. 276,R962–R973.
Mulkey, D.K., Stornetta, R.L., Weston, M.C., Simmons, J.R., Parker, A.,Bayliss, D.A., Guyenet, P.G., 2004. Respiratory control by ventralsurface chemoreceptor neurons in rats. Nat. Neurosci. 7, 1360–1369.
Nobin, A., Bojorklund, A., 1973. Topography of the monoamine neuronsystem in the human brain as revealed in fetuses. Acta Physiol. Scand.388, 1–40 supp.
Olson, L., Boreus, L.O., Seiger, A., 1973. Histochemical demonstration andmapping of 5-hydroxytryptamine and catecholamine-containing neuronsystems in the human fetal brain. Z. Anat. Entwickl-Gesch 139,259–282.
Olszewski, J., Baxter, D., 1954. Cytoarchitecture of the Human Brainstem.Karger.
Ootsuka, Y., Blessing, W.W., 2006. Thermogenesis in brown adipose tissue:increase by 5-HT2A receptor activation and decrease by 5-HT1Areceptor activation in conscious rats. Neurosci. Lett. 395, 170–174.
Paterson, D.S., Knoedler, V.L., Belliveau, R.A., Kinney, H.C., 2003.Serotonergic neurons and NK1 receptors in putative chemosensitivenuclei in the human infant medulla. Soc. Neurosci. Abst. 221.1.
Paterson, D.S., Belliveau, R.A., Trachtenberg, F., Kinney, H.C., 2004.Differential development of 5-HT receptor and the serotonin transporterbinding in the human infant medulla. J. Comp. Neurol. 472, 221–231.
Paterson, D.S., Thompson, E.G., Belliveau, R.A., Antalffy, B.A., Trachten-berg, F.L., Kinney, H.C., 2005. Serotonin transporter abnormality in thedorsal motor nucleus of the vagus in Rett Syndrome: potentialimplications for clinical autonomic dysfunction. J. Neuropathol. Exp.Neurol. 64, 1018–1027.
Paterson, D.S., Thompson, E.G., Kinney, H.C., 2006. Serotonergic andglutamatergic neurons at the ventral medullary surface of the humaninfant: observations relevant to the central chemosensitivity in early life.Auton. Neurosci. 124, 112–124.
Paxinos, G., Huang, X.F., 1995. Atlas of the Human Brainstem. AcademicPress, San Diego.
Pena, F., Ramirez, J.M., 2002. Endogenous activation of serotonin-2Areceptors is required for respiratory rhythm generation in vitro.J. Neurosci. 22, 11055–11064.
Penatti, E.M., Berniker, A.V., Kereshi, B., Cafaro, C., Kellly, M.L., Niblock,M.M., Gao, H.G., Kinney, H.C., Li, A., Nattie, E.E., 2006. Ventilatoryresponse to hypercapnia and hypoxia after extensive lesion of medullaryserotonergic neurons in newborn conscious piglets. J. Appl. Physiol.101, 1177–1188.
102 H.C. Kinney et al. / Autonomic Neuroscience: Basic and Clinical 132 (2007) 81–102
Rakic, P., Sidman, R.L., 1982. Development of the human nervous system.In: Haymaker, W., Adams, R.D., Thomas, C.C. (Eds.), Histology andHistopathology of the Nervous System. Springfield, Illinois, pp. 3–145.
Rasmussen, A.T., Peyton, W.T., 1946. Origin of the ventral arcuate fibersand their continuity with the striae medullares of the fourth ventricle ofman. J. Comp. Neurol. 84, 325–337.
Richerson, G.B., Wang, W., Hodges, M.R., Dohle, C.I., Diez-Sampedro, A.,2005. Homing in on the specific phenotype(s) of central respiratorychemoreceptors. Exp. Physiol. 90, 259–266.
Rubenstein, J.L., 1998. Development of serotonergic neurons and theirprojections. Biol. Psychiatry 44, 145–150.
Severson, C.A., Wang, W., Pieribone, V.A., Dohle, C.I., Richerson, G.B.,2003. Midbrain serotonergic neurons are central pH chemosensors. Nat.Neurosci. 6, 1139–1140.
Sevoz-Couche, C., Comet, M.A., Bernard, J.F., Hamon, M., Laguzzi, R.,2006. Cardiac baroreflex facilitation evoked by hypothalamus andprefrontal stimulation: role of the NTS 5-HT2A receptors. Am. J.Physiol. Regul. Integr. Comp. Physiol. 291, R1007–R1015.
Shen, W.Z., Luo, Z.B., Zheng, D.R., Yew, D.T., 1989. Immunocytochemicalstudies on the development of 5-HT (serotonin) neurons in the nuclei ofthe reticular formation of human fetuses. Pediatr. Neurosci. 15,291–295.
Smith, J.C., Ellenberger, H.H., Ballanyi, K., Richter, D.W., Feldman, J.L.,1991. Pre-Bötzinger complex: a brainstem region that may generaterespiratory rhythm in mammals. Science 254, 726–729.
Sood, S., Liu, X., Liu, H., Nolan, P., Horner, R.L., 2003. 5-HT athypoglossal motor nucleus and respiratory control of geniglossus musclein anesthetized rats. Respir. Physiol. Neurobiol. 138, 205–221.
Sundstrom, E., Kolare, S., Souverbie, F., Samuelsson, E.B., Pschera, H.,Lunell, N.O., Seiger, A., 1993. Neurochemical differentiation of humanbulbospinal monoaminergic neurons during the first trimester. BrainRes. Dev. Brain Res. 75, 1–12.
Törk, I., Hornung, J.P., 1990. Raphe nuclei and the 5-HT system. In:Paxinos, G. (Ed.), The Human Nervous System. Academic Press, SanDiego, pp. 1001–1022.
Trulson, M.E., Jacobs, B.J., 1979. Raphé unit activity in freely moving cats:correlation with the level of behavioral arousal. Brain Res. 163,135–150.
Tryba, A.R., Pena, F., Ramierz, J.M., 2006. Gasping activity in vitro: arhythm dependent on 5-HT2A receptors. J. Neurosci. 26, 2623–2634.
Ursin, R., 2002. Serotonin and sleep. Sleep Med. Rev. 6, 55–69.Wang, W., Tiwari, J.K., Bradley, S.R., Zaykin, A.V., Richerson, G.B., 2001.
Acidosis-stimulated neurons of the medullary raphe are serotonergic.J. Neurophysiol. 85, 224–235.
Yew, D.T., Chan, W.Y., 1999. Early appearance of acetylcholinergic,serotoninergic, and peptidergic neurons and fibers in the developinghuman central nervous system. Microsc. Res. Tech. 45, 389–400.
Zagon, A., 1993. Innervation of serotonergic medullary raphé neurons fromcells of the rostral ventrolateral medulla in rats. Neuroscience 55,849–867.
Zec, N., Filiano, J.J., Kinney, H.C., 1997. Anatomic relationships of the humanarcuate nucleus of the medulla: a DiI labeling study. J. Neuropathol. Exp.Neurol. 56, 509–522.
Zec, N., Filiano, J.J., Panigrahy, A., White, F.W., Kinney, H.C., 1996.Developmental changes in [3H]-lysergic acid diethylamide ([3H]LSD)binding to serotonin receptors in the human brainstem. J. Neuropathol.Exp. Neurol. 55, 114–126.
Zhou, F.C., Sari, Y., Li, T.K., Goodlett, C., Azmitia, E.C., 2002. Deviationsin brain early serotonergic development as a result of fetal alcoholexposure. Neurotox. Rev. 4, 337–342.
