Main

  • “By an apparent contradiction, it maintains its stability only if it is excitable and capable of modifying itself according to external stimuli, and adjusting its response to the stimulation. In a sense it is stable because it is modifiable—the slight instability is the necessary condition for the true stability of the organism.” (W.B. Cannon 1929).

In 1948, Rapport and colleagues purified and identified a chemical 5-hydroxytrptamine (5-HT) (Rapport et al. 1948) present in serum, which had powerful effects on smooth muscle contraction. They named it serotonin. This chemical exist in nearly every biological organism studied including plants, invertebrates, and vertebrates (Garattini and Valzelli 1965; Saxena et al. 1966; Smith 1971; Fischer 1971). The wide distribution in plants and animals indicates this system is phylogentically ancient and evolved very early. Consistent with this idea, the receptors for serotonin, especially the high-affinity 5-HT1 subtype, appears to be over a billion years old (Peroutka and Howell 1994). In mammals, high levels of 5-HT are seen in gut, lung, kidney, testis, superior ganglia and brain (Twarog and Page 1953; Verhofstad et al. 1981; Gershon 1991). These issues are raised here because whatever function serotonin plays in the brain, it should be relevant to its’ evolutionary history.

Despite this ubiquitous natural distribution, most functions attributed to serotonin have centered on its effects on specific neurons and distinct behavior. There are several ideas, which proposed a more general function for serotonin. Serotonin was proposed to be essential for normal mental health (Woolley 1961). This hypothesis was based on its similar structure to LSD, discussed more fully in the chapter by Whitaker-Azmitia (1999). Brodie and Shore (1957) proposed a metabolic role for serotonin in the neuronal activity of the brain. In their hypothesis, norepinephrine and serotonin modulated opposite systems in the brain based on Hess's (1954) concept of the functional integration of the autonomic system with the CNS. Serotonin was the modulator of the trophotrophic system, which integrates behavioral patterns that are recuperative in nature. This was considered a recessive system, which normally functions during sleep or hibernation. Using Golgi-stained brainstem material, a close relationship is seen between the raphe reticular neurons and blood vessels (Scheibel et al. 1975). Based on purely anatomical grounds, these neurons were proposed to function either as a chemoreceptor or a mechanoreceptor, although a neuroscretory role of a bioactive substance into the vascular system from these raphe neurons could not be ruled out. We now propose that the raphe neurons are a key component of neural tissue homeostasis because they are plastic and respond to a variety of neuronal and non-neuronal factors. Our dynamic concept encompasses the concepts of Woolley (1961), Brodie and Shore (1957), and Scheibel et al.(1975). The serotonin neurons evolved as general regulatory system to respond to external stimuli by continually modifying themselves. The fluctuations in serotonin levels are broadcast throughout the brain and serve to dynamically integrate and stabilize CNS structure and function.

THE BRAINSTEM RAPHE NUCLEI: ORGANIZATION AND CHEMICAL LOCALIZATION

Serotonin neuroanatomy is, at first examination, an enigma. The serotonergic raphe nuclei form a chemically homogenous reticular group of neurons which extend from the cervical spinal cord to the interpeduncular nucleus (see Jacobs and Azmitia 1992). The nuclear boundaries of the raphe neurons sometimes conform to classical nuclear designations such as the Nucleus Centralis Superior or Caudal Linear Nucleus (Figure 1). At other times, the cells seems scattered in unconventional locations such as the supralemiscal nucleus designated B9, para-gigantocellularis lateralis, or within the central reticularis region. With respect to development, the neurons form a superior and inferior group of immature cells which have distinct maturational and migrational patterns (Wallace and Lauder 1983; Lidov and Molliver 1982). Even within the superior group, there is evidence the 5-HT neurons form different subsets of cells. The Nucleus Centralis Superior (including the intrafascicular portion of the Dorsal Raphe Nucleus) comes from a different group of neurons than the Linear Caudal Nucleus and the rostral and dorsal part of the Dorsal Raphe Nucleus (see Azmitia and Gannon 1986).

Figure 1
figure 1

A schematic drawing of the main serotonergic projections in the human brain and spinal cord from the brainstem raphe nuclei. The figure shows the extensive efferents of this set of reticular nuclei. DRM, dosal raphe nucleus; CSD, centralis superior nucleus, pars dorsalis; CSM, centralis superior nucleus, pars medianus; RO, nucleus raphe obscruus; NRPa, nucleus raphe pallidus; RM, nucleus raphe magnus; VR, nucleus raphe ventricularis (modified from Azmitia and Gannon 1986).

Full size image

In fluorescence (Dahlstrom and Fuxe 1964) and immunocytochemical (Steinbusch 1981) studies, the cells containing serotonin appear either large or small, having either a multipolar or fusiform shaped perikaryon. The large cells with many spines were first drawn by Cajal (1911) using the Golgi method in Histologie du Systeme Nerveux published in 1911. In electron microscopic studies, the nucleus is seen as being highly invaginated and the cytoplasm has a well-defined Golgi apparatus and abundant microcanuculi (Leger and Descarries 1978; Wiklund et al. 1981; Johnson and Yee 1995) (Figure 2). Both small clear vesicles and dense core vesicles have been described. The major enzymes concentrated in these neurons are tryptophan hydroxylase and aromatic amino-acid decarboxylase, both of which are necessary to synthesize serotonin from tryptophan (Grahame-Smith 1964; Lovenberg et al. 1967). These cells also contain high levels of monoamine oxidase-B, the degradative enzyme having a low affinity for serotonin (Levitt et al. 1982). NADPH diaphorase, necessary for producing nitric oxide, is also found in the large raphe neurons (Rodrigo et al. 1994). Furthermore, the raphe cells have large amounts of the growth associated protein GAP-43, during development and in the adult brain (Curtis et al. 1993; Zou et al. 1996).

Figure 2
figure 2

An electron microscopic photograph taken of the Dorsal Raphe Nucleus of the adult rat. The neuron has a well-developed Golgi Apparatus and numerous organelles. The nucleus is indented. A pericellular oligodendroglia has a large area of specialized contact with the neuron.

Full size image

The raphe serotonergic neurons contain bioactive neuropeptides. The first neuropeptide reported in a serotonergic cell was substance-P in the Nucleus Raphe Magnus (Chan-Palay 1981; Magoul et al. 1986; Halliday et al. 1988a; Arvidsson et al. 1994). Glutamate, an excitatory neurotransmitter, is also co-localized with serotonin and substance-P in raphe neurons (Nicholas et al. 1992). Other peptides described within the serotonergic neurons are calretinin (Acsady et al. 1993), galanin (Arvidsson et al. 1991), enkephalin (Millhorn et al. 1989; Henry and Manaker 1998), N-acetyl-aspartyl-glutamate (Forloni et al. 1987), neuropeptide-Y (Halliday et al. 1988b; Krukoff et al. 1992), angiotensin II (Krukoff et al. 1992), and thyrotropin releasing hormone (Ulfhake et al. 1987; Sharif et al. 1989; Arvidsson et al. 1994). It has been suggested that serotonin is localized to the small clear vesicles and peptides are concentrated in the dense core vesicles (Pelletier et al. 1977; Johansson et al. 1980, Van Bockstaele and Chan 1997). There is evidence for the co-localization of peptides and serotonin in the dense core vesicles (Pelletier et al. 1981). It would be interesting to know if glutamate exists within both the small and dense core vesicles with serotonin.

The serotonergic neuronal cell bodies have variable sizes (15–60 um diameter), which can be dependent on the hormonal state of the animal. In adrenalectomized animals, lacking circulating adrenal glucocorticoids, the serotonergic neurons in all the raphe groups appear small with thin processes extending from the soma (Azmitia et al. 1993). If dexamethasone is placed in the drinking water, the size of the soma and the processes increases, with an approximately 80% increase within 24–72 hours in the volume of the tryptophan hydroxylase immunoreactive neurons (Figure 3). The interesting finding was that all serotonergic neurons changed, providing evidence of a common factor linking the plasticity of these neurons. Other hormones and cytokines influencing serotonin include estrogen, testosterone, thyroxine, aldosterone, cholesterol, and interleukins (see Azmitia and Whitaker-Azmitia 1997).

Figure 3
figure 3

A coronal section of the Dorsal Raphe Nucleus stained with an anti-tryptophan hydroxylase antibody in adrenalectomized (ADX) adult rats. The top photograph is from ADX rat maintained for one week on physiological saline solution alone while the bottom picture is from an ADX rat in which dexamethasone (1 mg/l) was added to the saline 72 hours before the end of the experiment (reprinted from Azmitia et al. 1993).

Full size image

THE SEROTONIN DENDRITES AND AFFERENTS

The dendrites of the serotonergic neurons receive processes from serotonergic and non-serotonergic axons and dendrites. The identified axonal terminals seen in the raphe nuclei contain serotonin (Dong and Shen 1986); norepinephrine (Baraban and Aghajanian 1981; Takagi et al. 1981; Lee et al. 1987; Dong and Shen 1986); dopamine (Ferre et al. 1994); acetylcholine (Chen et al. 1992; Honda and Semba 1994); GABA (Skinner et al. 1997; Wang et al. 1996), substance-P (Magoul et al. 1986); CLIP/ACTH (Leger et al. 1994); or neurotensin (Uhl et al. 1979) (Figure 3). Many anatomical regions provide connections to the raphe nuclei; and these include inputs from the cortex, hypothalamus, reticular nuclei, brainstem, spinal cord, and habenula (Hermann et al. 1997; Behzadi et al. 1990; Aghajanian and Wang 1977). Serotonergic dendrites extend into the axonal bundles of the medial longitudinal fasciculus, medial lemniscus, trapezoid body, cerebral peduncles, the superior cerebellar peduncles and other fiber bundles which traverse the brainstem carrying both ascending and descending fibers (Azmitia and Segal 1978). The spectrum of inputs from numerous anatomical sources having diverse chemical neurotransmitters and neuropeptides is not atypical for a large reticular neuron (Figure 4). But given the number of afferents and efferents of this single chemical system, it can transmit the varied signals to both neuronal and non-neuronal targets.

Figure 4
figure 4

A schematic drawing of a serotonergic neuron next to a pericellular oligodendroglia (OLIGO) with a number of molecules inside and outside the neuron: 5HT, serotonin; A, adrenal steroids; ACH, acetylcholine; ACTH, adrenal corticotrophin hormone; AngII, angiotensin II; BH4, tetra-hydro-biopterin; C, cholesterol; Cal, calretinin; DA, dopamine; E, estrogen; Enk, enkephalin; GABA, gamma-amino-butyric acid; Gal, galanin; GAP-43, growth associated protein; GLUT, glutamate; MAO-B, monoamine oxidase-B; NE, norepinephrine; NO, Nitric Oxide; NPY, neuropeptide-Y; Nt, neurotensin; SP, substance P; TRH, thyrotropin releasing hormone; WOH, tryptophan hydroxylase.

Full size image

Specialized contacts are seen as axo-dendritic, axo-somatic, and dendro-dendritic; however, the majority of the processes are seen as being closely appossed without a specialized contact (Malinsky and Malinska 1992; Kapadia et al. 1985; Chazal and Ralston 1987). In addition, the contacts made between the dendrites may have special significance (Pecci Saavedra et al. 1986; Chazal and Ohara 1986). The serotonergic dendrites contain tryptophan hydroxylase and have the ability to synthesize serotonin from tryptophan. The dendrites contain small clear vesicles that can serve as storage and release sites for serotonin. There is probably no cytoplasmic, non-calcium dependent release from dendrites since there is no evidence for the presence of the 5-HT transporter protein on dendrites (Tao-Cheng and Zhou 1997). Any serotonin released from the dendrites would remain in the neuropil for an extended period of time to interact with dendritic and somatic receptors. Eventually, the released neurotransmitter would be sequested by the numerous 5-HT terminals in the raphe nuclei.

An unusual arrangement in the raphe nuclei, is that the serotonergic soma appear to have extensive contacts with non-neuronal cells (Figure 2). Dendritic bundles from the raphe neurons intertwine with processes from tanocytes coming from the third ventricles (Cummings and Felten 1979). Pericellular oligodendroglia make specialized contacts with the soma of the raphe serotonergic neurons (Azmitia 1978). Contacts with endothelial cells, which surround the blood vessels, are common for both soma and dendrites (Scheibel et al. 1975; Azmitia 1978; Westlund et al. 1993; Gragera et al. 1994). Numerous S100β containing astrocytes flank the developing serotonergic neurons but become somewhat less prominent as the brain matures (Van Hartesveldt et al. 1986). The special relationship between serotonin and S100 is discussed below when we deal with issues of neuroplasticity. The glial, endocrine, and vascular contacts are consistent with serotonin's special role in integrating electrical, metabolic, and trophic systems within the brain (Figure 4).

THE SEROTONIN AXONS AND EFFERENTS

Cajal (1911) described the serotonergic neurons as large neurons with extensive but untraceable axonal projections. Using 3H-proline as a marker, at least five separate tracts were found ascending from the superior group of raphe nuclei (Azmitia and Segal 1978). Some serotonergic fibers are myelinated (Figure 5), whereas others are unmyelinated, and a variety of fiber diameters can exist within many brain regions (Kohler et al. 1980; Cropper et al. 1984). Some serotonergic neurons form synapses while others engage in non-synaptic interactions (Azmitia 1978; Beaudet and Descarries 1987; Hornung et al. 1990). The serotonergic neurons, present in all brains studied, have evolved numerous morphological characteristics within the CNS of vertebrates, properties that at times appear incompatible for accomplishing a unified role.

Figure 5
figure 5

An electron microscopic photograph taken of a myelinated axon immunocytochemically stained with an anti-5-HT antibody in an adult monkey brain in the medial forebrain bundle. Note the location of immunoreactivity in the small synaptic vesicles (reprinted from Azmitia and Gannon 1983).

Full size image

Some researchers have attempted to classify serotonergic neurons into two distinct categories principally based on fiber type. One fiber type, which is thick, relatively straight and non-varicose, is said to originate in the Median Raphe Nucleus and the other, which is thin, highly branched and varicose, from the Dorsal Raphe Nucleus (Hornung et al. 1990; Kosofsky and Molliver 1987). This rigid distinction is not universally accepted since it appears the morphology of the fibers may depend on the target region innervated (see Azmitia 1978) and appear to fluctuate during development or after injury. In fact, serotonergic fibers innervate the lateral ventricles from fibers that traveled in the medial forebrain bundle before entering the ventricles anteriorly at the level of the septum (Dinopoulos and Dori 1995). These fibers are initially thick, straight, and non-varicose at 1 week postnatal in the rat. However, these same fibers become thin, highly branched, and varicose by the third week of life. Thus, the appearance of the fiber may reflect their development stage and that of their target cells, rather than being a defining static characteristic of the parent neuron itself.

The innervation of the ependymal cells within the ventricle is only one type of non-neuronal projection, which the raphe neurons show. The fibers flood the entire ventral floor of the fetal brain and project into various target areas in the forebrain and spinal cord. The fibers grow along blood vessels and established fiber connections such as the fornix, fasciculus retroflexus, lemniscal pathways, medial longitudinal fasciculus, and pyramidal tracts (Azmitia and Segal 1978). Serotonergic fibers reach the sympathetic preganglion neurons (Bacon et al. 1990), the sensory glomeruli in the olfactory bulb, the intermediate lobe of the pituitary gland (Mezey et al. 1984), the endothelial cells of the choroid plexus (Napoleone et al. 1982), the lateral ventricles (Aghajanian and Gallager 1975; Lorez and Richards 1982), the motoneurons of brainstem (Connaughton et al. 1986), the spinal cord (Alvarez et al. 1998,) and nearly every region of the cerebral cortex (Molliver 1987). In these diverse targets, specialized contacts are common on motoneurons: the dopaminergic neurons in the substania nigra (Moukhles et al. 1997; Van Bockstaele et al. 1994; Herve et al. 1987) and the noradrenergic neurons in locus coeruleus (Pickel et al. 1978; Leger and Descarries 1978; Segal 1979); on the pace-maker neurons of the suprachiasmatic nucleus (Guy et al. 1987; Moga and Moore 1997); and on specialized calbindin GABAergic interneurons in the hippocampus (Hornung and Celio 1992; Freund 1992). In addition, close appositions are seen thoughout the brain on glial cells, pineal gland (Moller 1976), subcommisural organ (Mollgard and Wiklund 1979; Marcinkiewicz and Bouchaud 1986; Voutsinos et al. 1994), endothelial cells (Kobayashi et al. 1985, Azmitia 1978), and ependymal cells (Aghajanian and Gallager 1975; Lorez and Richards 1982). If the concept of volume transmission is accepted, we may say, with out too much effort, that every cell in the brain is exposed to serotonin.

An important insight to consider is that the flow of serotonin from the brainstem serotonergic neurons is not only anterograde, but also retrograde from the forebrain. Injection of 3H-5-HT is rapidly transported back into the raphe soma from the olfactory bulb (Araneda et al. 1980) and from the hippocampus (Azmitia 1981). The retrograde flow is as robust as the traditional movement of proteins such as horseradish peroxidase and signifies that the role of serotonin synthesis is not only for release at the terminal, but also for intracellular utilization within the perikaryon. Thus serotonin can act as a bi-directional cue from serotonergic terminals to either the target cells by release or to the brainstem raphe nuclei by retrograde transport.

NEUROPLASTICITY OF SEROTONERGIC NEURONS AND IMPACT ON TARGET CELL MATURATION

The extensive innervation pattern is complemented by the precocious growth shown by these fibers in the immature brain. In fact, in drosophila, the serotonergic innervation can be viewed as a consequence of segmentation of the brainstem under the control of the early expressing HOX genes (Hunter and Kenyon 1995). This role of serotonergic neurons as an early organizer suggests it serve to influence overall brain patterning. In addition, serotonin is trophic on its target cells in lobsters (Benton et al. 1997). In the brains of mammals, the serotonergic fibers are among the first fibers to innervate the forebrain and spinal cord from the brainstem (Lidov and Molliver 1982; Wallace and Lauder 1983). The serotonergic fibers innervate cerebral cortex by day 17 of gestation in the rat, just before the cerebral cortex begins to develop a laminar structure (Wallace and Lauder 1983). Studies have shown that delaying the ingrowth of serotonergic fibers, produced significant delays in neuronal maturation as evidenced by delayed final mitosis (Lauder and Krebs 1978), dendritic elongation and spine appearance (Yan et al. 1997), barrel field formation (Blue et al. 1991), and synaptogenesis (Mazer et al. 1997; Wilson et al. 1998). Similar results are seen in the mature brain (Cheng et al. 1994; Azmitia et al. 1995; Whitaker-Azmitia et al. 1995). Tissue culture studies of glial cells indicate that 5HT is a differentiating factor on non-neuronal target cells as well (Whitaker-Azmitia et al. 1990). Thus, serotonin can be considered a differentiating factor to its target cells within the brain, in both vertebrates and invertebrates.

Serotonergic fibers are very plastic in the mature brain. Damage to the long fibers projecting into the spinal cord or to the forebrain induces a vigorous regenerative sprouting response. In the spinal cord, the new regenerating fibers can establish contacts with motoneurons and restore normal spinal reflexes disrupted after the lesion (Nobin et al. 1973). In the forebrain, a similar dynamic response is found after lesioning of serotonergic fibers in the hypothalamus, which leads to loss of 5-HT, and hypersexuality (Frankfurt et al. 1985). Within a few days, new sprouts form on the damaged 5-HT axons and normal density is measured after about 4–5 weeks. The regenerative sprouts are able to correct the hypersensitive response to sexual hormones. The sprouting of serotonergic neurons can also come from undamaged serotonergic neurons when neighboring 5-HT fibers are removed (Azmitia et al. 1978). This homotypic collateral sprouting response can be maintained for over a year and results in both morphological and functional restoration. The signal for sprouting response is loss of serotonin and the mechanism appears to involve glial trophic factor S100 (Azmitia et al. 1990).

In the adult brain, removal of serotonin results in loss of synapses (Cheng et al. 1994; Wilson et al. 1998) reduction in the expression of synaptophysin and MAP-2 (markers for synapses and dendrites, respectively) (Azmitia et al. 1995; Whitaker-Azmitia et al. 1995), and decreased levels of S-100β (Azmitia et al. 1992; Haring et al. 1993). All these events are indicative of a return to an immature, undifferentiated brain stage. (See Figure 6.) This crucial role for serotonin is probably accomplished by its regulation of the soluble, calcium binding protein, S100β, from astrocytes (Whitaker-Azmitia et al. 1990). The gene for S100β is within the obligatory region for Down's syndrome on chromosome 21 (Allore et al. 1988). S100β, within neurons, functions to stabilize the microtubules and permit the elaborate branching pattern seen within the mature neurons in the brain (Nishi et al. 1996). In S100β transgenic mice, the neurons in the hippocampus develop dendritic branches at an accelerated rate (Whitaker-Azmitia et al. 1997). However, the increased rate of dendritic branching seen in the S100β transgenic mice may be a model for Down syndrome, since by six months the dendrites appear to be damaged and reduced in number. This is reminiscent of the usual development of Alzheimer disease by the Down patients in mid-life (Godridge et al. 1987).

Figure 6
figure 6

A schematic picture shows the dynamic changes (curved arrows) in neuronal morphology in the presence or absence of serotonin (5-HT). S100β is a glial-derived calcium binding protein, which had been shown to promote cytoskeleton assembly. S100β is released from astrocytes (see arrow marked with label S100β) by the action of 5-HT1A receptor agonist. In the presence of high Ca++ S100β forms a dimer called S100B. MAP-2 is a microtubule-associated protein which helps to stabilize the cytoskeleton in dendrites. Synaptophysin is a protein found associated with the vesicles in axons (see arrow with label).

Full size image

The functional consequences of removing 5-HT impacts upon nearly every behavioral and biological process studied. A sampling of these functions includes, but is not limited to aggression, learning, sexual behavior, pain, attention, temperature, appetite, steroid secretion, respiratory rate, sleep, and blood flow. Psychiatric and neurological disorders associated with altered brain serotonin activity include depression, schizophrenia, Down syndrome, Alzheimer disease, autism, attention deficit disorder, alcoholism, and sleep apnea (see Figure 7). This broad spectrum of functional changes is compatible with the wide distribution of the 5-HT fibers, but is intriguing because they appear to be associated with changes in 5-HT innervation. (Arango et al. 1997; Storga et al. 1996; Owens and Nemeroff 1994). The evidence for structural changes in the 5-HT system would provide an explanation for the lag-times for many serotonergic drugs to become effective in treating depression and schizophrenia (see Chapters by Mann (1999), Blier and de Montigny (1999), and Meltzer (1999). The serotonin involvement in functional and clinical disorders is even more compelling when the trophic effects of 5-HT on its targets cells is considered. When serotonergic fibers are removed from an area of the brain, both the structure and function of that region are compromised. Thus, the loss of serotonin in the mature brain could compromise the adaptability and stability of the neural tissue (e.g., neurons, glial, vascular) to dynamically and effectively respond to stimuli from the external environment. The result in loss of serotonin's role in neural tissue homeostasis would be mental illness.

Figure 7
figure 7

A schematic drawing of three interacting sphere of interaction showing how external factors act on various systems within the body to impinge on serotonergic neuronal plasticity. The inner sphere comprises brain homeostasis where the plasticity of 5-HT neurons trophic influences target-cell neuronal morphology. The structure of the brain is the basis of all neuronal networks and, thus function. In the outer sphere, the interactions between the environment and the brain are shown. An inability to properly adapt to the environment would have direct implications on evolution. Our model predicts serotonin dysfunction results in maladjustment of the organism to the environment because of deficits in brain morphology.

Full size image

SEROTONIN NEUROPLASTICITY: A KEY VARIABLE IN HOMEOSTASIS AND EVOLUTION

The multiple roles of serotonin throughout the life of the brain (development and aging) are as puzzling as the roles shown by serotonin throughout evolution. Attempts to provide a unified function for serotonin are elusive since the literature on this one chemical system is filled with apparent contradictions and enigmas. Serotonin, as a neurotransmitter, produces rapid postsynaptic effects at defined synapses, but as a trophic factor, it regulates the maturation of its target cells directly and by causing release of glial S100β. The complexity of function is matched by the complexity of anatomy. The raphe serotonergic neurons, which are located in the brainstem reticular formation, have extensive interactions with classical and peptidergic neurons throughout the neuroaxis, but they also have specialized contacts with endothelial, ependymal, endocrine, astrocytes, and oligodendroglial cells. These varied contacts with many different cell types gives this single chemical system an opportunity to influence all neural tissue, rather than selected neuronal circuits. Even the morphology of the serotonergic neurons itself is varied, and consistent with multiple functions. For example, the fibers, which extend from the serotonergic neurons, can be heavily myelinated to ensure rapid transmission of an electrical signal to a distant target. In the same pathway, other serotonergic fibers are unmyelinated and extensively branched, which provides a means for effective volume transmission of a trophic factor along the entire length of the fiber. Therefore, serotonergic neurons have a structural organization, which is compatible with its functions as both neurotransmitter and trophic factor, and indicates its role is more complex than can be explained by considering its effects on neuropsychopharmacology alone.

A critical question becomes what turns on this complex system? It has already been pointed out that connections between serotonergic neurons and a variety of cells exist. Most of the attention is given to the effects of neurotransmitters on the serotonergic neurons but these raphe neurons also contact blood vessels and respond not only to changes in blood pressures, but also to a variety of substances carried in the blood. These would include oxygen, glucose, amino-acids, steroids, carbohydrates, and toxins. Normally these chemicals are used to maintain the levels of serotonin but peripheral infections can drastically change serotonin levels and related functions (Freund et al. 1986). Besides these blood-related compounds, serotonin neurons are directly exposed to CSF. As mentioned above, serotonergic fibers appear to travel from brain tissue into the ventricles, indicating the same neuron integrates neuronal and CSF factors. This is an important point, because it implies serotonin's role as neurotransmitter is shaped by its non-neuronal interactions.

In this paper, I have proposed a structural and functional organization for serotonin to interact with a large number of cells within the brain by mechanisms influenced locally (e.g., synapses) or globally (e.g., vascular). However, a central issue in proposing a homeostatic role for serotonin involves feedback regulation. Serotonin responds to many neuropeptides, chemicals and steroids, and, in turn, influences the production and release of these same agents. For example, glucocorticoids increase the size of serotonergic neurons, its synthesis and release, as well as the levels of its postsynaptic receptors. In turn, serotonin, acting through a variety of receptor subtypes, regulates the release of CRF from the hypotalamus, which releases ACTH from the pituitary and ultimately cortisol from the adrenal, a feedback process involving brain, pituitary, vascular, and endocrine systems. These complex systems do not develop only in the brain or appear suddenly in mammals. These interactions evolve slowly and from the earliest known biological systems. Serotonin has important functions in plant root development and in invertebrate adaptive behaviors. Serotonergic neurons, present in the brain from the beginning of neural evolution, have established diverse and complex cellular phenotypes and neural interactions within the dynamic architecture of the brain. I propose that the serotonin system in the brain, and its plastic properties, are central to the ability of the brain to integrate with the peripheral organs of the body and with the outside environment as well (Figure 7). Changes in the functioning of serotonergic neurons, due to more effective transporter proteins or more sensitive receptors, all impact on the ability of the brain and the organism, to adapt and succeed in nature. Maladaption of the serotonin system in neural homeostasis is expected to be as harmful to mental functioning as disruption of the General Adaptive Syndrome in stress is harmful to normal health (Selye 1956). An aggressive, violent, hypersexual paranoid individual is not compatible with the social and psychiatric norms today; but the tendency to display these abnormal disorders maybe rooted in the evolutionary history of serotonin. Effective treatment of these disorders may require more consideration to trophic and plastic properties than currently shown.