Volume 20, Issue 6
Free Access

Dendritic Transmitter Release: A Comparison of Two Model Systems

F. Bergquist

Centre for Integrative Physiology, University of Edinburgh, Edinburgh, UK.

Search for more papers by this author
M. Ludwig

Centre for Integrative Physiology, University of Edinburgh, Edinburgh, UK.

Search for more papers by this author
First published: 10 July 2008
Cited by: 29
Mike Ludwig, Centre for Integrative Physiology, University of Edinburgh, Hugh Robson Building, George Square, Edinburgh EH8 9XD, UK (e‐mail: [email protected]).

Abstract

Information flow through neurones was historically considered to be linear, with dendrites receiving information from incoming synaptic terminals, the soma processing the information and the axon carrying it to the terminal that synapses upon another cell or end organ. However, recent studies have shown that dendrites can release transmitters themselves, and thereby communicate with neighbouring structures, whether these are adjacent neurones or incoming synapses. Due to their anatomical features, the magnocellular vasopressin and oxytocin containing neurones of the hypothalamic supraoptic and paraventricular nuclei and the dopamine neurones of the substantia nigra have revealed important aspects of dendritic function including mechanisms of dendritic transmitter release.

In 1945, in their review on the emerging concept of neurosecretion, Ernst and Berta Scharrer described the supraoptic nucleus (SON) as looking more like glandular than nervous tissue (1). At the time, their focus of interest was clearly the secretory appearance of neurone somata in the brain, and the fibre tract from the SON to the neurohypophysis was just described as ‘…another, very peculiar pathway through which the products of secreting nerve cells may be discharged’. When hormonal release from magnocellular hypothalamic neurones was gradually accepted as a physiological mechanism in the 1950s, this ‘peculiar pathway’ that released oxytocin and vasopressin into the blood became the focus of attention, resulting in pioneering studies on stimulus‐secretion coupling.

In the early 1960s, the notion of dendrites as active structures started to evolve (2) and has since largely replaced Ramon y Cajal’s 19th Century theory of the unidirectional neurone that only propagates signals from dendrites to soma to axon terminals. In the 1980s evidence for peptide release from the dendrites of magnocellular hypothalamic neurones started to emerge (3-6). The magnocellular neurones of the SON have since proved to be a tractable preparation for studies of dendritic neurotransmitter release, revealing important and novel aspects of neuronal function and demonstrating that dendrites can take their place as full players in both the transmitting and the receiving end of cellular communication. In this brief review, we discuss some of the advances that have been made in this field over the last 20 years. We use another model system that also contributed significantly to our understanding of dendritic neurotransmitter release, the nigrostriatal dopamine neurones, to highlight mechanisms and functions that may be general to dendritic neurotransmitter release or which reflect the large diversity of dendritic specialisations in the brain.

The hypothalamic‐neurohypophysial and the nigrostriatal cell systems as models for dendritic release

Magnocellular neurones in the hypothalamus are unusual in that they straddle the blood–brain barrier, with their dendrites arranged in dense layers under the ventricular ependyma in the central nervous system (CNS) and the axon terminals secreting their messengers directly into the peripheral circulation. Few if any of the axons find their way back into the cell bodies, so the dendritic field is virtually void of axon terminal vasopressin‐ or oxytocin‐release sites. As the blood–brain barrier effectively blocks the uptake of peripheral peptides into the brain, this gives the neuroendocrinologist two well separated compartments with which to study axon terminal and dendritic release. Similarly, the nigrostriatal dopaminergic neurones show very few retrogradely projecting axons and, in the rat, the axon terminals in the striatum and dendrites in the substantia nigra are separated by several millimetres (Fig. 1). Like hypothalamic magnocellular neurones, the nigrostriatal neurones comprise a neurochemically and anatomically well defined population, which minimises the possibility of ‘contamination’ of dendritic release by more abundant axon terminal release from neurones that are not part of the system.

The anatomical features of hypothalamic magnocellular neurones and nigrostriatal dopamine neurones make them attractive models for studying dendritic retrograde signalling. Both cell systems display distinct anatomical separation of the axon terminal and the somatodendritic regions, and utilise messengers that are only produced in restricted areas of the brain. The separation of the release sites makes it possible to study the different release mechanisms by in vivo microdialysis and blood sampling. Left insert is a fluorescence microscopy image of a supraoptic nucleus (SON) from a transgenic vasopressin‐enhanced green fluorescent protein rat. A subependymal layer of thick dendrites is indicated by arrows. The right insert shows an overview of the substantia nigra (SN) with the cell dense pars compacta (SNc). The SN pars reticulata (SNr) is perforated by dopamine containing dendrites, some of which extend several hundred micrometres as illustrated by the biocytin injected dopamine neurone on the right. Images are courtesy of Dr J. M. Tepper (SN) and Dr V. Tobin (SON). PVN, paraventricular nucleus; Pit, pituitary; ST, striatum.

Dendritic storage of neuropeptides and other neurotransmitters

Where?

The first evidence for dendritic release of neurotransmitters came from electrophysiological studies of mitral cells in the olfactory bulb combined with histological findings of dendrodendritic synaptic structures (2). The finding that dopamine was accumulated in dendrites in the substantia nigra, and depleted by reserpine treatment, established that neurotransmitters could be stored in, and perhaps released from, dendrites (7). Pow and Morris (6) were the first to visualise dendritic exocytosis from magnocellular oxytocin and vasopressin neurones when they treated hypothalamic tissue with tannic acid to ‘freeze’ the exocytosed peptide granules.

Neuropeptides are stored in large dense‐cored vesicles (LDCVs) that are found in significant numbers in somata, dendrites and axon terminals of all known peptide‐producing neurones, as well as other neurones with the capacity to co‐release neuropeptides along with other neurotransmitters such as amino acids or biogenic amines (8). However, dendritic storage of neurotransmitters does not adhere to a common theme. When classic neurotransmitters such as γ‐amino butyric acid (GABA) and glutamate are found in dendrites, they are either co‐stored with peptides in LDCVs, or stored in small synaptic vesicles (SSVs) associated with dendritic presynaptic specialisations (2, 9, 10). Some dendritically released neurotransmitters, such as nitric oxide and endocannabinoids, are synthesised on demand and not stored at all.

Dendritic storage of biogenic amines such as dopamine, noradrenalin and serotonin, occurs in dense‐cored or SSVs, but dendritic dopamine is also stored in tubulovesicular structures that resemble the smooth endoplasmic reticulum (Fig. 2) (7, 11-13). Interestingly, the vesicular monoamine transporter (VMAT2), which loads intracellular vesicles with monoamines, localises exclusively to LDCVs in pheochromocytoma PC12 cells (14). When transfected to hippocampal neurones, VMAT2 is sorted into an exocytotic pathway, which secretes both from axon terminals and soma/dendrites in fashion very similar to that of LDCVs (14).

Dendritic storage and release mechanisms in hypothalamic magnocellular neurones (a) and nigrostriatal dopamine neurones (b). (a) Neuropeptides are synthesised and packed in the soma and stored in a reserve pool (RP) containing large numbers of large dense cored vesicles. Release occurs by exocytosis, triggered by inositol‐1,4,5‐triphosphate (IP3) regulated mobilisation of intracellular calcium from the endoplasmic reticulum, or by a combination of extracellular and intracellular calcium fluxes. Intracellular calcium release can prime subsequent release by a process involving redistribution of vesicles from the RP to a readily releasable pool (RRP). (b) Dopamine dendrites contain few vesicles and dopamine is mainly stored in tubulovesicular structures resembling endoplasmic reticulum. Release is triggered by action potentials (a.p.), which trigger influx of extracellular calcium. SNARE‐dependent exocytosis is the main release mechanism, but cytoplasmic dopamine can sometimes be released by reversal of the dopamine transporter (DAT). Dopamine is synthesised in the dendrites by tyrosine hydroxylase (TH). Released dopamine is taken up and reused after being transported into storage‐ and release‐structures by the vesicular monoamine transporter VMAT2. R, membrane receptor; G, G‐protein complex; PLC, phospholipase C.

It has been suggested that the regulated LDCV exocytosis pathway and the dendritic dopamine release pathway are the same, and that the relative absence of LDCVs in dopamine neurones reflects that these neurones produce little of the chromogranins that form the dense core (14). As LDCVs are also used for storage and release of neuropeptides, one would expect significant similarities between dendritic release mechanisms of neuropeptides and monoamines but, as will be discussed, this is not necessarily the case.

How much neurotransmitter is found in dendrites?

Like other CNS neurones, magnocellular neurones store less neurotransmitter in soma and dendrites than in axon terminals. The rat pituitary contains approximately 1 μg oxytocin of which 2/9 (0.22 μg) originates from magnocellular oxytocin neurones in the SONs. The two SONs harbour 3.2 ng oxytocin in somata and dendrites (6), so approximately 1.4% of the oxytocin content of these neurones is somatodendritic. This ratio is very similar to the relation between brain and body weight in female rats (15) so, by crude measures, the dendrites of magnocellular neurones can supply as much oxytocin to the brain as the pituitary can supply to the body. Interestingly, the somatodendritic proportion of dopamine in dopamine neurones of the substantia nigra is also in this range, with axon terminals in one striatum containing some 290–300 ng of dopamine in rats, and the nigral somatodendritic region containing 2.2 ng, or approximately 1% of the total dopamine content of these neurones (16). Although the tissue levels of neurotransmitters are not reliable indicators of actual release activity, the relation between dendritic and axonal storage provides some information on the possible magnitudes of dendritic release in comparison to axon terminal release. It also highlights the difficulty of evaluating the importance of dendritic release of common neurotransmitters, such as glutamate and GABA, when the anatomical and synaptic arrangement does not allow good separation of dendritic and terminal release pathways.

The relatively small proportion of dendritically stored neurotransmitters may seem insignificant compared to the proportion that is found in axon terminals but, despite that, both hypothalamic dendritic neuropeptide release and somatodendritic dopamine release have physiologically significant effects.

Dendritic release mechanisms

The role of calcium

In axon terminals, the release of classical neurotransmitters is a process of high reliability, with every arriving action potential resulting in exocytosis (17). The available stores of SSVs are replenished by endocytotic recycling and they are quickly refilled by neurotransmitter transporters. Neuropeptides are not recycled into the releasing machinery and LDCVs have to be synthesised and packed in the cell body, so the threshold for LDCV release is higher to prevent excessive release and depletion. Compared to SSVs, LDCVs differ by requiring sustained increases in intracellular calcium to be released. As a consequence, LDCVs have longer latencies to release and require stronger stimulation to be released (e.g. bursts of electrical activity). LDCVs also differ from SSVs in that the associated calcium sensor that triggers release has a higher affinity for calcium. Consequently, it is not necessary for LDCVs to be located in close proximity to membrane calcium channels to be released, and synaptic specialisations are not a prerequisite for release. In oxytocin and vasopressin neurones, the release of LDCVs from dendrites can be regulated not only by the entry of extracellular calcium, but also by intracellular calcium mobilisation (18). The latter is usually not seen with fast SSV release from axon terminals, although constitutive release that occurs independently of action potentials may be triggered by intracellular calcium stores there as well (19).

Somatodendritic dopamine release is abolished by removing extracellular calcium (20-24), and even though intracellular calcium stores are intimately involved in the regulation of excitability of dopamine neurones (25, 26), it is not known if they play a direct role in triggering somatodendritic dopamine release. The calcium‐mediated regulation of dopamine release is not the same in dendrites and axon terminals. Dendritic dopamine release resists larger decreases in extracellular calcium than release from axon terminals (21, 23, 27). The reason for this is not entirely clear; one explanation may be differences in the release machinery, especially the calcium sensor affinity, but the more complex influences of extracellular calcium on electrical activity in dendrites and cell bodies compared to the straightforward release triggering function of calcium in terminals may be important. However, dopamine neurones stop firing in vitro when calcium currents are blocked (28), so it is unlikely that a decrease in calcium‐dependent release could be counteracted by an increase in electrical activity.

Unlike axon terminal dopamine release, release from dendrites is not blocked by N‐, or P/Q‐type calcium channel blockers but appears to be supported either by a multitude of calcium channels or possibly by R‐type channels (23, 29, 30). Taken together, the characteristics of calcium sensitivity of dendritic dopamine release suggest a weaker coupling between calcium fluxes through individual calcium channels and release than that usually seen in axon terminals. (31).

Priming of release

In dendrites of magnocellular neurones, vesicle exocytosis can be regulated not only by calcium entry, but also by intracellular calcium release. Large amounts of calcium are sequestered in the endoplasmic reticulum, and release from these stores is regulated by second‐messenger pathways. In oxytocin neurones, oxytocin itself binds to oxytocin receptors to activate this pathway. Activation of intracellular calcium release results in exocytosis by recruitment of vesicles from the reserve pool into the readily releasable pool (32). Priming is the regulated augmentation of the activity‐dependent readily releasable pool, and this can be a late and persistent consequence of activation of store‐regulated release (Fig. 2). Thus, priming involves preparing a system for some anticipated trigger that will occur at some uncertain time in the future; it involves making the secretory pool of the target cell available for rapid release in response to that future trigger (18, 33). This mechanism appears to be the key phenomenon underlying the intermittent burst discharge of oxytocin cells that they display in response to suckling during the milk‐ejection reflex (18, 34).

Priming is a mechanism, which magnocellular hypothalamic neurones share with gonadotrophs (35). At pro‐oestrous, the sensitivity of gonadotrophs to gonadothrophin releasing hormone (GnRH) increases with repeated exposure to GnRH, a process known as ‘self‐priming’. The mechanisms of self‐priming in gonadotrophs show remarkable similarities to priming in the oxytocin system. Priming cannot be induced by depolarising agents such as high potassium or calcium ionophores, and is independent of extracellular calcium concentration. Both systems require GnRH or oxytocin binding to their G‐protein coupled receptors to activate the phospholipase (PLC) pathway. PLC activation results in the production of inositol‐1,4,5‐triphosphate, which in turn triggers the release of calcium from stores of the endoplasmic reticulum.

It is not known whether priming occurs with other types of dendritic release, like monoamine release. Preliminary data indicate that dopamine release in the substantia nigra or the ventral tegmental area cannot be primed by thapsigargin, an intracellular calcium mobiliser (Fig. 3), but the influences of other intracellular calcium stores remain to be elucidated.

(a) Oxytocin release in vitro from an isolated supraoptoc nucleus (SON) evoked by depolarisation with high K+ solutions was strongly potentiated (primed) after pretreatment with the intracellular calcium mobilisers thapsigargin (TG). (b) By contrast, dopamine release in the substantia nigra (SN), measured by in vivo microdialysis, was not primed after activation of intracellular calcium stores by TG.

SNARE proteins: the exocytosis machinery

The process by which transmitter‐containing vesicles fuse with the plasma membrane and release their contents into the extracellular space is an adaptation of the constitutive exocytosis that occurs in all eukaryotic cells. Electron microscopy revealed the classical LDCV morphology in the dendrites and soma of magnocellular neurones. Tannic acid fixation of SON neurones uncovered ‘omega’ fusion profiles at the plasma membrane providing visual proof for exocytosis from SON dendrites. Furthermore, LDCV‐filled dilatations were shown at the distal parts of the dendrites (6), although no active zones appear to be required for dendritic exocytosis of vasopressin or oxytocin.

The somatodendritic release mechanism of dopamine has been the subject of a longstanding debate. Dopaminergic dendrites contain few synaptic vesicles so the origin of release was questioned early on. Pharmacological studies indicated that somatodendritic dopamine could be released by reversal of the dopamine transporter, DAT (36), and it was later shown that release can be mediated by the non‐exocytotic DAT pathway also in response to stimulation of glutamatergic/cholinergic afferents (37). However, in most studies, pharmacological block of DAT leads to increased, not decreased, dendritic dopamine release (38, 39). Furthermore, a major part of baseline and potassium evoked dendritic dopamine release in vivo can be abolished by clostridial toxins that cleave SNARE proteins (40), and this does not result from loss of excitatory afferents because similar effects are found in isolated dopamine neurones (27). The pronounced calcium sensitivity and tetrodotoxin‐sensitivity (20, 24, 36) also support a mainly exocytotic release mechanism, and quantal dopamine release events have been recorded from dopamine cell bodies (41). As with dendritic oxytocin and vasopressin release, active zones are not required for dendritic dopamine release.

One explanation to the different release profiles of axon terminal and dendritic secretion could be that different sets of exocytosis proteins mediate the release. The core complex SNARE proteins SNAP‐25 and syntaxin have been identified in dendrites from neonatal hypothalamic neurones (42). α‐SNAP mRNA was identified in dendrites from hypothalamic neurones (43), but dendritic isoforms of the various core‐complex proteins or of other exocytotic proteins, such as the postulated synaptotagmin (42), are essentially unknown. The terminals of vasopressin and oxytocin neurones in the posterior pituitary contain many SNARE proteins, some associated with neurosecretory granules, others with microvesicles (44, 45). We are currently determining the exact expression and localisation of SNARE proteins in dendrites of magnocellular neurones (46).

Stimulus‐evoked fusion, sensitive to neurotoxins, was described in dendrites from the hippocampus (47) and isolated magnocellular neurones (48), suggesting that proteins similar to those operating in axon terminals regulate dendritic exocytosis. Both dendritic release of dopamine in the substantia nigra (27, 40) and terminal release in the striatum (40) is inhibited by Botulinum toxin A (which cleaves SNAP‐25), and although Botulinum toxin B (which cleaves VAMP) does not inhibit nigral dopamine release in vivo (40), it effectively blocks it in cultured dopamine cells (27). Although the notion of different SNARE‐pathways in dendritic and terminal release is attractive, the evidence for or against it is currently insufficient.

Is dendritic release triggered by back‐propagating action potentials or by local independent processes?

Action potentials are usually initiated at the junction between cell body and axon, the axon hillock. In the classical view, an action potential then travels along the axon to the axon terminal where it triggers neurotransmitter release. What is less emphasised in most textbooks is that the action potential travels in all directions from its point of initiation. If the electrical properties of the dendrites can support an action potential, it can therefore spread into dendrites too, as is the case in many neurones (49). However, the spread of backpropagating action potentials is not a robust phenomenon.

In dopamine neurones, action potentials are generated in the dendritic tree (because dopaminergic axons branch off from a large dendrite rather than the cell body), and single action potentials backpropagate into dendrites (50). However, when the neurones fire in bursts, backpropagating action potentials often fail, and more so if dopamine D2 receptors are activated (51), suggesting that dopamine can self‐regulate its dendritic release by influencing the spread of action potentials. The failure of dendritic backpropagation of burst activity indicates that burst patterned firing does not promote dendritic dopamine release as it promotes axon terminal release (52). In that sense, dendritic dopamine release is at least semi‐independent from the electrical activity of the neurone. Serotonin neurones in the dorsal raphe nucleus also display dendritic release. In these neurones, action potentials are not backpropagated far, but local activation of N‐methyl‐d‐aspartate (NMDA)‐receptors can lead to dendritic release without action potentials occurring (53), illustrating that dendritic release can be completely independent of axon terminal release in some monaminergic neurones.

After dendritic priming, antidromic activation of magnocellular neurones leads to dendritic neuropeptide release (18) but, in most circumstances, antidromic activation of magnocellular neurones does not evoke release. It appears that action potentials cannot propagate backwards into the dendrites of these neurones, or that dendritic neuropeptide release is not triggered by back‐propagating action potentials.

So what is the physiological trigger of dendritic neuropeptide release? As in serotonin neurones, glutamate can trigger dendritic vesicle fusion in magnocellular neurones via NMDA‐receptor activation and without concurrent action potentials (54). In agreement with this, there is sometimes a clear dissociation between dendritic neuropeptide release and release from the terminals of the same neurones. A dissociation between dendritic and axon terminal oxytocin release is evident from the effects of α‐melanocyte‐stimulating hormone. Activation of MC4 receptors on oxytocin cells mobilises intracellular calcium and evokes dendritic oxytocin release, but the electrical activity of the cell is inhibited, leading to less oxytocin release into the periphery (55). Another example of dissociated release patterns is when vasopressin is released to counteract water loss from the kidneys in response to increased plasma osmolality. The axon terminal release of vasopressin after a systemic hypertonic saline injection increases immediately, but intra‐SON release of vasopressin only starts 1 h later, when peripheral release is subsiding. (56), illustrating a separation in time between the release events in dendrites and the terminals of the same neurones. Systemic osmotic stimuli activate both peripheral and central receptors on cells that project, directly or indirectly, to the SON and paraventricular nucleus (PVN), including the region anterior and ventral to the third ventricle (AV3V) (57). The ability of tetrodotoxin administration into the SON, or lesion of the AV3V, to abolish the dendritic peptide response to systemic osmotic stimulation suggests that dendritic release results from a cascade of events initiated by osmotic activation of synaptic pathways (58, 59).

Physiological functions of dendritically released neuropeptides and other neurotransmitters

Autocrine effects

The physiological functions of dendritically released neurotransmitters come in two flavours. One is the local autocrine effects that they have on the neurones from which they are released, and the other concerns the effects that they have on surrounding neurones and glia. Because the autocrine effects can ultimately change the axon‐terminal output from the neurones, dendritic release tends to have more far‐reaching effects than its relative proportion of neural transmission would indicate. A striking example of this is the way dendritically released oxytocin promotes the milk ejection reflex. In nonlactating females, as well as in males, oxytocin neurones are continuously active. During lactation, however, the firing pattern of oxytocin neurones shifts dramatically, and they display intense bursts of activity with approximately 5‐min intervals. The bursts are synchronised, also between the two SONs, and lead to pronounced but short‐lasting release of oxytocin from the pituitary. The peripheral oxytocin release promotes milk ejection from the mammary glands and the long intervals between the bursts allow the glands to replenish their milk content. This intermittent burst firing pattern of oxytocin neurones relies on the activation of oxytocin receptors in the SON and PVN, as it is inhibited by local application of oxytocin antagonist, and rescued by oxytocin given i.c.v. (60). The activation of SON and PVN oxytocin receptors are the result of locally released oxytocin because suckling is associated with increased oxytocin concentrations in the SON, but not near the SON (61). In lactating rats, the dendritic release of oxytocin can be evoked by antidromic stimulation of the pituitary stalk (61) but, in nonlactating rats, dendritic release does not reflect back‐propagating action potentials (18). If, however, the dendrites are primed by a surge in local oxytocin concentrations or by pharmacological mobilisation of intracellular calcium stores, the oxytocin dendrites will make a transition from action potential‐independent release to release that is triggered by back‐propagating action potentials (18). So it appears that priming of dendritic release could be the mechanism by which magnocellular neurones are rewired to deliver the pulses of oxytocin release needed for parturition and lactation. The priming mechanisms of SON dendrites have been studied in some detail and involve a redistribution of dense cored vesicles to locations closer to the cell membrane (32), as well as changes in actin remodelling (62).

A far more common autocrine effect of dendritic release is auto‐inhibition. Dopamine released in substantia nigra (37, 63), as well as noradrenalin in locus coeruleus (64), serotonin in nucleus raphe dorsalis (65), and histamine in tuberomamillary nuclei (66), inhibit the electrical activity of the neurones that they are released from by acting on autoreceptors. Vasopressin neurones discharge in a characteristic phasic pattern that optimises the efficiency of stimulus‐secretion coupling at the nerve terminals. Vasopressin released from dendrites modulates this phasic activity by a predominantly inhibitory action (67, 68). Interestingly, vasopressin also (like oxytocin) facilitates its own dendritic release (69). This may explain the time dissociation between peripheral and intra‐SON release of vasopressin after a hyperosmotic stimulus (56). Although systemic release of vasopressin occurs rapidly after an osmotic stimulus, the dendritic release of vasopressin evolves as a delayed and prolonged response. Mimicking dendritic release by retrodialysis of vasopressin onto vasopressin neurones inhibits vasopressin neurones by reducing their firing rate, sometimes silencing the electrical activity of vasopressin neurones by up to 1 h (67). Thus, dendritic vasopressin release may activate adjacent dendrites to facilitate its own release until the local concentration has reached a threshold sufficient to hyperpolarise the neurone and/or modulate inhibitory inputs. The auto‐inhibitory action of dendritic vasopressin may therefore serve to limit the extent of systemic vasopressin release in response to osmotic stimuli or volume depletion.

Neuromodulatory or paracrine effects of dendritic release

Dendritically released dopamine acts at both pre‐ and post‐synaptic receptors to modulate GABAergic and glutamatergic transmission in the substantia nigra pars reticulata. Activation of presynaptic dopamine D1 receptors in the substantia nigra reticulate increases GABA and dynorphin release from striato‐nigral neurones (70-73). Dopamine D1 receptors can also stimulate glutamate release from subthalamic afferents, but D1‐mediated inhibition is the predominant dopaminergic influence on nigral glutamate release in vivo (74, 75). Dopamine was also reported to excite GABAergic nigrothalamic neurones in the substantia nigra (76) but, in unrestrained rats in vivo, the overall effect of dendritic dopamine release is inhibition of these neurones, probably indirectly by promoting GABA‐release from striato‐nigral afferents (77). Dendritically released dopamine thereby disinhibits thalamocortical circuits, which in turn promote the initiation of movements. Accordingly, it was found that nigral dopamine release has a permissive role for motor performance, without altering axon terminal dopamine release in the striatum (78, 79). Local depletion of dendritic dopamine disrupts motor performance more effectively than local depletion of striatal dopamine (78). Dopamine is also released from dendrites in the ventral tegmental area, where it is necessary for the development of behavioural drug sensitisation (80-83), a phenomenon that is related to addiction.

Exogenously applied or endogenously released oxytocin also acts on afferent nerve endings (48, 84). As presynaptic oxytocin receptors could not be demonstrated in the SON (85), this paracrine action was likely to be indirect and was later shown to be mediated by oxytocin‐dependent endocannabinoid release (86, 87). Endocannabinoid signalling has a very short action radius (88), but the oxytocin signal spreads over larger areas, effectively broadcasting the message for presynaptic inhibition through the nucleus (87).

This brings us to the question of the action radius of dendritic release. After quantal release, dilution by diffusion has a major impact on the concentration at short ranges, and high synaptic concentrations after a release event may only be found within the nearest micrometre (89). At longer distances, uptake and breakdown processes are more important. In substantia nigra where the DATs are sparse, dopamine transients can travel up to 20 μm without being extinguished by uptake, although the dilution effect remains. The average extracellular dopamine concentration in substantia nigra is approximately 2.6 nm (40) which is enough to activate high affinity dopamine receptors. Although dendritically released dopamine has a longer action radius, its action on low‐affinity (D1) receptors is still mainly determined by diffusion dilution and therefore limited to some 10 μm (89).

Oxytocin‐ and vasopressin‐induced effects on behaviours are exerted at sites that, in some cases, richly express receptors but are innervated by few peptide‐containing projections. Could dendritically‐released peptides act at distant brain targets to evoke long‐lasting behavioural effects? Although extracellular neuropeptide concentrations differ from site to site, similar changes are often seen at widely separated sites. For example, after ‘forced swimming’, concentrations of vasopressin (but not oxytocin) are raised in both the SON and the mediolateral septum, whereas ‘social defeat’ results in raised concentrations of oxytocin (but not vasopressin) in both the SON and the lateral septum (90). In addition, direct osmotic stimulation of the SON results in a rise in vasopressin concentrations not only in the SON, but also in the septum, a finding that is difficult to explain except by diffusion from the SON (91).

Peptide release within the brain is not specifically targeted at synapses and, as the half‐lives of peptides in the CNS can be up to 20 min, there is time for considerable transport by diffusion and bulk flow transport in extracellular fluid and cerebrospinal fluid (CSF). As previously mentioned, the dendrites of magnocellular neurones project towards the brain surface and make close with ependymal cells that line the ventricular spaces. The reason for this may be two‐fold. The dendrites can register the neurochemical composition of CSF and they can send their messages into the CSF circulation. Neuropeptides administered i.c.v. lead to coherent and purposeful behaviours (34). Perhaps Scharrers’ suggestion that the SON looked like glandular tissue is close to the truth. Magnocellular peptides could reach distant targets in a hormone‐like fashion and, once there, can functionally reorganise neuronal networks (prime), providing a substrate for prolonged behavioural effects (34).

Concluding remarks

Over the last 20 years, hypothalamic magnocellular neurones and nigrostriatal dopamine neurones have become the best studied models of dendritic release. When compared head‐to‐head, these cell systems uncover similarities in some of the physiological mechanisms by which dendritic release regulates the activity of the releasing cells and local neurotransmission. However, when the cellular mechanisms for storage, release and regulation of release are compared, they reveal far‐reaching specialisations that may reflect the striking differences in dendritic morphology, as was already obvious from Ramon y Cajal’s groundbreaking studies. Understanding how dendritic release influences normal function and pathology in the nervous system will continue to be an exciting challenge for neuroscience.

Acknowledgements

We would like to thank Gareth Leng for critical comments on the manuscript and J. M. Tepper and Vicky Tobin for providing histological images. The work was supported by grants from the Swedish Research Council (F.B.) and BBSRC (M.L.).

      The full text of this article hosted at iucr.org is unavailable due to technical difficulties.