Elsevier

Progress in Neurobiology

Volume 68, Issue 4, November 2002, Pages 247-286
Progress in Neurobiology

Retrograde signaling in the regulation of synaptic transmission: focus on endocannabinoids

https://doi.org/10.1016/S0301-0082(02)00080-1Get rights and content

Abstract

This review covers recent developments in the cellular neurophysiology of retrograde signaling in the mammalian central nervous system. Normally at a chemical synapse a neurotransmitter is released from the presynaptic element and diffuses to the postsynaptic element, where it binds to and activates receptors. In retrograde signaling a diffusible messenger is liberated from the postsynaptic element, and travels “backwards” across the synaptic cleft, where it activates receptors on the presynaptic cell. Receptors for retrograde messengers are usually located on or near the presynaptic nerve terminals, and their activation causes an alteration in synaptic transmitter release. Although often considered in the context of long-term synaptic plasticity, retrograde messengers have numerous roles on the short-term regulation of synaptic transmission. The focus of this review will be on a group of molecules from different chemical classes that appear to act as retrograde messengers. The evidence supporting their candidacy as retrograde messengers is considered and evaluated. Endocannabinoids have recently emerged as one of the most thoroughly investigated, and widely accepted, classes of retrograde messenger in the brain. The study of the endocannabinoids can therefore serve as a model for the investigation of other putative messengers, and most attention is devoted to a discussion of systems that use these new messenger molecules.

Introduction

For most of the past century communication between nerve cells was considered to be polarized, i.e. that it took place at synapses between a sending cell that was presynaptic and a receiving cell that was postsynaptic. The impetus for the concept came from Ramon y Cajal in establishing the primacy of individual, morphologically delimited, neurons as the basic units of the nervous system (the “neuron doctrine”) (Ramon y Cajal, 1995). The contrary idea, prominently espoused by Golgi, was that the nervous system was a syncitium: a morphologically continuous reticulum of cells connected by very fine processes. Cajal deduced that a particular part of a cell was specialized to receive signals (the dendrites) and another part was specialized to send them (the axon); i.e. that cells were “dynamically polarized”. In the olfactory system, for example, it was clear that sensory inputs arising from the environment, and being carried by the olfactory nerve fibers, would first encounter the dendrites of the cells in the bulb, and would then get from the bulb to the brain via the axons. The arrangements of dendrites and axons throughout the nervous system permitted the generalization of the principle of dynamical polarization to cells in which inputs and outputs were not as readily distinguishable. Dynamical polarization of cells was important not only for its physiological relevance, but because it favored the neuron doctrine: it was not clear how the syncitial arrangement would allow for dynamical polarization—in that case signals, it seemed, should be able to go backwards and forwards equally well.

The widespread acceptance of the neuron doctrine and its association with the dynamical polarization of individual cells may have perpetuated a bias against the idea that signal transmission could go “backwards” across the synapse. There is no necessary link between the concepts, however, and indeed, retrograde signaling between anatomically distinct cells was fairly readily accepted as a necessity in the course of neuronal synaptic development and the establishment of neuronal circuits (Fitzsimonds and Poo, 1998).

Acceptance that retrograde signaling was important in moment-to-moment intercellular signaling came more slowly. Detailed information about the apparatus of synaptic transmission at nerve terminals led to expectations about the morphological and physiological underpinnings of chemical transmission in general: that transmitter would be pre-packaged in vesicles, that an active zone for vesicle release would be present; that release would be triggered by action potentials, etc. As most synaptic inputs in the CNS are made onto dendrites or somata, the lack of such morphological specializations in these postsynaptic elements, and the view of dendrites as passive conductors of electrical signals, weighed against the possibility of retrograde signaling. Finally, it was difficult to make the high-resolution measurements of synaptic transmission in the brain necessary to detect retrograde signaling.

Accumulating evidence has gradually removed these objections. Membrane-permeable, diffusible messengers do not require vesicles for storage or release, and indeed can be released from morphologically undifferentiated regions of the cell. Dendrites conduct action potentials and are electrically and functionally very complex (Llinas, 1975, Wong et al., 1979, Yuste and Tank, 1996, Johnston et al., 1999, Magee, 1999, Spruston et al., 1999). Neurophysiological evidence confirms that events in the postsynaptic cell can influence transmitter release from presynaptic terminals. Accepted procedures now exist for detecting the presence of a retrograde signal process.

This review is intended to provide a survey of recent neurophysiological developments in the field of retrograde signaling in the regulation of neurotransmitter release, mainly in the mammalian brain. Despite the possibility that retrograde signals can travel from the post- to the presynaptic cell via direct physical contact (Fitzsimonds and Poo, 1998), at present there is little evidence for this at mature synapses. Electronic coupling and gap junctions exist in the mature brain, but electrical signaling is no more retrograde than anterograde. I will not attempt to cover the trophic, or developmental, roles of retrograde signals, and will deal with issues of retrograde signaling in long-term plasticities only to discuss the specific molecules that have been proposed as messengers and to touch on a few recent findings. Although the study of long-term plasticities continues to provide fascinating new data and insights, the literature has been the subject of numerous authoritative reviews (e.g. Bear and Abraham, 1996, Frey and Morris, 1998, Daniel et al., 1998, Malenka and Nicoll, 1999, Bortolotto et al., 1999, Luscher et al., 2000, Hansel et al., 2001). Furthermore, focus on the long-term plasticities has perhaps fostered the notion that retrograde signaling is relevant only in this context, whereas the bias of this review is that retrograde signaling is more prevalent, more varied and has more profound implications for understanding the nervous system than are encapsulated by its roles in long-term plasticity.

The gases, nitric oxide and carbon monoxide, as well as the lipids, arachidonic acid and platelet-activating factor, were the first proposed retrograde signals in hippocampal LTP. Although their candidacy generated a great deal of interest and experimental work, it also generated controversy and contradictory results. Attempts at reconciliation have been made; experimental variables and multiple forms of LTP seem to be the most likely complicating factors, but even this is not yet certain. At this point it appears that consensus has not been reached on whether these agents are involved in LTP or LTD, and relatively little seems to be known about their effects on short-term regulation of synaptic transmission. It is also unclear what their final effectors are, and whether they are pre- or postsynaptically located. The liability and ubiquity of the gases, and the absence of well-defined surface receptors, together with the innate complexity of the phenomena they have been proposed to mediate, mitigated against simple answers to some basic questions that can be asked of a retrograde system. I will touch on only a few recent results, and published reviews should be consulted for discussion of previous research (Zhuo et al., 1994, Snyder, 1994, Holscher, 1997, Daniel et al., 1998, Son et al., 1998, Haley, 1998, Baranano et al., 2001, Grassi and Pettorossi, 2001).

The concept of “non-synaptic” transmission has been in use for many years (Vizi and Kiss, 1998, Zoli et al., 1999). “Non-synaptic” calls attention to the vast array of intercellular interactions that occur apart from conventional synapses, and is intended to be synonymous with “volume”, “paracrine” or “diffusion” transmission. Non-synaptic interactions represent “long distance communication” (>1 μm) between neurons, the range being determined by the distance traveled by the released neurotransmitter to its target receptors. The question arises whether “retrograde” is just another name for “non-synaptic”. Distinctions should be made, however. Non-synaptic is a very general term, and has been applied to increases in extracellular potassium, or ephaptic interactions in which the electrical fields set up by activity in one group of cells influence other cells. This usage of “non-synaptic” seems very appropriate, because synapses are not involved in any way, either as sources or targets of the communication. In some instances, what is called “retrograde signaling” would be better characterized as a type of paracrine transmission, as for example, when the release of nitric oxide affects numerous other cells in the environment without regard to their synaptic relationship to the releasing cell.

Other interactions, not involving traditional, closely apposed synaptic specializations, nevertheless can involve the modulation of traditional synapses or the release of transmitters from specialized regions to either pre- or postsynaptic targets. These often represent forms of synaptic interactions and include: volume conduction of neurotransmitters such as the monoamines and acetylcholine from presynaptic varicosities to distant targets, neurotransmitter “spillover” from one presynaptic nerve terminal to heteroreceptors on nearby nerve terminals, and autoreceptor feedback whereby transmitter released from a nerve terminal activates receptors on or near that same terminal (Kullmann and Asztely, 1998, Vizi and Kiss, 1998, MacDermott et al., 1999). As the transmitters are liberated by conventional mechanisms from specialized presynaptic regions, and their actions cause modifications of synaptic transmission, the designation of these phenomena as “non-synaptic” is potentially confusing, e.g. should “presynaptic inhibition” mediated by volume conduction of transmitter from a distant terminal be considered a synaptic or a non-synaptic mechanism?

The present review will focus on short-term (milliseconds to seconds) activity-dependent retrograde signaling in the mammalian brain, and particularly on presynaptic regulation of the strength of synaptic transmission (Fig. 1). The signaling involves polarized communication between cells (from post- to presynaptic cells), a specialized diffusible messenger molecule, and a receptor for the messenger. The messenger is released from regions of the cell that lack morphologically defined active zones, generally somata or dendrites. Retrograde signaling in this review does not apply to processes within a given cell, as for example, the signaling that travels up the axon from TrkB receptors on nerve terminals to cell bodies, although the name has been applied to these processes (Miller and Kaplan, 2001). Retrograde signaling is temporally delimited, but both release and reception of the signal occur over spatially diffuse regions of cells. The receptors can either be membrane-bound, or intracellular. The ultimate effect of the retrograde signal process is alteration of presynaptic neurotransmitter release by an action at or near the nerve terminal.

Even with these restrictions, it is clear that “retrograde signals” form a large class that comprises numerous kinds of molecules. This field is still in its earliest stages, and yet it is already very wide: virtually any question that can be posed regarding conventional, anterograde signaling, can be posed in the case of retrograde signaling. The more that is learned about retrograde signals, the more pervasive and remarkable they seem to be. It is too soon to tell whether the new insights will truly force a “paradigm shift” (Kuhn, 1970) in thinking about the organization and function of intercellular communication in the nervous system, but there can be no doubt that a thorough understanding of retrograde signaling will alter current concepts of many neural phenomena. With the aim of conveying some of the current excitement, the review will therefore concentrate on candidate diffusible messengers, and their various modes of action, release and function. Most attention will be given to the endocannabinoid system, as it is currently the most thoroughly studied retrograde signal system in the brain. The study of endocannabinoids is undergoing an explosion of interest and new developments, and can serve as a model for investigation of these systems.

In principle it would seem that anything that could be accomplished through retrograde signaling could also be accomplished by conventional synapses and neuronal circuitry. Why then does retrograde signaling exist, and what are its advantages over conventional signaling? A definitive answer is impossible, but potential factors include:

  • (a)

    Precision: Juxtaposition of the site of release of a retrograde messenger and the presynaptic terminal means that precise regulation of transmitter release from nerve terminals can occur.

  • (b)

    Speed: Retrograde interactions can be rapid because the signal can be locally mobilized and released. Somatic synaptic integration, and action potential generation and conduction, which are required for most synaptic communication, are not required to liberate the retrograde messenger, and the distance between sending and receiving elements can be as small as the width of the extracellular space.

  • (c)

    Efficiency: It is more efficient to regulate transmission at a single synapse by a directly influencing the properties of the presynaptic nerve terminal than by changing the integrative properties of the cell’s somatodendritic regions to alter its state of excitability. Minimal amounts of signal substance and receptors are required because only the properties of the terminals must be affected in order to alter transmitter release.

  • (d)

    Flexibility: If the retrograde signal is broadcast, i.e. not released by specialized active zones, then the presynaptic nerve terminals that are the targets of the signal do not have to remain fixed. For example, morphological plasticity may accompany long-term changes in synaptic influences; i.e. synapses may appear or disappear in response to changing patterns of use. If it is liberated diffusely, a retrograde signal will instantly provide, maintain, or re-establish signal interactions as soon as it is encountered by incoming input fibers carrying the appropriate receptors. In conventional signaling, specialized contacts must be formed.

  • (e)

    Specificity: A single cell can influence its own synaptic inputs without affecting those of even closely neighboring cells. For example, single GABAergic basket cells in the hippocampus can make a total of over 10,000 synapses each, and thereby innervate upwards of 1000 target cells (Freund and Buzsaki, 1996). Retrograde signaling provides a way of sculpting the output of such cells to provide locally appropriate input. Individual target cells can adjust the strengths of their inhibitory inputs, and behave differently from neighboring cells that may be receiving synapses from the same interneuron. Local sculpting of this kind would be difficult to achieve by neuronal circuits. An interesting feature of retrograde signaling in this context is that it degrades the information content of presynaptic somatic action potentials. Presynaptic activity detected by, e.g. extracellular unit recordings, is not necessarily conveyed to all of a cell’s targets, and analyses of circuit behavior based on the assumption that it is may be in error.

Functional classes of signaling by diffusible chemical mediators can be distinguished by several properties, e.g.: (1) the specificity of the signal itself; (2) whether or not its receptors are specific or also serve more general roles; (3) the nature of its release, whether this is temporally and spatially regulated and restricted, or whether it is a broadcast phenomenon; (4) its sphere of influence, whether it acts locally, or over a wide area. Practical tests (outlined in Table 1), analogous to those used for identifying a transmitter candidate, are available for identifying a retrograde signal process and the messenger involved. Few molecules have been subjected to the whole battery. Table 2 lists features of putative retrograde messengers with representative actions in selected brain areas. The list is intended to provide some points of comparison, and is not comprehensive.

Section snippets

Endocannabinoids as model retrograde signals

The active ingredient in marijuana is Δ9-THC (Martin et al., 1999, Ameri, 1999, Freund et al., 2002), and it produces central nervous system effects principally by binding to a specific, membrane-bound, G-protein coupled receptor (Herkenham et al., 1990, Westlake et al., 1994). Two forms of cannabinoid receptor, CB1R (with a splice variant, CB1aR) and CB2R, have been cloned (Matsuda et al., 1990, Pertwee, 1997, Onaivi et al., 2002). CB1R alone is found in brain, whereas CB2R is located

Dopamine

Dopamine is the major neurotransmitter in the substantia nigra and was the first transmitter found to be released from extra-synaptic sites, as it is in the pars reticulata from the dendrites of the nigro-striatal neurons (Geffen et al., 1976, Cheramy et al., 1981). As there are dendrites, but virtually no dopaminergic axon terminals in the pars reticulata, the dopamine collected there following stimulation could be attributed to dendritic release. Once released, dopamine can activate D2

Oxytocin and vasopressin

Magnocellular neurons of the supraoptic nucleus (SON) contain the neuropeptides oxytocin and vasopressin, and release them in response to physiological stimulation (Ludwig, 1998, Raggenbass, 2001). The peptides are present in vesicles in the somatodendritic regions, as well as the nerve terminals, of the cells. The G-protein-coupled peptide receptors are found on the magnocellular cell bodies, and on the terminals of glutamatergic axons present in the SON. Activation of the cells by antidromic

Nitric oxide (NO)

Nitric oxide is a membrane-permeant gas that is produced by the conversion of citrulline to arginine by the action of the enzyme, NO synthase (Snyder et al., 1998). The actions of NO are ubiquitous in peripheral nervous system where it mediates vasodilation and other effects. NO has been put forward as a retrograde signal in the establishment of LTP in the central nervous system, initially the hippocampus (O’Dell et al., 1991, Schuman and Madison, 1991, Zhuo et al., 1993), but this topic has

Non-cannabinergic lipids

The idea that certain lipids might be retrograde signals is appealing for at least two main reasons. (1) Lipid messengers can be synthesized quickly from membrane components; in effect they are available “on-demand”. The need for energetically expensive synthesis, transport, and storage systems is avoided. (2) Lipids are membrane-permeant, hence specialized release mechanisms are not needed, and their targets are not restricted to regions apposed to release sites. They can immediately diffuse

Conclusions and implications

Retrograde signaling is now established as a mechanism of synaptic regulation in the brain. Strong evidence supports the identification of several classes of chemicals as retrograde messengers, although crucial pieces of evidence are lacking in many cases, and more detailed information is needed in all. Common unresolved issues concern the actual mechanisms by which the messengers are liberated from the postsynaptic cell, and the physiological response patterns that are required to initiate

Acknowledgements

I thank Masako Isokawa, Scott Thompson, Darrin Brager, Jimok Kim and David Edwards for reading and commenting on the manuscript. I especially thank Evelyn Elizabeth, who performed outstanding editorial and clerical assistance for this and all of the related projects reported herein. Thanks also to Masako Isokawa, who prepared Table 2, and Tamas Freund, Istvan Katona and Daniele Piomelli, as well as Stephen Davies, R.G. Pertwee and G. Riedel, who shared their respective reviews of

References (268)

  • W.R Chen et al.

    Analysis of relations between NMDA receptors and GABA release at olfactory bulb reciprocal synapses

    Neuron

    (2000)
  • A.J Cochilla et al.

    Metabotropic glutamate receptor-mediated control of neurotransmitter release

    Neuron

    (1998)
  • H Daniel et al.

    Cellular mechanisms of cerebellar LTD

    Trends Neurosci.

    (1998)
  • M.A Desai et al.

    Metabotropic glutamate receptor-induced disinhibition is mediated by reduced transmission at excitatory synapses onto interneurons and inhibitory synapses onto pyramidal cells

    Neurosci. Lett.

    (1994)
  • V Di Marzo et al.

    Endocannabinoids: endogenous cannabinoid receptor ligands with neuromodulatory action

    Trends Neurosci.

    (1998)
  • J.P Fawcett et al.

    Detection of brain-derived neurotrophic factor in a vesicular fraction of brain synaptosomes

    J. Biol. Chem.

    (1997)
  • U Frey et al.

    Synaptic tagging: implications for late maintenance of hippocampal long-term potentiation

    Trends Neurosci.

    (1998)
  • S Grassi et al.

    Synaptic plasticity in the medial vestibular nuclei: role of glutamate receptors and retrograde messengers in rat brainstem slices

    Prog. Neurobiol.

    (2001)
  • R.W Greene et al.

    The electrophysiology of adenosine in the mammalian central nervous system

    Prog. Neurobiol.

    (1991)
  • P Gubellini et al.

    Selective involvement of mGlu1 receptors in corticostriatal LTD

    Neuropharmacology

    (2001)
  • N Hajos et al.

    Novel cannabinoid-sensitive receptor mediates inhibition of glutamatergic synaptic transmission in the hippocampus

    Neuroscience

    (2001)
  • F Helmchen et al.

    Ca2+ buffering and action potential-evoked Ca2+ signaling in dendrites of pyramidal neurons

    Biophys. J.

    (1996)
  • D.J Henry et al.

    Activation of inwardly rectifying potassium channels (GIRK1) by co-expressed rat brain cannabinoid receptors in Xenopus oocytes

    Neurosci. Lett.

    (1995)
  • C Holscher

    Nitric oxide, the enigmatic neuronal messenger: its role in synaptic plasticity

    Trends Neurosci.

    (1997)
  • A.C Howlett et al.

    The cannabinoid receptor: biochemical, anatomical and behavioral characterization

    Trends Neurosci.

    (1990)
  • A.J Irving et al.

    Functional expression of cell surface cannabinoid CB1 receptors on presynaptic inhibitory terminals in cultured rat hippocampal neurons

    Neuroscience

    (2000)
  • J.S Isaacson et al.

    Local and diffuse synaptic actions of GABA in the hippocampus

    Neuron

    (1993)
  • T Ishikawa et al.

    A single packet of transmitter does not saturate postsynaptic glutamate receptors

    Neuron

    (2002)
  • D Johnston et al.

    Regulation of back-propagating action potentials in hippocampal neurons

    Curr. Opin. Neurobiol.

    (1999)
  • B.E Alger et al.

    Retrograde signalling in depolarization-induced suppression of inhibition in rat hippocampal CA1 cells

    J. Physiol. (London)

    (1996)
  • A Al-Hayani et al.

    Cannabinoid receptor-mediated inhibition of excitatory synaptic transmission in the rat hippocampal slice is developmentally regulated

    Br. J. Pharmacol.

    (2000)
  • A Ameri et al.

    Effects of 2-arachidonylglycerol, an endogenous cannabinoid, on neuronal activity in rat hippocampal slices

    Naunyn-Schmiedeberg’s Arch. Pharmacol.

    (2000)
  • N Auclair et al.

    Cannabinoids modulate synaptic strength and plasticity at glutamatergic synapses of rat prefrontal cortex pyramidal neurons

    J. Neurophysiol.

    (2000)
  • D.E Baranano et al.

    Neural roles for heme oxygenase: contrasts to nitric oxide synthase

    Proc. Natl. Acad. Sci. U.S.A.

    (2001)
  • J.L Barker et al.

    Noradrenaline and acetylcholine responses of supraoptic neurosensory cells

    J. Physiol.

    (1971)
  • N.G Bazan

    Arachidonic acid in the modulation of excitable membrane function and at the onset of brain damage

    Ann. N. Y. Acad. Sci.

    (1989)
  • M.F Bear et al.

    Long-term depression in hippocampus

    Ann. Rev. Neurosci.

    (1996)
  • G.A Bohme et al.

    Enhanced long-term potentiation in mice lacking cannabinoid CB1 receptors

    Neuroscience

    (2000)
  • J.M Brundege et al.

    Modulation of excitatory synaptic transmission by adenosine released from single hippocampal pyramidal neurons

    J. Neurosci.

    (1996)
  • G Carlson et al.

    Endocannabinoids facilitate the induction of LTP in the hippocampus

    Nat. Neurosci.

    (2002)
  • M.P Caulfield et al.

    Cannabinoid receptor agonists inhibit Ca current in NG108-15 neuroblastoma cells via a Pertussis toxin-sensitive mechanism

    Br. J. Pharmacol.

    (1992)
  • M.P Caulfield et al.

    Neurotransmitters inhibit the omega-conotoxin-sensitive component of Ca current in neuroblastoma × glioma hybrid (NG108-15) cells, not the nifedipine-sensitive component

    Pflugers Arch.

    (1992)
  • B.T Chen et al.

    Novel Ca2+ dependence and time course of somatodendritic dopamine release: substantia nigra versus striatum

    J. Neurosci.

    (2001)
  • A Cheramy et al.

    Dendritic release of dopamine in the substantia nigra

    Nature

    (1981)
  • S Choi et al.

    Decreased frequency but not amplitude of quantal synaptic responses associated with expression of corticostriatal long-term depression

    J. Neurosci.

    (1997)
  • W.S Cobb et al.

    Distinct roles for nigral GABA and glutamate receptors in the regulation of dendritic dopamine release under normal conditions and in response to systemic haloperidol

    J. Neurosci.

    (2002)
  • W.A Coetzee et al.

    Molecular diversity of K+ channels

    Ann. N. Y. Acad. Sci.

    (1999)
  • P.J Conn et al.

    Pharmacology and functions of metabotropic glutamate receptors

    Annu. Rev. Pharmacol. Toxicol.

    (1997)
  • F Crepel et al.

    Effects of ACPD and AP3 on parallel-fibre-mediated EPSPs of Purkinje cells in cerebellar slices in vitro

    Exp. Brain Res.

    (1991)
  • S.A Deadwyler et al.

    Cannabinoids modulate potassium currents in cultured hippocampal neurons

    Receptors Channels

    (1993)
  • Cited by (513)

    View all citing articles on Scopus
    View full text