Elsevier

Neuropharmacology

Volume 76, Part C, January 2014, Pages 610-627
Neuropharmacology

Invited review
Pre- and postsynaptic twists in BDNF secretion and action in synaptic plasticity

https://doi.org/10.1016/j.neuropharm.2013.05.043Get rights and content

Highlights

  • Review focusing on available data regarding sensitive and reliable detection of endogenous BDNF.

  • Special focus on release sites and release time course of endogenous BDNF secretion.

  • Special focus on the role only of endogenous BDNF in synaptic plasticity at the single cell level.

  • Reviews the sparse knowledge regarding the role of BDNF in spike timing-dependent plasticity.

Abstract

Overwhelming evidence collected since the early 1990's strongly supports the notion that BDNF is among the key regulators of synaptic plasticity in many areas of the mammalian central nervous system. Still, due to the extremely low expression levels of endogenous BDNF in most brain areas, surprisingly little data i) pinpointing pre- and postsynaptic release sites, ii) unraveling the time course of release, and iii) elucidating the physiological levels of synaptic activity driving this secretion are available. Likewise, our knowledge regarding pre- and postsynaptic effects of endogenous BDNF at the single cell level in mediating long-term potentiation still is sparse. Thus, our review will discuss the data currently available regarding synaptic BDNF secretion in response to physiologically relevant levels of activity, and will discuss how endogenously secreted BDNF affects synaptic plasticity, giving a special focus on spike timing-dependent types of LTP and on mossy fiber LTP. We will attempt to open up perspectives how the remaining challenging questions regarding synaptic BDNF release and action might be addressed by future experiments.

This article is part of the Special Issue entitled ‘BDNF Regulation of Synaptic Structure, Function, and Plasticity’.

Introduction

There is certainly no doubt that BDNF is one of the central mediators and modulators of synaptic plasticity in the CNS. Numerous excellent previous reviews covered this topic and summarized the convincing evidence that BDNF promotes neuronal differentiation of stem cells, axonal and dendritic growth of neuronal processes, formation and maturation of glutamatergic and GABAergic synapses, and activity-dependent refinement of synaptic connections, like long-term potentiation (LTP), underlying learning and memory formation (see e.g. Bramham and Messaoudi, 2005, Gottmann et al., 2009, Park and Poo, 2013; review articles in this special issue). BDNF is such an attractive candidate for regulating synaptic transmission, since it is released locally at synapses in an activity-dependent manner, thus allowing local feedback at the level of individual synapses (Walz et al., 2006). In spite of the overwhelming evidence for BDNF in shaping synaptic plasticity, still surprisingly little is known regarding the exact location and time point of action of BDNF when synaptic plasticity is just being induced. Among the open questions are:

  • a)

    Is BDNF preferentially an ambient factor just favoring the induction and expression of synaptic plasticity (i.e. a synaptic modulator), or is it a mediator of synaptic changes representing the critical trigger for setting the changes in motion?

  • b)

    Is BDNF secreted from pre- or from postsynaptic elements to induce the specific synaptic changes, and what are the physiologically relevant patterns of synaptic activity that trigger synaptic BDNF secretion and decide whether it is pre- or postsynaptic?

  • c)

    Which parameters decide whether a certain BDNF-dependent synaptic change is mediated by pre- or postsynaptic alterations?

As will become evident from data discussed in this review, we are convinced that – depending on individual circumstances – BDNF i) can be both, a mediator or a modulator of synaptic plasticity, ii) can be released pre- and postsynaptically, and iii) can alter pre- and postsynaptic functions even simultaneously at the same individual synapse.

Importantly, intracellular as well as extracellular BDNF protein levels are extremely low, and BDNF release and action is local. Thus, to enable physiologically relevant insights into synaptic BDNF dynamics, experiments allowing space and time resolved analysis of endogenous BDNF are required, and will be focused on in this review, wherever respective data are available.

Secretion of BDNF requires the previous expression of BDNF mRNA, the subsequent translation of proBDNF protein, and the targeting of proBDNF and processed mature BDNF into exocytotic vesicles (for review compare: Lessmann et al., 2003, Lessmann and Brigadski, 2009). Alternatively, previously secreted BDNF can be endocytosed at primary sites of release and be recycled for additional rounds of secretion even at far distant sites in the same neuron (Santi et al., 2006, von Bartheld et al., 2001). Thus, any knowledge about the brain areas and subcellular neuronal compartments containing BDNF mRNA and protein give first important hints at loci of BDNF secretion.

The existence of mRNA is a prerequisite for the synthesis of proteins. Consequently, the presence of BDNF mRNA in rodents correlates very well with the site of synthesis of BDNF protein in specific tissues and cells (Conner et al., 1997). The mRNA for BDNF is widely distributed throughout the central nervous system (CNS) of rat and mice, including brain regions like the hippocampal formation (Conner et al., 1997, Connor and Dragunow, 1998, Ernfors et al., 1990, Hofer et al., 1990, Phillips et al., 1990, Schmidt-Kastner et al., 1996, Son and Winzer-Serhan, 2009, Wetmore et al., 1990, Wetmore et al., 1991), cerebral cortex (Conner et al., 1997, Hofer et al., 1990, Huntley et al., 1992, Phillips et al., 1990, Schmidt-Kastner et al., 1996, Timmusk et al., 1993, Wetmore et al., 1990, Wetmore et al., 1991), thalamus (Conner et al., 1997, Hofer et al., 1990, Schmidt-Kastner et al., 1996, Timmusk et al., 1993), hypothalamus (Conner et al., 1997, Hofer et al., 1990, Liao et al., 2012), olfactory bulb (Conner et al., 1997, Hofer et al., 1990, Phillips et al., 1990), amygdala (Conner et al., 1997, Phillips et al., 1990, Schmidt-Kastner et al., 1996, Wetmore et al., 1991), cerebellar granule cell layer (Hofer et al., 1990, Vazquez-Sanroman et al., 2013, Wetmore et al., 1990) and spinal cord (Conner et al., 1997, Hofer et al., 1990, Luo et al., 2001). In the hippocampus a similar distribution of BDNF mRNA was described for rodents and primates (Phillips et al., 1990). Similar to this pattern of BDNF mRNA expression, BDNF protein can be detected throughout the CNS, with highest expression levels in the hippocampal formation and cerebral cortex (Conner et al., 1997, Yan et al., 1997, Dugich-Djordjevic et al., 1995, Kokaia et al., 1996, Schmidt-Kastner et al., 1996, Wetmore et al., 1991). However, conflicting results regarding the expression of BDNF protein were observed in the hippocampal formation. While there is consensus in the literature that CA1 pyramidal cells of the hippocampus contain BDNF mRNA (An et al., 2008, Conner et al., 1997, Hofer et al., 1990, Kokaia et al., 1996, Schmidt-Kastner et al., 1996, Son and Winzer-Serhan, 2009, Timmusk et al., 1993, Tongiorgi et al., 2004, Wetmore et al., 1990, Wetmore et al., 1991; reviewed in Tongiorgi, 2008), the expression of BDNF protein in this cell type is discussed controversially. Numerous previous studies from different labs using distinct antibodies provided clear evidence that endogenous BDNF protein can be detected in somata and dendrites of CA1 pyramidal cells (see e.g. Conner et al., 1997, Dugich-Djordjevic et al., 1995, Schmidt-Kastner et al., 1996, Wetmore et al., 1991; reviewed in Lessmann et al., 2003). However, other studies (also using distinct antibodies) failed to observe such BDNF immunoreactivity in CA1 pyramidal cells, or observed only a weak immunoreactivity in some individual neurons in the temporal hippocampus (Dieni et al., 2012, Yan et al., 1997). There are several possible explanations for these discordant results: for example different BDNF antibodies are known to show cross-reactivity with non-BDNF protein species, which might give rise to false positive results for BDNF immunohistochemical detection (Matsuda et al., 2009, Matsumoto et al., 2008; compare Dieni et al., 2012). On the other hand, highly specific antibodies which avoid false positive BDNF detection might work at the expense of sensitivity for the low levels of BDNF expressed in CA1 of the hippocampus, which might explain the absence of BDNF protein detection in CA1 in some studies. Furthermore, critical parameters of immunohistochemical staining procedures, like fixation (Yan et al., 1997) or permeabilisation can decrease BDNF immunreactivity or give rise to differential access of antibodies to the interior of the distinct subsets of BDNF vesicles. In addition, BDNF expression is highly regulated not only by electrical activity but also by enriched environment, dietary restriction, light, stress, and circadian rhythm (reviewed in Chourbaji et al., 2011). Different housing conditions of the animals, like enriched environment or lower amount of stress, increase the expression of BDNF in CA1 dendrites (Chourbaji et al., 2011, Falkenberg et al., 1992, Ickes et al., 2000, Smith et al., 1995). Since the hippocampal formation is a brain region of utmost interest, further efforts are clearly required to solve these conflicting results about BDNF protein level in CA1 pyramidal cells. Due to the unambiguous consensus about BDNF mRNA level in this cell type, the question remains, whether BDNF expression might be regulated more tightly in CA1 pyramidal cells than in other brain areas.

At subcellular level, BDNF mRNA is distributed not only in somatic structures but also in dendritic compartments, both, in cultured neurons and in vivo (An et al., 2008, Capsoni et al., 1999, Lau et al., 2010, Liao et al., 2012, Righi et al., 2000, Tongiorgi et al., 1997, Tongiorgi et al., 2004, Tongiorgi et al., 2006, Tongiorgi, 2008, Waterhouse et al., 2012), but it is absent in axons (Tongiorgi et al., 1997; but see Ma et al., 2012). Dendritic BDNF mRNA was observed in hippocampal CA1, CA2 and CA3 pyramidal cells, granule cells, cortical neurons and hypothalamic neurons, and increased levels were detected in response to electrical activity (Wetmore et al., 1994; reviewed in Tongiorgi, 2008). In addition, dendritic localization of BDNF mRNA is dependent on polyadenylation of mRNA transcripts. In conditional transgenic animals which lacked the long 3′untranslated region (UTR) mRNA, dendritic targeting of BDNF mRNA was impaired in cortical and hippocampal neurons (An et al., 2008). While somatic localization of BDNF mRNA was observed for short and long 3′UTR BDNF mRNA, only long 3′UTR transcripts were targeted to dendrites of cortical and hippocampal CA1 neurons. Since the appearance of BDNF mRNA in dendrites indicates local protein synthesis followed by dendritic release of endogenous BDNF in these cells, this difference in dendritic BDNF mRNA localization has impact on physiological functions (An et al., 2008, Lau et al., 2010, Waterhouse et al., 2012). In accordance with this subcellular pattern of BDNF mRNA expression, endogenous BDNF protein was detected in somatic compartments and dendrites of cultured neurons as well as in dendrites of hippocampal neurons, cortical neurons and hypothalamic neurons in vivo (Adachi et al., 2013, Aoki et al., 2000, Jakawich et al., 2010, Ma et al., 2012, Matsuda et al., 2009, Swanwick et al., 2004, Tongiorgi et al., 2004, Wetmore et al., 1991, Yang et al., 2009b). However, Dieni and colleagues could not detect any endogenous BDNF in dendrites of hippocampal neurons. According to the conflicting results of BDNF protein expression in CA1 pyramidal cells, several explanations for this outcome are possible (see above). Besides dendritic localization, endogenous BDNF was also found in axons of cultured neurons and in vivo (Aoki et al., 2000, Buldyrev et al., 2006, Dieni et al., 2012, Fawcett et al., 1997, Matsuda et al., 2009, Michael et al., 1997, Swanwick et al., 2004). The subcellular distribution of BDNF protein has also been analyzed by electron microscopy. Localization of endogenous BDNF was observed in axon terminals within the lateral geniculate nucleus, the superior colliculus (Avwenagha et al., 2006), the amygdala (Salio et al., 2007) and in axonal terminals of dorsal root ganglia (Luo et al., 2001, Michael et al., 1997, Salio et al., 2007), trigeminal nociceptors (Buldyrev et al., 2006), dentate gyrus granule cells and CA3 pyramidal cells (Dieni et al., 2012). In addition, BDNF was observed in the postsynaptic density and dendritic shafts of cerebral cortex at the ultrastructural level (Aoki et al., 2000). Further analysis of subcellular BDNF distribution has been performed in cultured neurons which either expressed endogenous BDNF or fluorescently labeled BDNF. In these studies, BDNF was localized in organelles, like endoplasmic reticulum or Golgi apparatus, and in vesicles at extrasynaptic, as well as pre- and postsynaptic sites in axons and dendrites (Brigadski et al., 2005, Dean et al., 2012, Hartmann et al., 2001, Kolarow et al., 2007, Wu et al., 2004), suggesting extrasynaptic and synaptic release of BDNF. The distribution and the density of fluorescently labeled BDNF vesicles in these studies was similar to the density and distribution of endogenous BDNF in cultured neurons (Matsuda et al., 2009). Solely, the size of BDNF containing vesicles was significantly increased after expression of BDNF-GFP. These observations support the view that the transport and release of BDNF containing vesicles is comparable for endogenous and fluorescently labeled BDNF.

After transport of BDNF-containing vesicles into axons and dendrites, the neurotrophic factor is secreted either constitutively or in an activity-dependent manner in response to electrical activity (reviewed in Lessmann and Brigadski, 2009). Stimulus induced secretion of proteins is a highly dynamic process which is difficult to analyze with sufficient sensitivity and appropriate combined high spatial and time resolution. Currently these requirements can only be fulfilled by manipulating the secretory protein (BDNF) e.g. by attaching fluorescent tags. In contrast, immunohistochemical localization studies can indicate only potential sites of BDNF secretion at high spatial resolution, whereas this method represents only a snapshot of BDNF dynamics, which for example cannot distinguish between a pool of ready releasable BDNF vesicles and BDNF containing vesicles just being transported in anterograde or retrograde direction. Even more importantly, static vesicles detected with immunohistochemistry are not appropriate to judge BDNF secretion. Time resolved secretion analysis requires high sensitivity detection of small amounts of BDNF just secreted in a defined time interval. In 2008, Nakajima and colleagues introduced a cell-based fluorescence indicator which could detect endogenous BDNF in the picomolar range. This indicator – a chimeric receptor containing an extracellular BDNF binding domain and an intracellular tyrosine kinase domain – translates binding of BDNF into a fluorescent signal at high time resolution (Nakajima et al., 2008). Elegant as this assay is, it can, however, not resolve the site of activity-dependent secretion. Different methods can be used to measure endogenous BDNF secretion, but high spatio-temporal analysis of BDNF secretion can so far only be achieved using live cell imaging of fluorescently labeled BDNF (see 1.2.2 Subcellular secretion sites of BDNF, 1.2.3 Induction of BDNF secretion, 1.2.4 Kinetics of BDNF secretion).

In 1994, Wetmore and colleagues could provide an elegant first indication for stimulation induced BDNF secretion from somatic and dendritic sites in rat hippocampus (Wetmore et al., 1994). Using immunohistochemical analysis they observed a shift of BDNF immunoreactivity from somatic and dendritic regions to the surrounding neuropil in response to kainate stimulation. Two years later, blockade of BDNF function by application of TrkB receptor bodies, which scavenge released BDNF in extracellular space (see below), provided evidence for BDNF secretion in the hippocampus during LTP (Figurov et al., 1996). TrkB receptor bodies (TrkB-Fc; TrkB-IgG) are molecularly engineered scavengers of extracellular BDNF, consisting of the BDNF-binding part of the BDNF specific TrkB receptor fused at the cDNA level to the human Fc region of antibodies (Shelton et al., 1995). Upon extracellular application, TrkB-Fc bind and functionally inactivate the released BDNF. BDNF specific antibodies can serve the same purpose of extracellular BDNF scavenging but – due to the larger MW compared to TrkB-Fc – show decreased tissue penetration. While these scavengers are an elegant method to prove a functional role of endogenously released BDNF, they cannot resolve the site of BDNF release. Furthermore, to secure scavenging by TrkB-Fc (and even more so by BDNF antibodies) pre-incubation (at least 10 min – in most studies even longer) is required to enable sufficient access of TrkB-Fc also to synaptic clefts to block the action of BDNF at synaptic sites. Thus Figurov and colleagues could not distinguish whether constitutive or activity-dependent release was responsible for BDNF function in LTP, since application of the BDNF scavenger was performed for several hours prior to LTP induction. In addition, it is unknown whether the use of TrkB-Fc in vivo has the same power as for in vitro studies, since they can function as a carrier of BDNF even enhancing distribution and biological activity of BDNF in vivo (Croll et al., 1998; but see Cabelli et al., 1997). However, using this technique of blocking acute BDNF function, different groups could indicate release of endogenous BDNF in hippocampus (Amaral and Pozzo-Miller, 2007, Chen et al., 1999, Chytrova et al., 2008, Figurov et al., 1996, Griesbach et al., 2009, Kang et al., 1997, Kim et al., 2012, Li et al., 2010, Miladi-Gorji et al., 2011, Tanaka et al., 2008, Tyler et al., 2006, Vaynman et al., 2004), as well as in cortex (Aicardi et al., 2004, Akaneya et al., 1997, Cabelli et al., 1997, Clarkson et al., 2011, Inagaki et al., 2008, Sermasi et al., 2000), amygdala (Li et al., 2011, Ma et al., 2011, Meis et al., 2012, Ou et al., 2010, Ou and Gean, 2006), nucleus accumbens (Liang et al., 2011), substantia nigra (after injury (Canudas et al., 2005)), cerebellum (Xu et al., 2011), striatum (Jia et al., 2010) and spinal cord (Huie et al., 2012, Kerr et al., 1999, Munson et al., 1997, Pezet et al., 2002, Seebach et al., 1999, Slack and Thompson, 2002, Ying et al., 2008). Due to the required incubation time for TrkB-Fc and the necessarily indirect and time consuming readout of inhibition of BDNF function in this type of experiment, this technique enables analysis of BDNF release only at low temporal resolution without any information about subcellular sites of release.

The analysis of endogenously released BDNF at higher time resolution can be performed with ELISA measurements (see also 1.2.2). ELISA measurements require the sampling of the extracellular fluid superfusing the tissue during release and subsequent analysis in sampled aliquots, thus limiting the time resolution to ∼30 s. While ELISA measurements allow to determine secretion of endogenous BDNF from intact brain slice tissue, the release sites cannot be deduced from these experiments. Using ELISA techniques, Canossa and colleagues investigated for the first time stimulus dependent accumulation of endogenous BDNF in extracellular space. In this study, neurotransmitter and neurotrophin induced release, respectively, of endogenous BDNF from acute hippocampal slices at low time resolution was described (Canossa et al., 1997). According to the most abundant expression of BDNF in hippocampus and cortex, analysis of endogenous BDNF release was most frequently performed with ELISA measurements in acute slices of these two regions (Aicardi et al., 2004, Bergami et al., 2008, Canossa et al., 1997, Griesbeck et al., 1999). Further, accumulation of endogenous BDNF in extracellular space was analyzed in isolated dorsal horn slices (Lever et al., 2001, Slack et al., 2004) and in the supraoptic nucleus of the hypothalamus in vivo (Arancibia et al., 2007). However, most of the measurements analyzing stimulus dependent accumulation of endogenous BDNF in extracellular space have been performed with supernatants from cultured cells. Release of endogenous BDNF, in principle, has been observed in nodose and petrosal ganglion neurons (Balkowiec and Katz, 2000, Hsieh et al., 2010, Lin et al., 2011), hippocampal neurons (Babu et al., 2009, Balkowiec and Katz, 2002, Dean et al., 2009, Matsumoto et al., 2008), cortical neurons (Adachi et al., 2013, Bachis et al., 2012, Jeon et al., 2011, Lou et al., 2005, Wu et al., 2010), postnatal utricle explants (Chabbert et al., 2003), brainstem spinal cord preparations (Ba et al., 2005), cerebellar granule cells (Rabin et al., 2002, Xu et al., 2011), trigeminal neurons (Buldyrev et al., 2006), schwann cells (Faroni et al., 2013, Verderio et al., 2006), astrocytes (Hou et al., 2011, Hutchinson et al., 2009, Jean et al., 2008, Su et al., 2012, Zhang et al., 2012), microglia (Coull et al., 2005, Ferrini et al., 2013, Gomes et al., 2013, Hutchinson et al., 2009, Nakajima et al., 2002, Trang et al., 2009), Fibroblasts (Klein et al., 2012, Warnecke et al., 2012), endothelial cells (Wang et al., 2006), eosinophil granulocytes (Noga et al., 2003) and platelets (Fujimura et al., 2002, Hochstrasser et al., 2013, Stoll et al., 2011, Tamura et al., 2011, Watanabe et al., 2010). Nevertheless, detection of endogenous BDNF in neuronal cultures is a challenging issue. Protocols with specific modifications – like culturing of neurons in the presence of an immobilized BDNF antibody in the culture dish (ELISA in situ; Balkowiec and Katz, 2000) or additional presence of TrkB-Fc in the medium, followed by precipitation of receptor bodies and subsequent western blot analysis (Yang et al., 2009b) need to be used to reach the detection limit, especially in neuronal cultures. Therefore, most studies analyzing the extracellular accumulation of neuronal BDNF have been performed in cell cultures transfected with epitope-tagged or fluorescently labeled BDNF to use well established high sensitivity tag-directed antibodies for BDNF-detection, and to increase the expression level of BDNF. Although density and distribution of endogenous BDNF and BDNF-GFP seem to be similar in cultured neurons (Matsuda et al., 2009), the surplus in packing density might shift the protein level above detection limits. In spite of the overexpression and tagging which, both, can potentially lead to changes in targeting, these experiments investigating the BDNF content in extracellular space were very helpful in analyzing the intracellular signaling cascades contributing to BDNF release (reviewed in Lessmann et al., 2003).

ELISA measurements analyzing accumulation of extracellular BDNF or measurements using BDNF scavenger thereby blocking BDNF function provide information about the brain area in which endogenous BDNF secretion takes place. However, these methods neither allow to detect the cell type showing BDNF secretion nor the subcellular site of secretion. Therefore, measurements aiming at identifying BDNF secretion sites were often performed in cultured neurons transfected with fluorescently labeled or epitope tagged BDNF. Using time lapse imaging of BDNF-GFP or BDNF-pHluorin, release of the protein was observed in dendrites of hippocampal neurons (Brigadski et al., 2005, Dean et al., 2009, Hartmann et al., 2001, Kojima et al., 2001, Kolarow et al., 2007, Kuczewski et al., 2008a, Matsuda et al., 2009; compare Fig. 1 and Table 1). In addition to dendritic extrasynaptic secretion of BDNF, postsynaptic targeting of the protein and secretion at postsynaptic sites was observed (Brigadski et al., 2005, de Wit et al., 2009, Hartmann et al., 2001, Kojima et al., 2001, Kolarow et al., 2007, Lochner et al., 2008, Rind et al., 2005, Yang et al., 2009a). Using similar methods, axonal localization and axonal secretion of BDNF was observed in hippocampal as well as in cortical neuronal cultures (Dean et al., 2009, Haubensak et al., 1998, Kohara et al., 2001, Matsuda et al., 2009, Sadakata et al., 2012, Scalettar et al., 2012, Shinoda et al., 2011). In addition to neuronal BDNF secretion, uptake of BDNF and re-exocytosis in visual pathways of chicks (Butowt and von Bartheld, 2001, von Bartheld et al., 2001) as well as by astrocytes and hippocampal neurons has been described (Bergami et al., 2008, Santi et al., 2006). The fast endocytosis of BDNF by astrocytes and neurons might explain the low levels of BDNF in the extracellular space which hinder more sensitive detection of extracelluar BDNF by ELISA measurements.

Further methods to analyze subcellular sites of BDNF release are represented by mixed cultures, like e.g. co-cultures of neurons with different genetic backrounds (i.e. selective knockin/knockout of genes affecting BDNF release in specific brain areas or cell types). For example, experiments in co-cultures of xenopus nerve and muscle cells revealed an acticity-dependent proBDNF release from postsynaptic muscle cells which regulates presynaptic stabilization (Yang et al., 2009a). Furthermore, Kohara and colleagues suggest secretion of BDNF from postsynaptic neurons to act as a trophic factor for GABAergic synapses in cortical organotypic slice cultures as revealed by knockout of BDNF in single cells (Kohara et al., 2007). Using co-cultures of Syt-IV knockout and wildtype dissociated hippocampal neurons, Dean and colleagues suggested that postsynaptic release of BDNF modulates presynaptic transmitter release frequency as revealed by FM destaining (see also Magby et al., 2006; but compare Zakharenko et al., 2003) while presynaptic release of BDNF might modulate the amplitude of mEPSCs, thus speaking in favor of BDNF dependent postsynaptic modifications. While this study suggests pre- as well as postsynaptic release of BDNF in cultures, several studies addressed this topic in hippocampal slice preparations. Interestingly, rescue of LTP in CA1 of hippocampal slices from conventional BDNF knockout by virus mediated expression of BDNF in postsynaptic CA1 (but not in presynaptic CA3) neurons provided indirect evidence for a role of dendritically released BDNF during LTP (Korte et al., 1996). In conditional transgenic animals which lack the long 3′UTR transcripts, impaired dendritic targeting of BDNF mRNA – suggesting impaired dendritic release of BDNF – was also associated with reduced hippocampal LTP and therefore supports the role of dendritically released BDNF in hippocampal LTP (An et al., 2008). Zakharenko and colleagues, however, suggested that only axonal BDNF contributes to LTP in CA1. Using conditional BDNF knockout mice, they showed that selective virus mediated expression of BDNF in CA3 rescued hippocampal LTP induced by 200 Hz stimulation while infection of CA1 failed to show a similar rescue (Zakharenko et al., 2003). Given the conflicting results for BDNF protein expression in CA1 pyramidal cells (compare 1.1.2), these LTP data further stress the importance to investigate the mechanisms regulating endogenous BDNF protein expression, thereby regulating LTP in CA1, in future studies. From the data available at present, it seems likely that BDNF is secreted in an activity-dependent manner from both, axon and dendrites, possibly exerting distinct functions, depending on release site of BDNF. New methods like e.g. single cell transfection BDNF-GFP in organotypic slices in individual CA1 and CA3 neurons – preferably under the control of the endogenous BDNF promoter to avoid overexpression – are required to pinpoint the subcellular release site of BDNF to correlate BDNF function to its site of release. In this respect a very elegant recent study employing single cell ko of BDNF in individual cortical pyramidal neurons resulting in inhibition of spine growth selectively in this neuron provided strong evidence that endogenous BDNF is released from postsynaptic dendrites (English et al., 2012).

Various stimuli have been described to induce release of BDNF. Besides different patterns of electrical activity of neurons, several transmitters and related substances provoke an accumulation of BDNF in the extracellular space (reviewed in Lessmann et al., 2003, Kuczewski et al., 2008b, Kuczewski et al., 2008a, Lessmann and Brigadski, 2009; summarized in Table 1). The most common electrical stimulation protocols which were demonstrated to induce BDNF release are also known to induce LTP. Using either HFS or TBS, diverse studies reveal evidence for BDNF release in hippocampal or cortical neurons. (Aicardi et al., 2004, Balkowiec and Katz, 2002, Bergami et al., 2008, Gartner and Staiger, 2002, Hartmann et al., 2001, Lever et al., 2001, Matsuda et al., 2009, Nagappan et al., 2009, Santi et al., 2006). In addition, depolarization of hippocampal neurons as well as spontaneous electrical network activity, like rhythmic neuronal discharges have been described to induce BDNF accumulation in extracellular space (Ba et al., 2005, Canossa et al., 2002, Kuczewski et al., 2008a, Lever et al., 2001, Magby et al., 2006). In case of low frequency stimulation (LFS), different results regarding BDNF secretion were obtained. While Aicardi and colleagues could correlate a lower amount of endogenously released BDNF with long term depression of fEPSP in cortical slices (Aicardi et al., 2004), no change in the amount of endogenously released BDNF was described for hippocampal and dorsal horn neurons after LFS (Balkowiec and Katz, 2002, Lever et al., 2001). In contrast to both of these studies, Nagappan and colleagues even observed an increase in proBDNF-EGFP in supernatant of hippocampal cultures after LFS (Nagappan et al., 2009). These discordant results might be explained by differences in details of stimulation paradigms between studies, as well as by different states of basal electrical activity in cultured neurons compared to intact slices. Furthermore, future studies need to address the role of GABAergic inhibition in these synaptic networks on BDNF release, since inhibitory synaptic drive, which potentially reduces BDNF secretion, might be different depending on stimulation paradigm.

Consistent with the idea that increased electrical activity drives BDNF release, glutamate application as well as elevated potassium concentration in the extracellular space are known to induce BDNF release from neurons (Brigadski et al., 2005, Canossa et al., 1997, de Wit et al., 2009, Dean et al., 2009, Egan et al., 2003, Goodman et al., 1996, Griesbeck et al., 1999, Hartmann et al., 2001, Lou et al., 2005, Nagappan et al., 2009, Nakajima et al., 2008, Santi et al., 2006, Waterhouse et al., 2012, Xia et al., 2009). In addition, activation of different receptor classes – G protein coupled receptors, receptor tyrosine kinases, P2X purinoreceptors and transient receptor potential cation (TRPC) channels – has been shown to induce BDNF release. Thus, metabotropic glutamate receptor (mGluR) activation (Balkowiec and Katz, 2002, Canossa et al., 2001, Santi et al., 2006), activation of GABAB receptors (Fiorentino et al., 2009, Kuczewski et al., 2011), glycine by means of NMDAR facilitation (Babu et al., 2009), prostaglandin signaling via cAMP/PKA pathway (Hutchinson et al., 2009), neurotrophin signaling via TrkB and TrkC receptors (Canossa et al., 1997, Santi et al., 2006) and capsaicin (Lever et al., 2001) can – depending on neuronal subtype – all be effective secretagogues for neurons. Furthermore, adenosine triphosphate (ATP) was shown to induce release of endogenous BDNF from fibroblast and microglia (Klein et al., 2012, Trang et al., 2009). BDNF release induced by neurotrophins as well as BDNF release induced by activation of G protein coupled receptors, like mGluR or GABAB receptors, have been described to depend on phospholipase C (PLC) and protein kinase C (PKC) signaling (Canossa et al., 2001, Canossa et al., 2002, Kuczewski et al., 2011). Both electrically induced and chemically induced BDNF release depend on Ca2+-influx from extracellular space as well as Ca2+-mobilization from intracellular stores (reviewed in Lessmann et al., 2003, Lessmann and Brigadski, 2009). Whether the observation of BDNF release in response to application of hyperosmotic saline is – similar to neurotransmitter release (Rosenmund and Stevens, 1996) – due to unspecific Ca2+-independent stimulation of exocytosis remains to be determined.

All of the above mentioned neurotransmitters and neuromodulators might exert their effects on BDNF release at least in part by modulating synaptic network activity. Thus, future studies should aim at distinguishing between direct transmitter induced secretion vs. indirect modulation of networks, which indirectly might affect BDNF secretion without being a genuine secretagogue.

The characteristics of endogenous BDNF secretion have been studied using ELISA measurements in hippocampal, cortical or dorsal horn slices at relatively low time resolution (Aicardi et al., 2004, Bergami et al., 2008, Canossa et al., 2001, Lever et al., 2001). These studies described an increase of extracellular BDNF concentration after stimulation which takes about 5–20 min. Nakajima and colleagues analyzed the kinetics of endogenous BDNF release in cultured hippocampal neurons by using a cell-based fluorescent indicator (compare 1.2). In this assay, application of 10 μM glutamate induced release of 230 pM BDNF within 5 min. All other studies addressing BDNF release kinetics in hippocampal or cortical cultures employed live cell imaging of neuronal cultures expressing fluorescently labeled BDNF, measuring secretion as a decrease in intravesicular GFP levels. Irrespective of the stimulation protocol (e.g. electrical stimulation, glutamate or capsaicin application) the kinetics of release of fluorescently labeled BDNF was similar as described for peptide release in general (see e.g. Barg et al., 2002). After fusion pore opening of BDNF containing vesicles (either immediately with stimulation or with a delay of tens of seconds; compare Kolarow et al., 2007), the protein is released into the extracellular space. The time course of BDNF release could be fitted with a monoexponential function, yielding a half decay time in the range of 200–250 s (Brigadski et al., 2005). Importantly, BDNF release is regulated by intravesicular pH (reviewed in Lessmann and Brigadski, 2009), and intravesicular pre-release neutralization of initially acidic vesicles can speed up BDNF release, yielding a half decay time of about 50 s (Brigadski et al., 2005). Using BDNF-pHluorin, intravesicular pH of BDNF containing vesicles prior to release was shown to be more acidic in axons than in dendrites (Dean et al., 2009, Matsuda et al., 2009). Consistent with the pH dependence of BDNF release kinetics (with acidic granules delaying release because of slow dissolution of peptide cores; compare Lessmann and Brigadski, 2009), axonal BDNF release was observed less frequently (Matsuda et al., 2009). In axons, stationary BDNF containing vesicles exhibit transient fusion pore opening after TBS without any net release of BDNF, and intense stimulation was required to induce release of BDNF in axons (compare presumably presynaptic BDNF release upon intense 200 Hz LTP induction protocol; Zakharenko et al., 2003), while lower stimulation strength (like e.g., TBS stimulation) most often lead to full collapse of BDNF containing vesicles resulting in BDNF release in dendrites (Kuczewski et al., 2008a, Kuczewski et al., 2008b, Matsuda et al., 2009).

In addition to the first stimulus dependent release of BDNF within the first 5 min after stimulation, a second phase of BDNF release was described for microglial cultures, with a delay of 60 min after initial stimulation (Trang et al., 2009). This second release phase was associated with increased BDNF protein expression and exhibited a similar dependence on extracellular calcium than the first release phase (Trang et al., 2009). This additional increase in BDNF content in extracellular space might contribute to the consolidation of BDNF induced physiological changes and could act as a temporal feedback loop. Interestingly, a delayed second phase of BDNF secretion for consolidation of BDNF dependent fear learning in the amygdale has been described (Ou et al., 2010). Thus, whether a delayed second round of BDNF secretion can also be observed in neurons and thereby accounts for memory consolidation effects remains to be determined.

The release of BDNF represents a dynamic process with high impact for a multitude of biological functions. Finding out whether constitutive or stimulus dependent release of BDNF as well as the site and time course of release can be correlated with distinct physiological effects of BDNF at its site of release will be challenging tasks for the future. Several functions of BDNF have been shown by scavenging of endogenously released BDNF. Due to these studies, it is known that endogenous BDNF is involved in neurogenesis, plasticity processes and cell survival, demonstrating the importance of this small neurotrophic factor for the CNS.

Section snippets

BDNF action in synaptic plasticity

An involvement of BDNF in long-term potentiation (LTP) – which is believed to be a neuronal correlate of learning and memory (for reviews see e.g., Bliss and Collingridge, 1993, Malenka and Bear, 2004) – was first reported in studies showing impaired LTP in CA1 of hippocampal slices of heterozygous BDNF knockout (ko) animals (Korte et al., 1995, Patterson et al., 1996, Pozzo-Miller et al., 1999; for review e.g., Gottmann et al., 2009, Yoshii and Constantine-Paton, 2010, Park and Poo, 2013).

Summary and conclusion

Although detailed knowledge regarding the action of endogenous BDNF in activity-dependent synaptic plasticity is still sparse, the overall evidence as discussed in our review supports the notion that – depending on individual circumstances –BDNF can be either a mediator or a modulator of synaptic plasticity, it can be released pre- and postsynaptically, and can alter pre- and postsynaptic functions even simultaneously at the same individual synapse.

Most BDNF effects in LTP seem to be

Futher perspectives

Developing new experimental tools and approaches allowing to selectively interfere with BDNF secretion, or allowing to follow the subcellular sites, kinetics and the duration of endogenous BDNF secretion at physiological expression level will be crucial to elucidate BDNF functions in synaptic plasticity in sufficient detail.

With these new approaches we believe future research in the field should be centered on the following aspects:

  • i.

    Investigation of secretion sites of endogenous BDNF at defined

References (245)

  • S. Capsoni et al.

    Differential regulation of brain-derived neurotrophic factor messenger RNA cellular expression in the adult rat visual cortex

    Neuroscience

    (1999)
  • S. Chourbaji et al.

    Altering BDNF expression by genetics and/or environment: impact for emotional and depression-like behaviour in laboratory mice

    Neurosci. Biobehav Rev.

    (2011)
  • B. Connor et al.

    The role of neuronal growth factors in neurodegenerative disorders of the human brain

    Brain Res. Brain Res. Rev.

    (1998)
  • S.D. Croll et al.

    Co-infusion with a TrkB-Fc receptor body carrier enhances BDNF distribution in the adult rat brain 3

    Exp. Neurol.

    (1998)
  • I. Duguid et al.

    Novel presynaptic mechanisms for coincidence detection in synaptic plasticity

    Curr. Opin. Neurobiol.

    (2006)
  • M.F. Egan et al.

    The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function

    Cell

    (2003)
  • P. Ernfors et al.

    Identification of cells in rat brain and peripheral tissues expressing mRNA for members of the nerve growth factor family

    Neuron

    (1990)
  • T. Falkenberg et al.

    Cortical transynaptic activation of tyrosine kinase receptor trkB messenger RNA expression in rat hippocampus

    Neuroscience

    (1992)
  • J.P. Fawcett et al.

    Detection of brain-derived neurotrophic factor in a vesicular fraction of brain synaptosomes9

    J. Biol. Chem.

    (1997)
  • D.E. Feldman

    The spike-timing dependence of plasticity

    Neuron

    (2012)
  • S. Fujisawa et al.

    BDNF boosts spike fidelity in chaotic neural oscillations

    Biophys. J.

    (2004)
  • L.J. Goodman et al.

    Regulated release and polarized localization of brain-derived neurotrophic factor in hippocampal neurons

    Mol. Cell Neurosci.

    (1996)
  • G.S. Griesbach et al.

    Exercise-induced improvement in cognitive performance after traumatic brain injury in rats is dependent on BDNF activation

    Brain Res.

    (2009)
  • E.P. Huang

    Synaptic plasticity: going through phases with LTP

    Curr. Biol.

    (1998)
  • Y.Z. Huang et al.

    Zinc-mediated transactivation of TrkB potentiates the hippocampal mossy fiber-CA3 pyramid synapse

    Neuron

    (2008)
  • J.R. Huie et al.

    Brain-derived neurotrophic factor promotes adaptive plasticity within the spinal cord and mediates the beneficial effects of controllable stimulation

    Neuroscience

    (2012)
  • G.W. Huntley et al.

    Developmental expression of brain derived neurotrophic factor mRNA by neurons of fetal and adult monkey prefrontal cortex

    Brain Res. Dev. Brain Res.

    (1992)
  • A.J. Hutchinson et al.

    Activation of EP2 prostanoid receptors in human glial cell lines stimulates the secretion of BDNF

    Neurochem. Int.

    (2009)
  • B.R. Ickes et al.

    Long-term environmental enrichment leads to regional increases in neurotrophin levels in rat brain

    Exp. Neurol.

    (2000)
  • T. Inagaki et al.

    Brain-derived neurotrophic factor-mediated retrograde signaling required for the induction of long-term potentiation at inhibitory synapses of visual cortical pyramidal neurons

    Neurosci. Res.

    (2008)
  • S.K. Jakawich et al.

    Local presynaptic activity gates homeostatic changes in presynaptic function driven by dendritic BDNF synthesis

    Neuron

    (2010)
  • H. Kang et al.

    Neurotrophins and time: different roles for TrkB signaling in hippocampal long-term potentiation

    Neuron

    (1997)
  • R.J. Kelleher et al.

    Translational regulatory mechanisms in persistent forms of synaptic plasticity

    Neuron

    (2004)
  • W.C. Abraham

    Metaplasticity: tuning synapses and networks for plasticity

    Nat. Rev. Neurosci.

    (2008)
  • N. Adachi et al.

    Phencyclidine-induced decrease of synaptic connectivity via inhibition of BDNF secretion in cultured cortical neurons

    Cereb. Cortex

    (2013)
  • G. Aicardi et al.

    Induction of long-term potentiation and depression is reflected by corresponding changes in secretion of endogenous brain-derived neurotrophic factor

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

    (2004)
  • Y. Akaneya et al.

    Brain-derived neurotrophic factor enhances long-term potentiation in rat visual cortex

    J. Neurosci.

    (1997)
  • J. Alder et al.

    Early presynaptic and late postsynaptic components contribute independently to brain-derived neurotrophic factor-induced synaptic plasticity

    J. Neurosci.

    (2005)
  • M.D. Amaral et al.

    TRPC3 channels are necessary for brain-derived neurotrophic factor to activate a nonselective cationic current and to induce dendritic spine formation

    J. Neurosci.

    (2007)
  • C. Aoki et al.

    Localization of brain-derived neurotrophic factor and TrkB receptors to postsynaptic densities of adult rat cerebral cortex

    J. Neurosci. Res.

    (2000)
  • A. Artola et al.

    The involvement of n-methyl-d-aspartate receptors in induction and maintenance of long-term potentiation in rat visual cortex

    Eur. J. Neurosci.

    (1990)
  • F. Ba et al.

    Brain-derived neurotrophic factor release with neuronal activity in fetal rats

    Neuroreport

    (2005)
  • H. Babu et al.

    Synaptic network activity induces neuronal differentiation of adult hippocampal precursor cells through BDNF signaling

    Front Neurosci.

    (2009)
  • A. Bachis et al.

    Human immunodeficiency virus type 1 alters brain-derived neurotrophic factor processing in neurons

    J. Neurosci.

    (2012)
  • A. Balkowiec et al.

    Activity-dependent release of endogenous brain-derived neurotrophic factor from primary sensory neurons detected by ELISA in situ

    J. Neurosci.

    (2000)
  • A. Balkowiec et al.

    Cellular mechanisms regulating activity-dependent release of native brain-derived neurotrophic factor from hippocampal neurons

    J. Neurosci.

    (2002)
  • G. Barrionuevo et al.

    Associative long-term potentiation in hippocampal slices

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

    (1983)
  • M. Bergami et al.

    Uptake and recycling of pro-BDNF for transmitter-induced secretion by cortical astrocytes

    J. Cell Biol.

    (2008)
  • J. Bischofberger et al.

    Timing and efficacy of transmitter release at mossy fiber synapses in the hippocampal network

    Pflugers Arch.

    (2006)
  • T.V. Bliss et al.

    A synaptic model of memory: long-term potentiation in the hippocampus

    Nature

    (1993)
  • Cited by (0)

    View full text