Biochemical Society Transactions

Neurotrophins: Mechanisms in Disease and Therapy

Brain-derived neurotrophic factor and control of synaptic consolidation in the adult brain

J. Soulé, E. Messaoudi, C.R. Bramham

Abstract

Interest in BDNF (brain-derived neurotrophic factor) as an activity-dependent modulator of neuronal structure and function in the adult brain has intensified in recent years. Localization of BDNF and its receptor tyrosine kinase TrkB (tropomyosin receptor kinase B) to glutamate synapses makes this system attractive as a dynamic, activity-dependent regulator of excitatory transmission and synaptic plasticity in the adult brain. Development of stable LTP (long-term potentiation) in response to high-frequency stimulation requires new gene expression and protein synthesis, a process referred to as synaptic consolidation. Several lines of evidence have implicated endogenous BDNF–TrkB signalling in synaptic consolidation. This mini-review emphasizes new insights into the molecular mechanisms underlying this process. The immediate early gene Arc (activity-regulated cytoskeleton-associated protein) is strongly induced and transported to dendritic processes after LTP induction in the dentate gyrus in live rats. Recent work suggests that sustained synthesis of Arc during a surprisingly protracted time-window is required for hyperphosphorylation of actin-depolymerizing factor/cofilin and local expansion of the actin cytoskeleton in vivo. Moreover, this process of Arc-dependent synaptic consolidation is activated in response to brief infusion of BDNF. Microarray expression profiling has also revealed a panel of BDNF-regulated genes that may co-operate with Arc during LTP maintenance. In addition to regulating gene expression, BDNF signalling modulates the fine localization and biochemical activation of the translation machinery. By modulating the spatial and temporal translation of newly induced (Arc) and constitutively expressed mRNA in neuronal dendrites, BDNF may effectively control the window of synaptic consolidation. These findings have implications for mechanisms of memory storage and mood control.

  • brain-derived neurotrophic factor
  • hippocampus
  • long-term potentiation
  • neurotrophin
  • protein synthesis
  • synaptic plasticity

Introduction

The function of the neurotrophin family of secretory peptides, including nerve growth factor, BDNF (brain-derived neurotrophic factor) and neurotrophin-3, has been studied extensively in the context of neuronal development. Diverse functions of these neurotrophins in regulating neuronal survival, signalling and activity-dependent synaptic plasticity in the adult nervous system have also been identified [1]. Of the neurotrophin family, the BDNF–TrkB (tropomyosin receptor kinase B) system has emerged as the major regulator of excitatory synaptic transmission and plasticity, a role that fits with the expression of BDNF and TrkB at glutamate synapses.

Activity-dependent changes in synaptic strength, as exemplified by LTP (long-term potentiation), are thought to underlie memory storage and other adaptive mechanisms such as mood stability and drug addiction [2,3]. Research in the last decade suggests multiple, distinct functions for BDNF in LTP in the hippocampus [4]. The contributions of BDNF signalling to LTP can be classified as permissive or instructive. Permissive mechanisms make synapses competent for LTP. As an example of this, basal (non-evoked) release of BDNF promotes docking of neurotransmitter vesicles to the active zone, enabling sustained presynaptic transmission during LTP induction [5]. Instructive mechanisms are initiated in response to HFS (high-frequency stimulation) and required for subsequent development of LTP. Immediate release of BDNF modulates the induction and early maintenance phase of LTP [6], possibly through activation of a novel voltage-dependent sodium channel [7]. The depolarization elicited by activation of these channels facilitates NMDA (N-methyl-D-aspartate) receptor activation and calcium influx into spines during LTP induction [7]. In contrast, formation of stable LTP is coupled with sustained BDNF release and activation of TrkB receptors [8,9]. Genetic and pharmacological studies establishing the role of BDNF–TrkB in LTP have been reviewed in detail elsewhere [4]. This mini-review summarizes recent progress in identifying how BDNF regulates the protein synthesis-dependent component of LTP. A model depicting some of these mechanisms is presented in Figure 1.

Figure 1 BDNF as a trigger of synaptic consolidation

The mechanism of stable LTP formation at glutamatergic synapses is presented as a two-stage process: translation activation and Arc-dependent consolidation. In the translation activation stage, patterned HFS leads to sustained post-synaptic release of BDNF and activation of TrkB receptors pre- and post-synaptically. Post-synaptic TrkB leads to (i) rapid activation and translocation of the translation machinery in dendritic spines and (ii) Arc transcription in granule cell bodies. Translation activation is mediated by phosphorylation of the cap-binding protein eIF4E, and possibly by more mRNA-specific mechanisms such as relief of miRNA-mediated translation repression. Spines activated in this way may effectively capture and translate local mRNA pools. In this model, transcripts liberated from local RNA storage granules are translated first, followed by dendritic transport and sustained translation of newly synthesized Arc mRNA. During Arc-dependent consolidation, sustained translation of Arc is necessary for cofilin phosphorylation, local F-actin expansion and formation of stable LTP. This is a working model; several points require further experimental validation (see text). AMPAR, AMPA receptor; NMDAR, NMDA receptor; PSD, post-synaptic density.

Arc (activity-regulated cytoskeleton-associated protein), actin dynamics and synapse expansion

Persistent LTP is thought to occur when small spines are converted into large mushroom-shaped spines through a mechanism dependent on actin polymerization [10]. Fukazawa et al. [11] showed that LTP at medial perforant path–granule cell synapses of the dentate gyrus is associated with an increase in F-actin content at activated synapses and enhanced phosphorylation of cofilin. Phosphorylation of cofilin on Ser3 inhibits activity and promotes actin polymerization. LIMK (LIM domain kinase), one of the major cofilin kinases in brain, controls actin dynamics and morphology of dendritic spines through regulation of cofilin. Mice lacking the LIMK1 gene exhibit small, actin-poor spines that are unable to express stable LTP [12]. Taken together, this suggests a major role for cofilin regulation in actin-dependent enlargement of synapses and consolidation of LTP.

What are the events that couple gene expression and protein synthesis with synaptic consolidation? LTP is associated with the induction of a number of immediate-early genes and only one of these, Arc, is known to undergo transport to dendritic processes of granule cells [13,14]. Arc mRNA is enriched at stimulated synapses and Arc protein is transiently elevated in dendritic spines after LTP induction [1517]. Arc co-sediments with crude F-actin and is found in the postsynaptic density of excitatory synapses [14,18]. Arc is dynamically expressed in principal neurons of many cortical and limbic structures during behavioural training and this expression is necessary for long-term memory in a variety of memory tasks ([20,21], and N. Plath, O. Ohana, B. Dammermann, H. Welzl, R. Waltereit, A. Bick-Sander, D.P. Wolfer, E. Therstappen, H. Husi, M.L. Errington, V. Blanquet, W. Wurst, T.V.P. Bliss, S.G. Grant, H.P. Lipp, M. Bosl and D. Kuhl, personal communication).

Work by Guzowski et al. [20,21] using Arc antisense oligodeoxynucleotides suggested that Arc synthesis may contribute to LTP in the dentate gyrus. This issue has now been examined in detail in a recent study by Messaoudi and co-workers (E. Messaoudi, T. Kanhema, J. Soule, A. Tiron, G. Dagyte, B. da Silva and C.R. Bramham, unpublished work). These authors have found that Arc antisense application at 2 h (but not 4 h) after LTP induction leads to rapid and complete reversal of LTP. LTP reversal is coupled with rapid knockdown of newly synthesized Arc mRNA and protein, dephosphorylation of hyperphosphorylated cofilin and loss of new F-actin at medial perforant path–granule cell synapses. Co-immunoprecipitation experiments indicate that Arc and cofilin are found in the same protein complex. Furthermore, introduction of the F-actin stabilizer jasplakinolide blocks the reversal of LTP. Taken together, these findings strongly support a role for Arc synthesis in regulation of actin polymerization and stable LTP formation (Figure 1). Interestingly, LTP is only transiently inhibited when Arc antisense is given shortly before (5 min) or after (15 min) HFS. This suggests that sustained synthesis of Arc during a protracted time-window is necessary to consolidate LTP. A sequential mechanism is envisioned in which translation of pre-existing Arc mRNA contributes to early LTP expression, while translation of new Arc mRNA mediates consolidation.

Inhibition of endogenous BDNF–TrkB signalling blocks the formation of late phase LTP. Furthermore, simple exogenous application of BDNF induces a protein synthesis-dependent increase in synaptic efficacy termed BDNF-LTP [23,24]. Induction of BDNF-LTP in the dentate gyrus is blocked by inhibitors of RNA synthesis and occluded by prior expression of late phase, but not early phase, LTP [24,25]. BDNF-LTP is also independent of NMDA receptor activation. These results suggest that exogenous BDNF mimics the effects of endogenous BDNF in triggering synaptic consolidation. Like LTP, BDNF-LTP requires ERK (extracellular-signal-regulated kinase) activation and is coupled with up-regulation, dendritic transport and translation of Arc mRNA [24]. Finally, Arc antisense experiments show that Arc synthesis is necessary for the induction and time-dependent consolidation of BDNF-LTP (E. Messaoudi, T. Kanhema, J. Soule, A. Tiron, G. Dagyte, B. da Silva and C.R. Bramham, unpublished work).

A recent microarray-based screen has identified and confirmed a panel of five genes that are co-up-regulated with Arc in dentate granule cells during BDNF-LTP in live rats [26]. Two of these genes (Narp and neuritin) have functions in AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid) receptor clustering and excitatory synaptogenesis. Arc is therefore likely to be part of a co-ordinate process of synaptic strengthening and remodelling involving BDNF-regulated transcription. Although lasting presynaptic actions of BDNF in the adult brain are little understood, BDNF-LTP is associated with sustained increases in evoked glutamate release and activation of CREB (cAMP-response-element-binding protein) in the entorhinal cortex, suggesting retrograde nuclear signalling [27].

BDNF and translation control in dendrites

Protein synthesis occurs in dendrites of many different kinds of principal neurons in adult brain. Synaptic activity may therefore direct spatial and temporal regulation of the translation machinery (mRNA, ribosomes and translation factors) to produce protein synthesis-dependent alterations at individual spines or dendritic regions. Dendritic protein synthesis in hippocampal neurons is regulated by a variety of neurotransmitters, most notably glutamate, dopamine and BDNF [2830]. The BDNF–TrkB system stands out as a potentially autonomous regulator at glutamate synapses. In the CA1 region of adult hippocampal slices, BDNF-LTP is induced in synaptic regions isolated from the pyramidal cell somata [23]. In synaptodendrosome preparations, subcellular fractions containing resealed terminal-spine contacts, BDNF rapidly stimulates translation of several mRNAs coupled with LTP or spine morphogenesis, including α-CaMKII (Ca2+/calmodulin-dependent protein kinase II), Arc, LIMK1 and GluR1 (glutamate receptor 1) [3133].

Recent work has examined the biochemical mechanisms by which BDNF modulates local protein synthesis (Figure 2). Phosphorylation of the eIF4E (eukaryotic initiation factor 4E) is considered the rate-limiting step for translation of most mRNAs (those with a 7-methyl-guanosine ‘cap’ at the 5′-end). Phosphorylation of eIF4E on Ser209 is correlated with enhanced rates of translation, whereas hypophosphorylation is associated with decreased translation. eIF4E is phosphorylated by Mnk1 [MAPK (mitogen-activated protein kinase) integrating kinase 1], whose activity is regulated by ERK and p38 MAPK. The availability of eIF4E is controlled by several binding proteins, most notably 4E-BPs (eIF4E-binding proteins). Signalling through receptor-coupled PI3K (phosphoinositide 3-kinase) and mTOR (mammalian target of rapamycin) leads to phosphorylation of 4E-BP and liberation of eIF4E. BDNF stimulates cap-dependent translation in dendrites through TrkB-coupled PI3K–mTOR and Ras–ERK [3237]. Genetic or pharmacological inhibition of mTOR or ERK impairs LTP maintenance and abolishes BDNF-induced LTP [35,36,38,39]. BDNF-LTP in the dentate gyrus is coupled with transient ERK-dependent phosphorylation of eIF4E and enhanced expression of eIF4E protein in granule cell somata and dendrites [33]. Cap-independent initiation of transcripts at internal ribosomal entry sites could also be important but these mechanisms have yet to be explored in the context of synaptic plasticity.

Figure 2 TrkB and translation control in dendritic spines

The cartoon depicts some of the major signalling pathways coupling TrkB with regulation of eIF4E and eEF2. TrkB activation of PI3K–mTOR and Ras–ERK promotes eIF4E phosphorylation and enhances translation initiation. Phosphorylation of eEF2 stalls ribosomes and arrests peptide chain elongation. BDNF–TrkB signalling has bidirectional effects on eEF2 phosphorylation. In isolated synaptodendrosomes, BDNF treatment has no effect on EF2 phosphorylation state [33].

Protein synthesis in synaptic plasticity is also controlled at the level of peptide chain elongation. eEF2 (eukaryotic elongation factor 2) is a GTP-binding protein that mediates translocation of peptidyl-tRNAs from the A-site to the P-site on the ribosome. Phosphorylation of eEF2 on Thr56 inhibits eEF2–ribosome binding and arrests elongation [40]. LTP is associated with decreases as well as increases in specific protein synthesis. eEF2 phosphorylation observed during LTP may therefore contribute to translation arrest [41]. Paradoxically, however, certain transcripts undergo maintained or enhanced translation under conditions of reduced global protein synthesis and eEF2 phosphorylation. This is the case for Arc and CaMKII mRNA, both of which are critical for stable LTP formation ([41,42], and E. Messaoudi, T. Kanhema, J. Soule, A. Tiron, G. Dagyte, B. da Silva and C.R. Bramham, unpublished work). BDNF-LTP in the dentate gyrus and consolidation of taste memory in neocortex are both associated with transient phosphorylation of eEF2 [33,43]. eEF2 is phosphorylated by eEF2 kinase which itself is subject to tight regulation by calcium/calmodulin, mTOR, ERK and protein kinase A through multiple phosphorylation sites. Recent studies have shown bidirectional effects of BDNF on eEF2 phosphorylation depending on the preparation studied [33,44]. At synapses, eEF2 is phosphorylated in response to NMDA, but not BDNF treatment [33,42]. While eEF2 appears to be important in transcript-specific and compartment-specific translation control in synaptic plasticity, further work is needed to resolve these mechanisms.

Translocation and positioning of the translation machinery are another feature of activity-dependent regulation. RNA storage granules in dendrites discharge mRNA in response to strong depolarization [45,46]. During LTP, ribosomes move from dendritic shafts to spines [47], and pre-existing α-CaMKII mRNA shifts to the synaptodendritic compartment [48]. Finally, BDNF induces a redistribution of eIF4E to an mRNA granule-rich cytoskeletal fraction [49].

A recent study by Schratt et al. [50] has provided evidence for another key player in BDNF-mediated translation control: miRNAs (microRNAs). miRNAs are small, non-coding RNAs that bind to the 3′-untranslated region of target mRNAs and either block translation or bring about transcript degradation. Schratt et al. [50] showed that miR-134, a brain-specific miRNA found in the synaptodendritic compartment, negatively regulates dendritic spine morphogenesis in cultured hippocampal neurons by repressing translation of LIMK1 mRNA. Application of BDNF, which promotes spine morphogenesis, relieves miR-134-mediated repression of LIMK1 translation.

Model of BDNF-controlled synaptic consolidation

In summary, current evidence supports a model for synaptic consolidation triggered by BDNF signalling and involving regulation of gene expression and local mRNA translation (Figure 1). Bursts of excitatory synaptic activity trigger sustained release of BDNF and activation of post-synaptic TrkB receptors. TrkB signalling rapidly activates translation in spines and induces transcription of Arc mRNA in cell bodies. Translation activation in spines consists of global (eIF4E), and probably more mRNA-specific (miRNA), mechanisms. Spines activated in this way may effectively capture and translate mRNAs liberated from local RNA storage granules, as well as newly induced Arc mRNA in transit along dendrites. Sustained translation of Arc is necessary for cofilin phosphorylation, local F-actin expansion and formation of stable LTP. We predict that translation of Arc, and possibly LIMK1, underlies actin polymerization-dependent enlargement of synapses and dendritic spines.

Footnotes

  • Neurotrophins: Mechanisms in Disease and Therapy: Biochemical Society Focused Meeting held at School of Chemistry, Bristol, U.K., 6 April 2006. Organized by S. Allen and D. Dawbarn (Bristol). Edited by D. Dawbarn.

Abbreviations: Arc, activity-regulated cytoskeleton-associated protein; BDNF, brain-derived neurotrophic factor; CaMKII, Ca2+/calmodulin-dependent protein kinase II; eIF4E, eukaryotic initiation factor 4E; 4E-BP, eIF4E-binding protein; eEF2, eukaryotic elongation factor 2; ERK, extracellular-signal-regulated kinase; HFS, high-frequency stimulation; LIMK, LIM domain kinase; LTP, long-term potentiation; MAPK, mitogen-activated protein kinase; miRNA, microRNA; Mnk1, MAPK integrating kinase 1; mTOR, mammalian target of rapamycin; NMDA, N-methyl-D-aspartate; PI3K, phosphoinositide 3-kinase; Trk, tropomyosin receptor kinase

References

View Abstract