New Insights in the Biology of BDNF Synthesis and Release: Implications in CNS Function

Activity-dependent BDNF transcription in the development of cortical inhibition

The Bdnf gene is comprised of at least eight distinct promoters that initiate transcription of multiple distinct mRNA transcripts, each of which contains an alternative 5′ exon spliced to a common 3′ coding exon that contains the entire open reading frame for the BDNF protein (Aid et al., 2007). Through the use of alternative promoters, splicing and polyadenylation sites, at least 18 transcripts can be produced, but remarkably, each encodes an identical initial BDNF protein product. Although the significance of this complex transcriptional organization is cryptic, one hypothesis is that it provides multiple layers of regulation, through alternative promoter usage, differential mRNA stability, or differential subcellular localization of either mRNA or protein.

In the cortex, promoter IV-dependent Bdnf transcription accounts for the majority of the neuronal activity-induced Bdnf expression (Tao et al., 2002; Timmusk et al., 1994). To determine which, if any, of the physiological functions of BDNF in the nervous system are specifically controlled by the neuronal activity-regulated component of its expression, a subtle knock-in mutation was introduced into Bdnf promoter IV and expressed in the cortex (Hong et al., 2008). The resulting animal exhibits disrupted sensory experience-dependent induction of Bdnf expression. Neurons from these animals form fewer inhibitory synapses and have fewer spontaneous inhibitory quantal events (Hong et al., 2008). Significantly, these animals exhibit reduced expression of inhibitory presynaptic markers in the cortex, indicating a specific requirement for activity-dependent Bdnf expression in the development of cortical inhibition. Interestingly, these animals do no exhibit altered survival or differentiation of GABAergic neurons, demonstrating a selective role for activity-dependent Bdnf expression in establishing appropriate connectivity in the CNS (Hong et al., 2008). The lack of other phenotypes associated with reduced BDNF expression, such as hyperphagia and obesity, in these animals highlights the ability to this approach to delineate promoter-specific effects as well as activity-dependent effects on BDNF action.

Regulation of dendritic trafficking of Bdnf mRNA modulates synaptic plasticity and spine development

In addition to utilization of different promoters, Bdnf transcripts are polyadenylated at either of two alternative sites, leading to two populations of mRNAs: those with a short 3′UTR and those with a long 3′UTR (Timmusk et al., 1993). These pools of Bdnf mRNAs are localized to distinct subcellular compartments of neurons, with long 3′UTR Bdnf mRNA enriched in dendrites of cortical neurons when compared with total Bdnf mRNAs. This observation raises the possibility that the long 3′UTR targets one pool of Bdnf mRNA to dendrites. To investigate this hypothesis, an existing mouse strain (Bdnf^flox/flox) in which the long 3′UTR is disrupted was used (Gorski et al., 2003), and indeed few Bdnf mRNAs were present in dendrites of cortical and hippocampal neurons from these animals. These findings indicate that the short 3′UTR Bdnf mRNA is restricted to the soma, whereas the long 3′UTR Bdnf mRNA is also targeted to dendrites for local translation (An et al., 2008).

Numerous mRNA species have been localized to neuronal dendrites, including transcripts encoding cytosolic, cytoskeletal, integral membrane, and secreted proteins (Bramham, 2008). Although local translation of these dendritic mRNAs is required for lasting synaptic plasticity, the in vivo function of local protein synthesis remains largely unknown. Despite of the truncation of the long 3′UTR, Bdnf^flox/flox mice still produce normal amounts of Bdnf mRNAs and protein in the brain (An et al., 2008), permitting the use of this animal model to dissect the in vivo function of dendritic BDNF synthesis. As anticipated, the level of BDNF protein in Bdnf^flox/flox hippocampal neurons was increased in the soma but decreased in dendrites, suggesting that a portion of dendritic BDNF protein is from local translation of dendritic Bdnf mRNA. Interestingly, Bdnf^flox/flox mice exhibited denser and thinner dendritic spines of CA1 pyramidal neurons and reduced hippocampal long-term potentiation (An et al., 2008). As application of BDNF to cultured neurons stimulates spine formation, the spine phenotype in Bdnf^flox/flox mice highlights the effects of locally translated BDNF in dendrites.

Development of spines involves overproduction, growth, and pruning, and spine density increases with age in early development, followed by a reduction to reach levels in mature animals. In humans, spine density increases dramatically during the first few months of postnatal life; however, ∼40% of synapses are later selectively eliminated (Huttenlocher, 1979). Maturation and pruning of dendritic spines is dependent on sensory experience and has been implicated in the activity-dependent refinement of synaptic connections (Churchill et al., 2002). The abnormal spine morphology in Bdnf^flox/flox mice likely results from a deficit in spine pruning, as spine density at 3 weeks of age is normal (An et al., 2008). These animals, however, fail to reduce spine density from 3 weeks to 2 months of age, as do wild-type mice. Since dendritic mRNAs are likely to remain dormant until stimulated, these observations suggest that neuronal activity may promote pruning and growth of dendritic spines by inducing local translation of dendritic Bdnf mRNA. Collectively, these studies suggest that in the Bdnf gene, specific 3′UTRs are used to direct Bdnf transcripts to dendrites, whereby locally translated BDNF may act to promote dendritic spine remodeling, in later postnatal development.

Regulation of BDNF action by altering conversion of proBDNF to mature BDNF

Although the above studies have focused upon mechanisms of regulation of Bdnf mRNA, BDNF actions can also be altered at the protein level. BDNF is initially synthesized as a precursor protein (preproBDNF) in endoplasmic reticulum. Following cleavage of the signal peptide, proBDNF is transported to the Golgi for sorting into either constitutive or regulated secretory vesicles. ProBDNF may be converted into mature BDNF (mBDNF) intracellularly in the trans-Golgi by members of the subtilisin-kexin family of endoproteases such as furin, or in the immature secretory granules by proprotein convertases (Mowla et al., 1999). It has long been thought that only secreted mBDNF is biologically active, and proBDNF is exclusively localized intracellularly, serving as an inactive precursor. However, recent observations of proBDNF secretion and its conversion to mBDNF in vitro by plasmin and matrix metalloproteases suggest that proBDNF may be biologically active (Pang et al., 2004), although the efficiency of intracellular cleavage is controversial (Matsumoto et al., 2008) and may vary among neuronal cell types.

BDNF and the tissue plasminogen activator (tPA, a serine protease) have both been implicated in late-phase long-term potentiation (L-LTP) and long-term memory. In vitro, proBDNF can be converted to mature BDNF by plasmin through the tPA-dependent activation of plasminogen. In electrophysiological studies, mature BDNF, but not uncleavable proBDNF, rescues L-LTP in tPA and plasminogen knock-out mice (Pang et al., 2004). These studies suggest that the conversion of proBDNF to mBDNF promotes L-LTP expression in the hippocampus and provides a mechanistic link between tPA and BDNF, the only two secreted proteins implicated in long-term memory.

Proneurotrophins, including proBDNF, bind and activate the p75^NTR receptor (Teng et al., 2005). To determine if proBDNF alters synaptic activity, effects of recombinant proBDNF were evaluated in wild-type and p75^NTR mutant mice. Perfusion of uncleavable proBDNF significantly enhances LTD, an effect not observed in p75^NTR knock-out animals (Woo et al., 2005). LTD is completely disrupted in juvenile (3–4 week) hippocampal slices from the p75^NTR mutant mice. These results suggest that proBDNF and mature BDNF elicit opposing synaptic effects through activation of two distinct receptors; p75^NTR and Trk receptor tyrosine kinases. Interestingly, complete deletion of the Bdnf gene, which eliminates both proBDNF and mature BDNF, did not impact LTD (Matsumoto et al., 2008), suggesting that complete deletion may obscure detection of potentially opposing phenotypes.

To better understand the mechanisms that regulate the conversion of proBDNF to mBDNF, experiments were designed to uncover physiologically relevant conditions under which the specific isoforms of endogenous BDNF are secreted. Under conditions that induce LTD (low-frequency stimulation), the majority of the secreted BDNF was proBDNF (Nagappan et al., 2009). In contrast, when the neurons were subjected to high-frequency stimulation (HFS—a condition that induces LTP), substantial amounts of secreted mBDNF were detectable. Interestingly, tPA, the protease involved in the extracellular conversion of proBDNF to mBDNF, was secreted only under HFS. Thus, both low- and high-frequency neuronal activities increased proBDNF in the extracellular milieu, but only high-frequency neuronal activity induced tPA secretion resulting in extracellular conversion of proBDNF to mBDNF. These results demonstrate how high-frequency neuronal activity may regulate opposing functions of BDNF isoforms.

The role of the proBDNF/p75^NTR signaling and LTD in adult animals has also been evaluated in vivo. In the uninjured adult brain, p75^NTR is almost exclusively expressed in basal forebrain cholinergic neurons, which project to the hippocampus. Upon exposure to acute stress, LTD can be induced in the adult hippocampus, in contrast to nonstressed adult animals, which do not express LTD. Adult p75^NTR-null animals fail to express LTD in this paradigm, suggesting that p75^NTR plays a central role. Interestingly, the p75^NTR-null mice exhibit anxiety-like behavior and have difficulties coping with stress. Moreover, blockade of LTD by a LTD-specific peptide inhibitor also led to impaired stress coping ability and decreased resiliency. It is possible that acute stress may enhance secretion of proBDNF, which through p75^NTR facilitates LTD in the adult hippocampus. The proBDNF-p75^NTR-LTD may serve as a stress-coping pathway that aids the animal's recovery from stress reactions.

Regulation of intracellular trafficking of endogenous BDNF

Endogenous BDNF levels are extremely low, and specific antibodies are difficult to generate because of the high conservation among mammalian BDNFs. However, new tools, including a monoclonal antibody to the prodomain, and a gene-targeted knock-in mouse expressing epitope-tagged Bdnf, have facilitated quantitative detection of endogenous proBDNF and mature BDNF (Yang et al., 2009). In addition, these reagents permit analysis at the light and ultrastructural level of the intracellular localization of BDNF isoforms. Using these tools, proBDNF is expressed at significant levels at early postnatal stages (<4 weeks), whereas mature BDNF is the prominent isoform in the adult. In the juvenile brain, proBDNF immunoreactivity is diffusely observed in the developing hippocampus, in contrast to adults, in which proBDNF immunoreactivity is largely restricted to the mossy fibers of dentate granule neurons. In parallel, p75^NTR immunoreactivity is most abundant in the juvenile hippocampus and is markedly downregulated by 8 weeks. Collectively, these results suggest that the spatial and temporal expressions of proBDNF and p75^NTR are coordinated (Yang et al., 2009), with higher expression during the perinatal window when axonal outgrowth and synapse maturation is robust and more localized expression during adolescence/adulthood.

These tools also permit localization of BDNF isoforms across development and following injury or stress paradigms to distinct vesicular compartments in axons and dendrites. Several intracellular chaperone proteins, including carboxypeptidase E (Lou et al., 2005), which binds to the mature domain of BDNF, and sortilin, which binds to the prodomain of BDNF (Chen et al., 2005), have been identified and are candidate proteins to potentially regulate intracellular localization of BDNF within neurons. The interactions of sortilin with proBDNF are important in dictating intracellular trafficking, as expression of a dominant-negative form of sortilin lacking a cytoplasmic tail redirects BDNF from the regulated secretory pathway to the constitutive secretory pathway (Chen et al., 2005). This mouse model permits the dissection of the roles of sortilin and other BDNF-chaperone proteins in regulating the intracellular trafficking of proBDNF/BDNF to endogenous axons or dendrites.

Summary

BDNF, the product of a single coding exon, elicits potent effects on neurons to regulate differentiation, synaptic plasticity, and dendritic arborization and furthermore modulates the establishment of neuronal circuits to regulate complex behaviors. Although models of BDNF haploinsufficiency have uncovered diverse cognitive and behavioral phenotypes, more recent studies, including those described above, underscore the multiple levels by which BDNF function can be manipulated across development and in the setting of stress paradigms. The utilization of distinct promoter elements, the selective use of 3′UTRs to direct dendritic trafficking, the regulation of proBDNF to mature BDNF cleavage, and utilization of multiple intracellular chaperones to direct intracellular trafficking provide a diverse repertoire of regulatory mechanisms that permit central neurons to exquisitely and rapidly modulate BDNF expression, localization, and synaptic delivery.