Fear extinction and BDNF: translating animal models of PTSD to the clinic


  • R. Andero,

    1. Department of Psychiatry and Behavioral Sciences, Center for Behavioral Neuroscience, Yerkes National Primate Research Center, Emory University School of Medicine, Atlanta, GA, USA
    Search for more papers by this author
  • K.J. Ressler

    Corresponding author
    1. Department of Psychiatry and Behavioral Sciences, Center for Behavioral Neuroscience, Yerkes National Primate Research Center, Emory University School of Medicine, Atlanta, GA, USA
    2. Howard Hughes Medical Institute, Chevy Chase, MD, USA
    Search for more papers by this author

K. J. Ressler, MD, PhD, Department of Psychiatry and Behavioral Sciences, Center for Behavioral Neuroscience, Yerkes Research Center, Emory University, 954 Gatewood Road, Atlanta, GA 30329, USA.
E-mail: kressle@emory.edu


Brain-derived neurotrophic factor (BDNF) is the most studied neurotrophin involved in synaptic plasticity processes that are required for long-term learning and memory. Specifically, BDNF gene expression and activation of its high-affinity tropomyosin-related kinase B (TrkB) receptor are necessary in the amygdala, hippocampus and prefrontal cortex for the formation of emotional memories, including fear memories. Among the psychiatric disorders with altered fear processing, there is post-traumatic stress disorder (PTSD) which is characterized by an inability to extinguish fear memories. Since BDNF appears to enhance extinction of fear, targeting impaired extinction in anxiety disorders such as PTSD via BDNF signalling may be an important and novel way to enhance treatment efficacy. The aim of this review is to provide a translational point of view that stems from findings in the BDNF regulation of synaptic plasticity and fear extinction. In addition, there are different systems that seem to alter fear extinction through BDNF modulation like the endocannabinoid system and the hypothalamic–pituitary adrenal axis. Recent work also finds that the pituitary adenylate cyclase-activating polypeptide and PAC1 receptor, which are upstream of BDNF activation, may be implicated in PTSD. Especially interesting are data that exogenous fear extinction enhancers such as antidepressants, histone deacetylases inhibitors and d-cycloserine, a partial N-methyl d-aspartate agonist, may act through or in concert with the BDNF–TrkB system. Finally, we review studies where recombinant BDNF and a putative TrkB agonist, 7,8-dihydroxyflavone, may enhance extinction of fear. These approaches may lead to novel agents that improve extinction in animal models and eventually humans.

BDNF, synaptic plasticity and long-term potentiation

Neurotrophins are a group of proteins that support different neuronal processes, such as synaptic plasticity and its electrophysiological correlate, long-term potentiation (LTP) (Chao 2003). Synaptic plasticity is the process by which connections between two neurons, or synapses, change in strength. These changes involve structural and functional alterations of synapses which underlie LTP (Minichiello 2009; Xu et al. 2000). The most abundant neurotrophin in the central nervous system (CNS) related to these processes is brain-derived neurotrophic factor (BDNF) which has two receptors, its high-affinity receptor tropomyosin-related kinase B (TrkB) and the low-affinity receptor p75NTR which is a common non-specific receptor for all neurotrophins (Reichardt 2006). BDNF and TrkB are found in the CNS in different areas such as cortex, hypothalamus, hippocampus and amygdala. BDNF is essential for synaptic plasticity processes which are required for long-term learning and memory (Minichiello 2009). Because of the breadth of the current literature, we will mainly consider studies in this review that include LTP and/or fear extinction in humans and rodents where BDNF and TrkB are known or hypothesized to play a key role.

Fear learning and memory, structures involved and PTSD

Among learning and memory processes, fear memories are crucial in some psychiatric disorders like post-traumatic stress disorder (PTSD). PTSD is an anxiety disorder that results in some individuals after repeated or single exposure to an extreme traumatic event directly or by witnessing these situations in others. The traumatic event is often re-experienced following trauma reminders and recurrent nightmares. PTSD is also generally accompanied by hyperarousal symptoms (DSM-IV-TR 2000). Moreover, persistent highly aversive memories related to the trauma and the inability of these memories to be extinguished is a main characteristic of this disorder. The most effective learning-based psychotherapy involves the extinction of these highly aversive memories by the acquisition of an incompatible or inhibitory competing memory (Choi et al. 2010b). The formation of these aversive memories and fear extinction has been studied according to the Pavlovian learning paradigm. This associative learning process consists of the pairing of a neutral conditioned stimulus (CS) with an aversive unconditioned stimulus (US) eliciting a conditioned fear response (see Myers & Davis 2007 for a review). The CS can be a cue (e.g. a tone) or a context (e.g. a room). Examples of conditioned fear responses are increased fear potentiated startle and heart rate. Fear extinction consists of a new inhibitory learning after repeated or prolonged CS presentations, without the US, which causes a gradual decrease in the magnitude and/or frequency of the conditioned response (Choi et al. 2010b; Myers & Davis 2007). However, the fear memory after extinction is not erased, but inhibited, base on its re-appearance in three different phenomena: spontaneous recovery, renewal and reinstatement (Myers & Davis 2007). Spontaneous recovery refers to the CS re-emerging after extinction with the passage of time. Renewal consists in the extinguished CS fear return when the CS is presented outside of the extinction context. Reinstatement denotes unsignaled exposure to the US after extinction which leads to re-appearance of the CS.

Interestingly, the most important cerebral structures involved in both contextual and cue-dependent fear regulation are also relevant in PTSD; those more involved in emotional processes: amygdala, hippocampus and the prefrontal cortex (PFC) (Mahan & Ressler 2012; Myers & Davis 2007). The amygdala is the storage structure for fear memories and the hippocampus is especially important for the processing of contextual information being both necessary for fear acquisition and extinction (Myers & Davis 2007; Sierra-Mercado et al. 2011). The most important substructures related to fear processing in the PFC are the dorsal anterior cingulate cortex (prelimbic cortex, PL, in rodents) and the ventromedial prefrontal cortex (infralimbic cortex, IL, in rodents) (Milad & Quirk 2012).

PL and IL, both substructures of the medial prefrontal cortex (mPFC), appear to have opposing influences on fear processes. Muscimol inactivation suggests that PL is necessary for fear acquisition but not extinction, whereas IL is required for fear extinction but not fear acquisition (Sierra-Mercado et al. 2011).

Fear learning due to externally experienced traumas represents a critical environmentally determined risk for the development of PTSD. However, evidence also suggests that risk for PTSD is in part genetically determined (Cornelis et al. 2010; Koenen et al. 2003,2006; Norrholm & Ressler 2009; Skelton et al. 2012). Thus, it is also important to study genetic variability which may help to explain why some individuals develop the disorder while others do not.

BDNF single-nucleotide polymorphisms, PTSD and fear extinction

The study of single-nucleotide polymorphisms (SNP) related to PTSD is useful to establish genetic risk factors for the onset and development of the disorder and improve its treatment (Mahan & Ressler 2012). Examples of genetic variation related to PTSD risk are FKBP5, serotonin transporter gene variants SLC6A4 and 5-HTTLPR and RGS2 (Amstadter et al. 2009; Binder et al. 2008; Koenen et al. 2009; Mehta et al. 2011; Mercer et al. 2012; Wang et al. 2011). A SNP that has also been suggested to be relevant for PTSD is the Val66Met polymorphism which consists in the substitution of Met for Val at position 66 in the pro-region of BDNF (Frielingsdorf et al. 2010). The Val66Met SNP causes decreased hippocampal volume, deficits in declarative memory and impaired fear extinction (Bueller et al. 2006; Egan et al. 2003; Soliman et al. 2010). Concordantly, studies in transgenic mice containing a knock-in allele of the human Val66Met allele and cell cultures show reduced: neuronal BDNF availability, neuronal survival, hippocampal dendritic arborization, hippocampal volume and LTP which may result in the declarative memory and fear extinction deficits found in these BDNF Met66 knock-in mice (Chen et al. 2004, 2005, 2006; Cao et al. 2007; Frielingsdorf et al. 2010; Ninan et al. 2010; Soliman et al. 2010; Spencer et al. 2010; Yu et al. 2009).

Unfortunately, there are few specific studies addressing the association between PTSD and the BDNF Val66Met polymorphism (Rakofsky et al. 2012). These reports show no effect although they have a small number of participants, probably given the low percentage of homozygous individuals in the population (Lee et al. 2006; Valente et al. 2011; Zhang et al. 2006). More studies are needed to clarify this lack of effect because there are other studies suggesting a relationship between blood BDNF levels and PTSD. Peripheral BDNF blood levels have been found to be altered in PTSD patients (Dell’osso et al. 2009; Hauck et al. 2010) and low BDNF levels predict a higher recovery rate for PTSD symptoms under selective serotonin reuptake inhibitor (SSRI) treatment (Berger et al. 2010). However, the meaning of BDNF in peripheral blood levels is unclear because the relationship between peripheral concentrations and brain function is not known. All these data on BDNF and SNP studies suggest that altered BDNF levels are related to extinction learning deficits. This could be relevant for developing new PTSD biological markers and improving the treatment by enhancing BDNF activity. These studies are in agreement with the findings that BDNF is necessary in the amygdala for fear conditioning acquisition and consolidation of fear extinction (Chhatwal et al. 2006; Rattiner et al. 2004a). Moreover, in vitro administration of recombinant BDNF (rBDNF) or 7,8-dihydroxyflavone (7,8-DHF), a putative TrkB agonist, lowers the threshold of LTP induction in the amygdala (Li et al. 2011). Concordantly, rBDNF administered in the hippocampus in vivo or in slices induces long-lasting enhancement of synaptic plasticity (Kang et al. 1995; Messaoudi et al. 2002; Ying et al. 2002). Also, genetic alterations of BDNF and TrkB in animal models have given interesting insight in the study of synaptic plasticity and fear memories, as discussed below.

Genetic manipulations of BDNF/TrkB in synaptic plasticity and fear extinction

BDNF or TrkB homozygous knockout mice rarely survive beyond the third week of life and if they survive have serious health problems and extreme phenotype abnormalities (Ernfors et al. 1994; Klein et al. 1993). However, BDNF heterozygous knockout mice present a viable, but altered, phenotype including impaired hippocampus-LTP and deficits in the acquisition of contextual fear conditioning, which are both rescued by rBDNF infusion (Korte et al. 1995; Liu et al. 2004; Patterson et al. 1996). On the contrary, transgenic mice overexpressing TrkB receptors have enhanced contextual fear conditioning (Koponen et al. 2004). Together, these knockout mouse lines demonstrate that BDNF and TrkB are necessary for CNS development and BDNF/TrkB play a key role in synaptic plasticity and the formation of fear memories. Since this altered gene expression during the lifespan could present compensations that might occur during development, the generation of conditional knockout mice is more desirable. Deletion of BDNF by injecting Cre recombinase expressing lentivirus into specific brain regions of floxed BDNF transgenic mice has shown that fear extinction effects are regional dependent. While this BDNF deletion in the hippocampus leads to cue-dependent fear extinction deficits (Heldt et al. 2007), no effect is found in extinction of cue-dependent fear when the BDNF deletion is restricted to the PL (Choi et al. 2010a). However, this latter study also shows that BDNF in the PL is necessary for cue-dependent fear acquisition and the deficit in learned fear is rescued by 7,8-DHF, which mimics endogenous BDNF presumably by activating TrkB receptor.

Epigenetic regulation of BDNF in the extinction of fearful memories

The BDNF gene in rodents has at least nine 5′ non-coding exons each containing a unique promoter region and a 3′-coding exon (IX), which codes for the BDNF prepropeptide (see Musumeci & Minichiello 2011 for a review). It has been recently reported that epigenetic regulation of BDNF gene could be crucial in depression (Fuchikami et al. 2011). Specifically, DNA methylation of the CpG island at the promoter 1 of the BDNF gene might be a biological marker of depression although the study is limited by a low number of subjects and needs to be replicated. Animal models of PTSD suggest that epigenetic regulation of the BDNF gene may be also crucial for this disorder. For example, specific exon-containing BDNF mRNAs seem differentially regulated in fear processes in rats depending on the procedure (cue-dependent or contextual conditioning), previous stress, and which brain region is analysed. Note that BDNF exon nomenclature of the papers reviewed below follow the new nomenclature proposed by Aid et al. (2007).

In brief, cue-dependent fear acquisition increases BDNF exon I and IV mRNA in the amygdala (Rattiner et al. 2004b), while re-exposure to the contextual fear conditioning box causes increased exon IV mRNA in the hippocampus (Lubin et al. 2008). Interestingly, contextual fear extinction elicits an increase in BDNF exon I and IV mRNA in the mPFC, although in this study PL and IL were not differentiated (Bredy et al. 2007). A recent study suggests that single prolonged stress causes an increase in the levels of BDNF exon I, IV and IX mRNAs in the hippocampus after contextual fear conditioning when compared to animals with similar fear conditioning exposure but no stress (Takei et al. 2011). These data suggest that BDNF gene transcriptional changes could be modulated by previous exposure to stress which might be relevant for understanding PTSD.

Thus, BDNF gene expression is needed for fear learning processes. In addition, chromatin structure remodelling and dynamic changes in the nucleosomal packaging of DNA appears to have a crucial role in its regulation (Musumeci & Minichiello 2011). Chromatin structural remodelling that includes reversible histone acetylation/deacetylation of the lysine residues in the N-terminal tail are mediated by histone acetyltransferases (HATs) and histone deacetylases (HDACs), respectively. Specific studies of fear processes suggest that histone H3 acetylation and phosphoacetylation levels are enhanced and histone H4 acetylation levels decreased in promoter 4 in the hippocampus after contextual fear conditioning (Lubin et al. 2008; Takei et al. 2011). Again, regional differences might explain an increase in H4 acetylation in the promoter 4 within the PFC after contextual fear extinction (Bredy et al. 2007). Moreover, cue-dependent fear reconsolidation is accompanied by increased levels of acetylated histone H3 but not H4 in the lateral amygdala (LA) (Maddox & Schafe 2011). The use of drugs that act as histone deacetylase inhibitors (HDACi) as agents that may enhance fear extinction is reviewed below.

Fear extinction modulators which act or may act through BDNF

Seeing that BDNF is extremely important in fear learning, and extinction, and is also regulated by these processes, BDNF–TrkB modulation might be clinically relevant. In this section, we will review some of the endogenous and exogenous compounds that modulate fear extinction in both humans and rodents, either directly or indirectly via BDNF/TrkB signalling mechanisms. In particular, we review effective fear extinction enhancers that mediate or may mediate its actions through the BDNF system accompanied by fear extinction training: Antidepressants, HDACi, rBDNF, TrkB agonists and d-cycloserine (DCS; Fig. 1). Moreover, the potential role of the endocannabinoid system and the hypothalamic–pituitary adrenal (HPA) axis as BDNF modulators in fear extinction is also reviewed. Suffice to say, these compounds and their modes of action have implications for PTSD. Pharmacotherapy is important for PTSD treatment because it has been shown that some molecules can enhance extinction learning and memory potentiating the effects of psychotherapies based on fear extinction (Choi et al. 2010b).

Figure 1.

Simplified schematic diagram of BDNF modulators acting at the synaptic level during fear extinction processes. NMDA receptor, HDACi, CB1 receptor, 5-HT receptor and Pac1 receptor activation lead to epigenetic BDNF gene changes which affect BDNF synthesis. At a presynaptic level, BDNF is released from vesicles and its binding of the postsynaptic TrkB receptor leads to autophosphorylation. This causes activation of downstream signaling pathways (PI3K/Akt, MEK/ERK and PKC), leading to gene transcription changes and synaptic plasticity. In addition, PACAP has been suggested to transactivate the TrkB receptor through Src and causing the activation of TrkB downstream signalling. Moreover, BDNF and TrkB modulate glutamate release and NMDA phosphorylation.


The most common pharmacological treatment for PTSD is antidepressants alone and there are few studies combining antidepressants with exposure psychotherapy (Choi et al. 2010b; Hetrick et al. 2010). Results are disappointing in PTSD clinical trials using antidepressants alone and when combined with psychotherapy there is not enough data to support or refute an increased effectiveness (Choi et al. 2010b; Hetrick et al. 2010). There are two rodent studies related to fear processing, suggesting that the antidepressant fluoxetine (SSRI) enhances synaptic plasticity in wild-type mice animals but not in mice with a genetic alteration of the BDNF functioning. The first study shows that fluoxetine enhances synaptic plasticity and fear extinction through BDNF in the amygdala and hippocampus in mice (Karpova et al. 2011). The second one shows how the BDNF Met66 knock-in mice present impaired fluoxetine-induced enhancement of hippocampal synaptic plasticity (Bath et al. 2012). Thus, altered BDNF functioning in PTSD patients, as suggested above, may explain why treatment with currently available antidepressants has not been fully effective in this disorder.

Histone deacetylases inhibitors

Pharmacological HDACi catalytic activity appears to enhance emotional learning processes (Bredy & Barad 2008; Maddox & Schafe 2011). The use of compounds, such as valproic acid (VPA), trichostatin A or sodium butyrate, seems to enhance contextual fear extinction when injected systemically and when infused in the hippocampus (Bredy et al. 2007; Bredy & Barad 2008; Lattal et al. 2007). Nonetheless, trichostatin A enhances cue-dependent fear memory reconsolidation when injected in the LA (Maddox & Schafe 2011). In contrast, in the same study, the DNA methyltransferase inhibitor 5-aza-cytidine (AZA) impairs cue-dependent fear memory reconsolidation. In addition, systemic VPA induces increased mRNA BDNF levels in exon IV within the mPFC and increases histone 4 acetylated levels around the promoter 1 and 4 after fear extinction (Bredy et al. 2007). This is concordant with an in vitro study where VPA is shown to strongly activate promoter 4 of BDNF in cortical neurons (Yasuda et al. 2009). More studies are needed in this direction to elucidate the epigenetic changes elicited by HDACi in structures involved in emotional processing. Several HDACi could be tested to enhance fear extinction in humans since they are approved for the treatment of mood disorders and as anti-epileptics.

Recombinant BDNF and the putative TrkB agonist, 7,8-DHF

rBDNF has been used extensively in rodent studies. For example, an interesting experiment showed that a single rBDNF infusion in the IL of rats enhances cue-dependent extinction in rats, even without extinction training. This BDNF effect is dependent on N-methyl d-aspartate (NMDA) receptors (Peters et al. 2010). In contrast, as mentioned above, BDNF deletion in the PL causes deficits in cue-dependent fear acquisition but not extinction or innate fear (Choi et al. 2010a). Thus, these data support the widely accepted idea explained above that there is a clear dissociation in the role of different mPFC areas in fear learning and memory processes, with IL involved in fear extinction and PL in fear acquisition. Moreover, these two studies show how BDNF is a key modulator in both of them. In contrast to the enhanced extinction found when rBDNF is injected in the IL, rBDNF injected in the dorsal CA1 region of the hippocampus causes impairment in the consolidation of contextual fear extinction (Kirtley et al. 2010). This latter apparently contradictory study and the paper by Heldt et al. (2007), where hippocampal BDNF deletion causes impairment of cue-dependent fear extinction, may suggest that BDNF levels in the hippocampus functions follow an inverted-U shape function. Extremely high and low BDNF levels in the hippocampus may impair its functions in relation to memory processes.

Notably, rBDNF in humans seems to have limited or no therapeutic properties since clinical trials of acute ischaemic stroke or the neurodegenerative disease studies reported lack of effect, probably due to the poor pharmacokinetic properties of BDNF (Wu 2005). An interesting alternative was identified with a potential small molecule TrkB receptor agonist, 7,8-DHF (Jang et al. 2010b). This flavone can be obtained from the plant Godmania aesculifolia or through chemical synthesis. Most recent studies have used the commercially available synthetic version. When systemically administered in mice, 7,8-DHF crosses the blood brain barrier (BBB), provokes TrkB receptor dimerization and autophosphorylation and activation of downstream signalling (Jang et al. 2010b). Moreover, 7,8-DHF presents better pharmacokinetic properties than BDNF and higher TrkB binding affinity. A single systemic dose of 7,8-DHF activates TrkB receptors within the amygdala and enhances both fear acquisition and extinction in naïve mice (Andero et al. 2011). More importantly, in the same study, a systemic 7,8-DHF injection rescues a deficit in fear extinction in a potential mouse model of PTSD. In addition, other recent studies have suggested interesting actions of 7,8-DHF in rodents. 7,8-DHF blocks long-term declarative memory deficits elicited by acute stress immobilization (Andero et al. 2012), prevents depressive symptoms and structural changes caused by a chronic stress procedure (Blugeot et al. 2011) and reverses memory deficits in an Alzheimer mouse model (Devi & Ohno 2012). For all these reasons, 7,8-DHF seems an interesting candidate that ought to be tested for its potential to enhance fear extinction in humans.

Other flavonoid derivatives have been described also as relatively specific TrkB agonists and present antidepressant-like actions in mice, including 4’-dimethylamino-7,8-dihydroxyflavone (Liu et al. in press) and deoxygedunin (Jang et al. 2010a). Moreover, deoxygedunin also enhances cue-dependent fear conditioning learning similarly as 7,8-DHF (Jang et al. 2010a).

LM22A-4 appears to be another potential TrkB agonist, which when chronically administered intranasally in rodents crosses the BBB and activates its downstream signalling (Massa et al. 2010). In addition, treatment with LM22A-4 for 2 weeks restores motor learning in a traumatic brain injury model in rats. We have no knowledge of fear learning experiments performed with this agent. More studies involving potential beneficial activities of this compound in vivo and more detailed pharmacodynamics would be desirable.


BDNF and TrkB are found in glutamatergic neurons at both pre-synaptic and post-synaptic sites (see Minichiello 2009 for a review). Moreover, there are interactions between BDNF/TrkB and glutamate receptors. For example, BDNF increases glutamate release, enhances ionotropic glutamate receptor NMDA subtype activity and increases phosphorylation of NMDA receptor subunits (Minichiello 2009). All these processes may influence synaptic plasticity.

Nonetheless, studies involving NMDA antagonists in rodents impair fear extinction and LTP, while agonists enhance those processes (see Myers et al. 2011 for a review). DCS is an antibiotic that enhances excitatory neurotransmission mediated by NMDA receptors, by binding to the strychnine-insensitive glycine recognition site of the NMDA receptor complex and increasing calcium influx without causing neurotoxicity. Thus, it acts as a NMDA receptor partial agonist and it has been shown to be an effective drug to facilitate consolidation of fear extinction in different anxiety disorders (Guastella et al. 2008; Hofmann et al. 2006; Kushner et al. 2007; Otto et al. 2010; Ressler et al. 2004; Storch et al. 2010; Wilhelm et al. 2008). DCS is especially effective in patients with maladaptive fear and rodents with previous exposure to stress (Myers et al. 2011). Specifically, DCS enhances fear memory extinction when systemically injected and when locally infused into the basolateral amygdala (BLA), mPFC and hippocampus of rats suggesting regional specific effects (Myers et al. 2011). In addition, BDNF Met66 knock-in mice have impaired NMDA receptor-dependent synaptic plasticity (Ninan et al. 2010) and delayed fear extinction which is rescued by a systemic DCS injection (Yu et al. 2009).

At a molecular level, in vitro studies suggest that NMDA receptor activation increases BDNF expression (Marini et al. 1998; Yaka et al. 2003). In particular, NMDA receptor stimulation leads to specific activation of the BDNF promoter 3 (Tabuchi et al. 2000). Moreover, DCS enhancement of fear extinction is dependent on activation of mitogen-activated protein kinase and phosphatidylinositol-3 kinase (PI3K) signalling cascades in the BLA (Yang & Lu 2005). Interestingly, these pathways are also two of the ones by which TrkB receptors exert its neurotrophic actions, including learning and memory processes. So, taking all these data together, there are important bidirectional interactions between BDNF/TrkB and NMDA receptors, which may be key for fear extinction although more studies are needed in this direction to further clarify this issue.

Endocannabinoid system

The cannabinoid type 1 receptor (CB1) is part of the endogenous endocannabinoid system. CB1 has been related to anxiety and required for the extinction of conditioned fear (see Chhatwal & Ressler 2007 for a review). Recent findings have found interesting interactions between the CB1 receptors and BDNF/TrkB. For example, endocannabinoids have been suggested to be released as retrograde messengers induced by post-synaptic TrkB activation (Lemtiri-Chlieh & Levine 2010). Interestingly, BDNF incubation increases the expression of CB1 receptor transcripts and protein (Maison et al. 2009). In addition, BDNF incubation causes an increased sensitivity of the CB1 receptor to endocannabinoid agonists, 2-arachidonylglycerol and noladin, in cerebellar cell culture. Also, BDNF release triggered by CB1 receptor stimulation mediates the neuroprotective effects of cannabinoids in hippocampal cultures (Khaspekov et al. 2004). Although there is limited evidence of in vivo functional interaction of endocannabinoids and BDNF, there is a study where CB1 receptor knockout mice present decreased basal BDNF levels in the hippocampus and rBDNF infusion rescues despair-like behaviour (Aso et al. 2008). In humans, it has been suggested that habitual δ-9 tetrahydrocannabinol users have decreased serum BDNF levels (D’Souza et al. 2009), although this study is limited by a low number of subjects. Taken together, these data suggest that there are interactions between the endocannabinoid and BDNF/TrkB systems which would be important for the modulation of fear extinction.

HPA axis and pituitary adenylate cyclase-activating polypeptide

One of the physiological systems that more strongly alter BDNF is the HPA axis which is activated in response to stressful situations. There are myriad studies in rodents showing how stress alters BDNF and TrkB levels. Although the picture is complex because these stress-induced BDNF/TrkB alterations are regionally dependent and influenced by several factors, such as stress paradigm, stress duration and sex (Greenwood et al. 2007; Lin et al. 2009; Marmigere et al. 2003; Molteni et al. 2001; Molteni et al. 2010; Smith et al. 1995). All these changes in BDNF/TrkB might have in common the later alteration of extinction processing. Corticosterone and adenylate cyclase-activating polypeptide (PACAP) are discussed below as potential modulators of BDNF, fear extinction and PTSD.

There is only one study about the HPA axis–BDNF regulation combined with fear extinction (Gourley et al. 2009). The authors found that 3 weeks of exposure to a high dose of corticosterone in rats leads to impaired fear extinction to contextual freezing. Moreover, cortical BDNF mRNA was decreased in the orbitofrontal cortex but not in mPFC, although this subarea has been extensively related to fear processes. A possible explanation of this lack of effect on the BDNF mRNA levels in the mPFC after chronic stress exposure might be that stress differentially regulates PL and IL BDNF levels. Then, when analysing BDNF in the whole mPFC this effect may be masked. More specific studies are needed for a more deep understanding of the HPA regulation of BDNF and fear extinction processes, a cornerstone of PTSD.

A recent study, also related to stress, shows how PACAP and its specific G protein-coupled PAC1 receptor (PAC1R) are associated with PTSD (Ressler et al. 2011). In addition, experiments with animal models suggest that PACAP and PAC1R in the bed nucleus of the stria terminalis (BNST) and the amygdala are related to fear conditioning and chronic stress (Hammack et al. 2009; Ressler et al. 2011). The BNST is a structure of the HPA axis key for stress and anxiety responses, whereas the amygdala, besides being crucial for fear processes as discussed above, is an important HPA axis modulator. The first study suggest that 2 h after fear conditioning in mice, PAC1R gene (ADCYAP1R1) levels are increased in the amygdala and positively correlated to freezing behaviour (Ressler et al. 2011). The second study shows increased PACAP levels in the dorsal BNST in rats after chronic stress exposure (Hammack et al. 2009).

The potential PACAP effects on BDNF and fear extinction processes come from in vitro and in vivo studies. In vitro studies suggest that hippocampal neurons incubated with PACAP enhance the activity of the NMDA receptor channel which in turn increases BDNF activity, specifically the BDNF transcript IV (Kidane et al. 2008; Yaka et al. 2003). Also, PACAP has been suggested to transactivate the TrkB receptor trough Src kinase activating its downstream signalling, crucial for learning and memory, in the absence of neurotrophins (Minichiello 2009). In addition, there is an in vivo study showing an interaction between PAC1R and BDNF since PAC1R knockout mice present decreased transcriptional regulation of BDNF in the hippocampus maybe trough altered excitatory transmission (Zink et al. 2004). Taking all these data together, it can be hypothesized that deregulation in PACAP and PAC1R affecting BDNF/TrkB might be critical in fear extinction processes although more studies are needed in this direction.


Herein, we have reviewed the known and hypothesized roles of BDNF in the synaptic plasticity processes involved in fear learning and memory, including human and translational genetic and epigenetic studies examining these processes. We then reviewed the results of a number of experimental studies demonstrating that exogenous enhancers of fear extinction may act, at least in part, through BDNF/TrkB processes. The BDNF/TrkB neurotrophin system is clearly critical to the regulation of emotional learning and memory. Further understanding of the pathways mediating BDNF/TrkB action along with potential specific small molecule agonists and antagonists of this system may provide exciting and important future approaches to understanding the neurobiology of fear as well as in treating or preventing fear-related disorders in humans.


The authors would like to thank Nicole Gouws for her help designing the Fig. 1. Support was provided by NIH (F32MH085443, R01DA01962), the Burroughs Wellcome Fund and by the National Center for Research Resources P51RR165 and is currently supported by the Office of Research Infrastructure Programs/OD P51OD11132. Dr. K.J.R. is a founding member of Extinction Pharmaceuticals/Therapade Technologies, which have provided no equity or income, but which exist to develop d-cycloserine for use to augment the effectiveness of psychotherapy.