These two elegant studies by Barbara Hempstead, Bai Lu, and colleagues on brain-derived neurotrophic factor (BDNF) clearly demonstrate that both proBDNF and mature BDNF (mBDNF) can be released from neurons in culture. Which form gets released is an important question in basic neuroscience and potentially also in neurodegeneration, because, similarly to proNGF and NGF, pro- and mature BDNF activate different receptors. Their respective signalling consequence in the target neuron can be as different as attenuating or enhancing synaptic transmission, or even promoting cell death or survival. The Yang and Nagappan studies are complementary, but both are in disagreement with previous data of Matsumoto et al. (Matsumoto et al., 2008). This group had reported that cultured hippocampal neurons (derived from BDNF-myc knock-in mice) stored and secreted mBDNF but not proBDNF. However, over the last years, the idea that processing of proBDNF takes place extracellularly following the activity-dependent secretion of proBDNF from neurons has gained...  Read more

These two elegant studies by Barbara Hempstead, Bai Lu, and colleagues on brain-derived neurotrophic factor (BDNF) clearly demonstrate that both proBDNF and mature BDNF (mBDNF) can be released from neurons in culture. Which form gets released is an important question in basic neuroscience and potentially also in neurodegeneration, because, similarly to proNGF and NGF, pro- and mature BDNF activate different receptors. Their respective signalling consequence in the target neuron can be as different as attenuating or enhancing synaptic transmission, or even promoting cell death or survival. The Yang and Nagappan studies are complementary, but both are in disagreement with previous data of Matsumoto et al. (Matsumoto et al., 2008). This group had reported that cultured hippocampal neurons (derived from BDNF-myc knock-in mice) stored and secreted mBDNF but not proBDNF. However, over the last years, the idea that processing of proBDNF takes place extracellularly following the activity-dependent secretion of proBDNF from neurons has gained momentum, and these latest reports strongly support this notion.

To facilitate the detection of proBDNF and mBDNF at extremely low abundance, the authors of Yang et al. generated a knock-in mouse in which the coding exon of the endogenous BDNF gene was replaced by a C-terminal HA-tagged exon (hemagglutinin epitope tag). The authors showed that proBDNF was released from cultured neurons prepared from these mice, both if cultures were devoid of glial cells or in mixed cultures in the presence of a plasmin inhibitor. Both proBDNF and mBDNF release was increased in response to K+ depolarization of neurons. Yang et al. also reported that proBDNF levels as well those of its receptor p75 are developmentally regulated, being high prenatally and declining in adulthood.

In the second study, Nagappan et al. demonstrated that low-frequency stimulation induced predominantly proBDNF secretion from cultured neurons. By contrast, high-frequency stimulation (HFS) preferentially increased extracellular mBDNF and also the secretion of tissue plasminogen activator (tPA), a key protease involved in the extracellular conversion of proBDNF to mBDNF. These authors showed that after HFS, intracellular tPA localizes to dendritic spines and that HFS-induced increase in extracellular mBDNF occurred in dendrites. Thus, the majority of BDNF secreted by the regulated secretory pathway in response to neuronal activation is proBDNF, which is then converted to mBDNF by extracellular cleavage.

One important challenge now is to determine whether proBDNF is also released in vivo, i.e., in the hippocampus, and if its release is more abundant in aged animals or in Alzheimer disease animal models. It should be kept in mind that glial cells represent more than half of total cells in the brain (Pope, 1978) and that glia are able to rapidly process proBDNF to BDNF. Thus, measurements of basal or stimulated proBDNF release in vivo will help the field determine the physiological relevance of these new findings.

In AD patients, both proBDNF and mBDNF are decreased in the parietal cortex and hippocampus even in preclinical stages (Peng et al., 2005; Michalski and Fahnestock, 2003). In contrast, we have reported that proBDNF levels are increased in aged rat hippocampus (Silhol et al., 2007). We also measured BDNF release from hypothalamus of adult rats under stress conditions (Givalois et al., 2004; Arancibia et al., 2007) or from posterior cingular cortex in vivo (Arancibia et al., 2008) using push-pull perfusion experiments.

Therefore, if it can be demonstrated that the release of proBDNF shown in these two papers also occurs in vivo, then peptidase activators could become a promising therapeutic avenue for reducing extracellular concentrations of proBDNF in some pathologies where it can act as an apoptotic agent. Regulation of the expression or activation of specific proteases in particular physiological or pathological conditions (stress, aging, neurodegeneration, inflammation, injury) could therefore be extremely important in determining pro- or antiapoptotic cellular responses. Given the opposite biological effects of proBDNF and mBDNF, as well as the protective effect of BDNF against Aβ-induced toxicity in vitro and in vivo (Arancibia et al., 2008; Tapia-Arancibia et al., 2008), the possibility of modifying the ratio of mBDNF/proBDNF in order to increase cellular mBDNF availability is an important issue. The positive effect of BDNF on memory formation (Barnes and Thomas, 2008), a process impaired in neurodegenerative diseases, further reinforces this point.

Jacobsen et al. (Jacobsen et al., 2008) documented the stimulatory effects on tPA activity of the brain-penetrant small molecule (PAZ-417). It inhibits PAI-1, the endogenous inhibitor of tPA. Because Aβ40 and Aβ42 are substrates for plasmin, PAZ-417 lowered brain and plasma levels of Aβ peptides by increasing their catabolism. Although BDNF or proBDNF were not measured in this study, it is possible that PAZ-417 works doubly, increasing catabolism of Aβ peptides and also increasing proBDNF conversion to BDNF. One finding that suggests this might be the case is that high concentrations (30 or 100 mg/kg) of PAZ-417 decreased only around 20-25 percent of Aβ peptides accumulated in the brain, but it completely reversed LTP and cognitive deficits in transgenic AD mice.

As a number of proteases (neprilysin, insulin-degrading enzyme, endothelin-converting enzyme, and plasmin) have also been implicated in the proteolytic clearance of Aβ in the CNS, the exploration of these enzymatic pathways is also worthy of development.

References:
Pope A. Neuroglia, quantitative aspects. In E. Schoffeniels, G. Franck, L. Hertz and D.B. Tower (Eds). Dynamic Properties of Glial Cells. Pergamon, Oxford, 1978, pp 13-20.

View all comments by Lucia Tapia-Arancibia

These two papers from the Lu and Hempstead labs provide a thorough and interesting set of mechanistic experiments dealing with the longstanding observations that proBDNF is sorted through a regulated secretory pathway and is released primarily without cleavage in an activity-dependent fashion (Mowla et al., 1999). Previous studies by the Hempstead and Lu groups demonstrated that proBDNF induces LTD via the p75NTR receptor (Woo et al., 2005), and that tPA, which activates plasmin for conversion of proBDNF to mature BDNF, is critical for hippocampal late-phase LTP (Pang et al., 2004).

The articles under discussion here use unique reagents, antibodies specific for proBDNF and mature BDNF, and an epitope-tagged BDNF knock-in mouse, to describe how activity regulates the secretion of proBDNF and its processing to the mature form of BDNF. The Nagappan article provides novel information demonstrating that while low- and high-frequency stimulation...  Read more

These two papers from the Lu and Hempstead labs provide a thorough and interesting set of mechanistic experiments dealing with the longstanding observations that proBDNF is sorted through a regulated secretory pathway and is released primarily without cleavage in an activity-dependent fashion (Mowla et al., 1999). Previous studies by the Hempstead and Lu groups demonstrated that proBDNF induces LTD via the p75NTR receptor (Woo et al., 2005), and that tPA, which activates plasmin for conversion of proBDNF to mature BDNF, is critical for hippocampal late-phase LTP (Pang et al., 2004).

The articles under discussion here use unique reagents, antibodies specific for proBDNF and mature BDNF, and an epitope-tagged BDNF knock-in mouse, to describe how activity regulates the secretion of proBDNF and its processing to the mature form of BDNF. The Nagappan article provides novel information demonstrating that while low- and high-frequency stimulation increase secretion of proBDNF, high-frequency stimulation induces tPA secretion resulting in the extracellular processing of proBDNF to mature BDNF. Thus, differential frequencies modulate the balance of proBDNF and BDNF within the extracellular space. It is most interesting that activation of the tPA-related metalloprotease pathway also plays a role in the formation and degradation of NGF (Bruno and Cuello, 2006), suggesting a convergent pathway for the expression and regulation of different members of the NGF family of neurotrophins. However, the lack of detectable mature NGF in normal human brain (Fahnestock et al., 2001) indicates rapid in vivo degradation of mature NGF, which does not appear to be the case for mature BDNF.

BDNF and proBDNF protein levels are reduced in Alzheimer disease (AD) (Michalski and Fahnestock, 2003; Peng et al., 2005). It is therefore relevant to ask how the findings from the Lu and Hempstead labs relate to the disease process. Decreased BDNF transcription (Holsinger et al., 2000) may result in decreased proBDNF synthesis, making less proBDNF available for conversion to BDNF and resulting in attenuated signaling. This may set up a self-perpetuating spiral: reduced BDNF signaling may decrease synaptic potentiation, resulting in less neuronal activation, further reductions in proBDNF and tPA secretion, and reduced BDNF production. Reduced tPA activity in AD and increased PAI-1 are likely to further lessen BDNF levels compared to proBDNF, although this effect appears to be rather small (Peng et al., 2005).

However, this interpretation may be complicated by the observation that the high-affinity TrkB receptor for BDNF is also downregulated in AD (Salehi et al., 1996; Allen et al., 1999; Ferrer et al., 1999), with no change in the pan-neurotrophin receptor, p75NTR (Ginsberg et al., 2006). A decrease in catalytic TrkB receptor without a concomitant change in p75NTR may tip the balance towards decreased LTP and increased LTD. On the other hand, proBDNF is also capable of activating TrkB (Mowla et al., 2001; Fayard et al., 2005). Therefore, the reduced signaling of one or both BDNF isoforms through TrkB may contribute to the AD process and, along with the balance of proBDNF to BDNF, should be included in the equation when considering changes in the cellular milieu due to disease or other biological processes.

View all comments by Margaret Fahnestock

Ever since the discovery of NGF, the quantification of neurotrophins in tissues and body fluids has been a source of difficulty and controversy, in large part due to the very low abundance of these proteins. This is an intrinsic aspect of their physiology that contributes to make these molecules interesting in the first place. What was difficult with NGF many years ago (e.g., Suda et al., 1978) now creates similar problems with BDNF, and published data on its levels in the brain, for example, indicate values that differ by orders of magnitude. The extremely low abundance of BDNF even in the adult brain makes its detection by simple techniques such as Western blot a priori problematic, as it remains very difficult to detect one part in a million (on a wet weight basis) by separating brain lysates by gel electrophoresis without prior purification. This alone has led to considerable confusion in the field as one of the many contaminants detected by BDNF antibodies has a molecular weight very similar to proBDNF. This unknown molecule is...  Read more

Ever since the discovery of NGF, the quantification of neurotrophins in tissues and body fluids has been a source of difficulty and controversy, in large part due to the very low abundance of these proteins. This is an intrinsic aspect of their physiology that contributes to make these molecules interesting in the first place. What was difficult with NGF many years ago (e.g., Suda et al., 1978) now creates similar problems with BDNF, and published data on its levels in the brain, for example, indicate values that differ by orders of magnitude. The extremely low abundance of BDNF even in the adult brain makes its detection by simple techniques such as Western blot a priori problematic, as it remains very difficult to detect one part in a million (on a wet weight basis) by separating brain lysates by gel electrophoresis without prior purification. This alone has led to considerable confusion in the field as one of the many contaminants detected by BDNF antibodies has a molecular weight very similar to proBDNF. This unknown molecule is present at comparable levels in the adult brain of wild-type and mutant mice devoid of BDNF (see Matsumoto et al., 2008, for details). While the necessary controls can be performed with mouse tissues, they are not easy to provide with rat and human tissues.

The situation is even more challenging with NGF. Its precise quantification may be beyond reach at this point as its levels have been estimated to be about 50 times lower than those of BDNF in, for example, mouse hippocampus, both at the protein and mRNA level (see Hofer et al., 1990). The notion that the cleavage of proBDNF would occur extracellularly after its activity-dependent release from neurons needs to be viewed in the context of pulse-chase experiments examining the biosynthesis of endogenous BDNF in neurons. These results indicate that proBDNF is rapidly converted to mature BDNF, in compartments that are inaccessible to antibodies able to capture proBDNF (for details, see Matsumoto et al., 2008). This is perhaps the most direct piece of evidence at this point that, in neurons, proBNDF behaves like many other propeptides and prohormones and is processed intracellularly.

View all comments by Yves Barde