First, the BDNF is translated as a precursor form and then processed by the intracellular/extracellular proteases into mature BDNF (mBDNF). Transformation of proBDNF into mBDNF is a critical step to regulate the mBDNF/TrkB signaling because proBDNF preferentially binds with p75/sortilin receptor but not with TrkB. In addition, contrarily to the positive influence of mBDNF on CNS neurons, proBDNF induces synaptic depression and apoptosis via p75/sortilin signaling
[33,34][33][34]. In hippocampal tissue and CSF from AD patients, the expression of proBDNF, sortilin, and the ratio of proBDNF/BDNF were increased in comparison with healthy controls
[9]. proBDNF in AD patients was highly modified with advanced glycation end products which prevents proteolytic cleavage by proteases. Administration of CSF from AD patients caused apoptosis in the primary culture of hippocampal neurons via p75/sortilin receptor-dependent manner. In addition to post-translational modification in the BDNF protein, the alternation of protease activity also influences the BDNF processing. Proteolytic cleavage of proBDNF into mBDNF was regulated by intracellular and extracellular proteases including mammalian Kexin-like proteases (Furin, PC 1/3/7), tissue plasminogen activator (tPA)/plasmin system (plasminogen, plasmin, tPA, plasminogen activator inihibitor-1 (PAI-1), and alpha (2)-antiplasmin), and matrix metallopeptidases 9 (MMP9)
[33]. Interestingly, tPA/the plasmin system and MMP9 also cleave and degrade Aβ peptide. Therefore, activity of these proteases may impact both Aβ accumulation and mBDNF expression. In the neocortex of AD patients, the levels of plasminogen (pro-type of plasmin) and plasmin were lower than healthy controls
[35]. Cai et al. reported that AD model mice showed a lower plasmin level in the hippocampus
[36]. They also found that a treatment of spinosyn, a flavonoid isolated from Zizyphus jujuba var spinosa seeds, increased expression and activity of hippocampal plasmin and synaptic plasticity in AD model mice. The beneficial effect of spinosyn on synaptic plasticity was blocked by a plasmin inhibitor. Activity of tPA, an endogenous activator of plasmin, is also downregulated in the AD brain in accordance with an increased expression of neuroserpin, an endogenous inhibitor of tPA
[37]. In addition, an upregulation of PAI-1, another endogenous inhibitor of tPA, was also reported in Aβ-treated primary neurons, the hippocampus of AD model mice and frontal cortex of AD patients
[38]. Aβ peptide induced PAI-1 upregulation through the JNK/c-Jun pathways. The PAI-1 inhibitor resulted in improved BDNF maturation and cognitive function without affecting the burden level of amyloid. Further, cognitive stimulation given at a pre-plaque and pre-symptomatic phase protected from cognitive decline in accordance with an increase of PAI-1, decreased activity of tPA, and enhanced production of mBDNF in AD model mice. On the other hand, neuronal overexpression of MMP-9 in AD model mice resulted in increased soluble amyloid precursor protein α(sAPPα) levels, decreased Aβ oligomers, enhanced insulin signaling, and increased mBDNF levels, and prevented the cognitive deficits
[39]. Hu et al. showed that PL402, rhamnoside derivative, suppressed the Aβ level via upregulation of MMP3/9 in cell models
[40]. PL402 also attenuated Aβ pathology and cognitive defects in AD model mice with the consistent increase of MMP3/9. MMP9 also has a role in leukocyte migration from circulation into the brain under inflammatory conditions
[41]. Indeed, knockout of MMP2 and MMP9 showed neuroprotective effects against inflammation in a murine model of experimental autoimmune encephalomyelitis
[42,43][42][43]. Considering that neuroinflammation is one of the major pathologies of AD, it is required to carefully assess the possible impact of a protease enhancer on neuroinflammation. Collectively, enhancement of the BDNF and Aβ processing via proteases is a potential therapeutic strategy in AD, while it is required to take into account side effects.
3.3. Types of TrkB Receptors and AD
There are two type of TrkB receptors: full-length TrkB (TrkB-FL) and truncated TrkB (TrkB-TC); the latter lacks the C-terminal catalytic domain. TrkB-TC negatively regulates BDNF/TrkB signaling because heterodimer of TrkB-TC/TrkB-FL fail to stimulate the downstream signaling. TrkB-TC is generated through two pathways: transcription of
TrkB gene (
NTRK2) isoform and proteolytic processing of TrkB-FL. Santos et al. reported that Aβ treatment selectively increased mRNA levels for
TrkB-TC isoform without affecting
TrkB-FL mRNA levels in rat primary culture
[44]. In addition, they found that Aβ induced calpain-dependent cleavage of TrkB-FL into TrkB-TC
[44]. AD model mice also exhibited an age-dependent relative increase in cortical (but not hippocampal) TrkB-TC receptor levels compared with TrkB-FL
[45]. Of note, overexpression of
TrkB-TC in AD model mice exacerbated their spatial memory impairment while the overexpression of
TrkB-FL alleviated it
[45]. In the postmortem brain of AD patients, TrkB-TC was increased while TrkB-FL was decreased in prefrontal cortex regions
[46]. The researchers also reported that TrkB-FL immunoreactivity was largely decreased in tangle-bearing neurons although increased TrkB-TC immunoreactivity in neurons and reactive astrocytes was confirmed. These studies suggest that an imbalance of TrkB-FL and TrkB-TC contributes to the disturbance of BDNF/TrkB signaling in AD.
3.4. The BDNF Polymorphism and AD
The
BDNF gene carries more than 100 polymorphisms, which may influence BDNF signaling and the pathophysiology of AD
[47]. The single nucleotide polymorphism rs6265, also known as the Val66Met polymorphism, is one of the most characterized polymorphism in relation to BDNF function and risk of brain disease
[47]. This polymorphism leads to valine (Val) to methionine (Met) substitution at position 66 in the prodomain of the
BDNF. In primary hippocampal neurons, the
BDNFMet was inefficiently sorted into secretory granules, resulting in abnormal intracellular trafficking and failed localization in synapses
[48]. As a result, activity-dependent secretion of the
BDNFMet was lower than that of the
BDNFval, which lead to an impairment of synaptic plasticity in the
BDNFMet carrier
[49]. Consistently, in human subjects, the met allele was associated with poorer episodic memory, with abnormal hippocampal activation monitored with fMRI
[50]. The relationship between AD and rs6265 polymorphism was reported from several groups
[51,52,53,54][51][52][53][54]. A preliminary study with a small sample size in Australian populations reported that Met carriers showed a significant and large decline in episodic memory and hippocampal volume but did not show significant changes in the rate of Aβ accumulation
[51]. Another group also demonstrated that Met carriers exhibited a sharp decline in verbal learning/memory and speed/flexibility in a cohort consisting of 89% White, 8% Black, and 2% Hispanic populations
[52]. In Japanese populations, a significant allelic association between rs6265 and AD was found in women, but not men
[53]. On the other hand, study with Chinese Han populations reported no significant allelic association between rs6265 and AD
[54]. These controversial results imply that the risk impact of rs6265 on AD is influenced by ethnic groups and sex.
3.5. Pathogenic Role of the BDNF in AD
Since the BDNF regulates neuronal survival, synaptic plasticity and memory, many research studies focused on the pathological role of BDNF downregulation in AD. In the primary culture of hippocampal neurons, a deprivation of NGF or BDNF induces Aβ production via upregulation of amyloidogenic pathway components, APP, presenilin-1
[55]. It was also reported that BDNF neutralization or knockout leads to increased pro-inflammatory cytokines and activates the JAK2/STAT3/C/EBPβ pathway, resulting in the upregulation of asparagine endopeptidase-mediated APP and Tau cleavage, generating Aβ and neurotoxic Tau N368
[56]. Rohe et al. reported that the BDNF upregulated SOLRA expressions through the ERK signaling pathway and reduced amyloidogenic processing in primary cortical neurons
[7]. BDNF treatment also dephosphorylated tau protein at S202, T205, AT180, and S262 sites in neuronally differentiated P19 mouse embryonic carcinoma cells through the PI3K/Akt signaling pathway
[8]. Although in vitro studies suggest an involvement of the BDNF function in amyloid processing and tau phosphorylation, there are conflicting reports on the role of the BDNF in Aβ and tau pathology in AD mice models. Braun et al. performed conditional depletion of hippocampal BDNF in adult AD model mice
[57]. Interestingly, the BDNF depletion reduced noradrenergic neurons in the locus coeruleus (LC) and noradrenergic projection in the hippocampus, frontal cortex, and molecular layer of the cerebellum. In addition, the number of microglia was decreased while Aβ plaque was increased in the cortex of the BDNF-depleted AD model mice. This study suggested that loss of hippocampal BDNF affects the cortex via decreasing LC projection and leads to increased Aβ plaque due to reduced microglial phagocytosis. The small molecule TrkB agonist, 7,8-DHF, decreased cortical Aβ plaque deposition and protected cortical neurons against reduced dendritic arbor complexity in AD model mice
[58]. On the other hand, Castello et al. demonstrated that both Aβ and tau pathology in AD model mice were not influenced when the mice were crossed with heterozygous
BDNF knockout (+/−) mice
[59]. Heterozygous knockout of
TrkB in the AD model mice also did not show any change in Aβ pathology although memory decline was exacerbated
[60]. The
BDNF overexpression of astrocyte in AD model mice rescued the BDNF signaling activity associated with an improvement of dendritic spine density and morphology, synaptic plasticity, and behavioral memory/cognitive performance without any improvement of Aβ plaque deposition
[61]. The inconsistency on the impact of BDNF signaling on Aβ and tau pathology may be due to different experimental strategies to achieve BDNF downregulation or upregulation. Further studies are required to clarify the BDNF function in AD pathology.
3.6. The BDNF, Dysregulation of the Cholinergic System, and Dementia in Down Syndrome
Decreased function of the central cholinergic nervous system, including lack of acetylcholine neurotransmitter, is also recognized as one of the key mechanisms underlying the cognitive dysfunction in AD patients
[62]. It was demonstrated that typical phenotypes of AD progression including loss of cholinergic neurons and reduced activity of choline acetyltransferase in the nucleus basalis of Meynert are closely related to deficits in memory and the cognitive function of AD because the cholinergic neurons in the region innervate to the hippocampus and neocortex (see the review by Cheng et al. 2021)
[63]. Importantly, basal forebrain cholinergic neurons (BFCNs) depend their synaptic function and cell survival on the BDNF, which is retrogradely released from BFCN targets. Interestingly, using an in vitro aging model with rat BFCNs, Shekari and Fahnestock found impaired axonal transport of the BDNF protein
[64]. They also observed that proNGF was transported in BFCNs and its activity was diminished during in vitro age progression. The aged BFCNs displayed decreased expression of both TrkA and TrkB, although levels of p75 were not changed during in vitro aging, implying the vulnerability of BFCNs in AD may be due to the downregulation of neurotrophin transport. It was reported that Rh2, a rare ginsenoside, has neuroprotective effects including increased activity of choline acetyl transferase (ChAT) against scopolamine-induced memory deficits in mice
[65]. Furthermore, treatment with Rh2 induced the significant upregulation of activated ERK and CREB, and expression of the BDNF in the hippocampus of the memory-deficit model. Recently, the effect of dl-3-n-butylphthalide (NBP) on hippocampal expression levels of ChAT, acetylcholinesterase (AChE), vesicular acetylcholine transporter (VAChT), and the BDNF was examined. A rat model of vascular dementia established by bilateral common carotid artery ligation exhibited decreased spatial learning and memory evaluated with a Morris water maze test, although administration of NBP reversed the decreased learning and memory function
[66]. NBP treatment also increased ChAT, AChE, VAChT, and BDNF expressions, suggesting that central cholinergic dysfunction is involved in vascular dementia pathogenesis and NBP is effective to protect the cholinergic system via activating the BDNF signaling
[66]. Interestingly, the influence of focused ultrasound (FUS)-mediated blood–brain barrier opening in the cholinergic degeneration rat model was reported
[67]. Significant cholinergic degeneration, decreased hippocampal neurogenesis, and deficits in spatial memory function after an administration of 192 IgG-saporin (a cholinergic immunotoxin), were all rescued by FUS-mediated brain–blood barrier opening, with a marked upregulation of the BDNF.
Critical involvement of cholinergic dysregulation in the AD pathogenesis is likely, judging from molecular and behavioral phenotypes of Down syndrome (DS). DS, a genetic developmental disorder due to trisomy 21, exhibits dementia by the third or fourth decade of life such as in early-onset AD. As expected, significant atrophy of BFCNs in DS was also shown
[68]. Previously, using a DS model animal, Ts65Dn mouse, which is recognized as a useful animal model for DS and AD, significant deficits in the Morris water maze performance, cholinergic neuronal degeneration of ChAT-positive neurons, and increased APP protein levels in the hippocampus were demonstrated
[69]. It was also shown that the Ts65Dn mouse exhibited decreased neurogenesis in the hippocampus
[70]. Interestingly, additional choline (approximately 4.5 times than control) supplementing the maternal diet improved the performance of the adult Ts65Dn offspring in a radial arm water maze task with partial normalization of the decreased hippocampal neurogenesis
[70]. Although no therapies for intellectual disability in DS are established, the potential effect of 7,8-DHF on the behavior of the Ts65Dn model is very interesting
[71] as 7,8-DHF, a small molecule agonist for TrkB, was demonstrated to exert beneficial effects in various brain disease models including AD animals. Giacomini et al. found that Ts65Dn mice treated with 7,8-DHF during postnatal days P3-15 did not display any learning and memory improvement (Morris Water Maze) at 1 month later. Ts65Dn mice receiving 7,8-DHF for about 40 days starting from 4 months of age also did not show any improvement in learning and memory. The same group previously reported that the early treatment with 7,8-DHF (for 40 days, from P3 to 45) restored deficits in learning and memory and decreased hippocampal neurogenesis in Ts65Dn mice, suggesting that a timing of treatment with the flavonoid, 7,8-DHF, is very critical
[72]. Considering that an improvement of cholinergic degeneration and resultant decreased hippocampal neurogenesis after the disease onset is very difficult, available combined application (e.g., FUS and 7,8-DHF) using the DS or the other early-onset AD models during development may be beneficial in future studies.