Drug treatment
Currently, the FDA-approved drugs for AD include acetylcholinesterase inhibitors (AChEIs), an NMDAR antagonist, and the IgG1 anti-Aβ monoclonal antibody (aducanumab). Approved AChEIs—including donepezil, galantamine, and rivastigmine, and the approved NMDAR antagonist (memantine) are symptomatic treatments that do not treat the underlying pathological cause of AD. Thus, aducanumab is the first and only disease-modifying drug licensed for AD [
6]. Most (if not all) of the drugs approved for AD treatment are known to influence the level of BDNF. In the following section, we will discuss molecular mechanisms underlying the association between BDNF signaling and drugs that have been approved for the treatment of AD.
AChEIs
A pathological hallmark of AD is that the cholinergic neurons of the basal forebrain are the first to fall prey to neurodegeneration [
250]. AChEIs such as donepezil enhance cholinergic transmission and have been approved for the treatment of AD on the basis that they were found to delay the progression of cognitive decline in clinical trials. Notably, experimental studies have also shown that AChEI administration enhances the cholinergic tone in cholinergic neurons of the basal forebrain in mice, and that these effects are mediated by the activation of Trk receptors [
251]. Similarly, BDNF promotes the survival and differentiation of cholinergic neurons in the same region of the rat brain [
252]. These findings suggest that AChEI administration may have some neuroprotective effects in AD, which is conferred by the activation of neurotrophic signaling. In support, clinical studies have shown that the AChEI donepezil increases the level of CNS BDNF in AD patients [
208].
As the neuroprotective effects of AChEIs are transient at best, a more provocative question is what mechanism governs their regulation of neurotrophin signaling. One possible explanation comes from studies on the effect of AChEIs in other neurodegenerative conditions. Administration of donepezil has been found to protect against vascular dementia by inhibiting the nuclear translocation of histone deacetylase 6 (HDAC6) and the binding of HDAC6 to
BDNF promoter IV, which enhances
BDNF expression [
253]. HDAC6 is upregulated in the cortex and hippocampus of AD patients [
254,
255]. The consequences of HDAC6-BDNF binding have previously been studied in the context of other risk factors for AD. For example, ApoE4 has been shown to promote the nuclear translocation of HDACs in human neurons, resulting in decreased BDNF expression [
255]. Specifically, ApoE4 has been found to induce HDAC6 to bind to
BDNF promoter IV, thereby inhibiting the expression of BDNF. Therefore, these findings suggest that inhibiting the HDAC6-BDNF binding in the cortex could increase BDNF levels and exert neuroprotective effects in AD. Another key question is what BDNF signaling pathways do AChEIs activate. Previous experimental studies have shown that administration of donepezil or galantamine in mice enhances the production of BDNF, thereby suppressing neuronal apoptosis via the activation of PI3K/Akt and ERK pathways and phosphorylation of CREB [
256]. However, it is worth re-stating that the neuroprotective effects of AChEIs do not prevent the progression of AD. Therefore, although these studies suggest that AChEIs can exert neuroprotective effects via enhancing endogenous BDNF levels, more investigations are required.
Antidepressants
Depressive symptoms are common in patients with cognitive impairment. The overall prevalence of depression in AD patients is up to 50% [
257‐
259]. A large-scale longitudinal study has found that the depressive symptoms in AD patients reflect prodromal features of dementia, and dementia is not likely a consequence of long-term depression [
260]. This suggests that the pathological mechanisms may differ from those of depressive symptoms in adulthood–that is, in adults without dementia. Despite these differences, antidepressants are still the only treatment option available for the depressive symptoms in dementia[
261]. In general, the effect of antidepressants on BDNF expression is not well understood. Several studies suggest that antidepressants like the selective serotonin reuptake inhibitor (SSRI) fluoxetine increase BDNF levels and are dependent on normal TrkB signaling to elicit their behavioral effects [
262,
263]. This implies that the therapeutic efficacy of SSRIs may be dependent upon activation of the BDNF/TrkB pathways. However, other studies have reported that certain SSRIs (i.e., fluoxetine, paroxetine, and sertraline) regulate the expression of BDNF mRNA in a dose- and time-dependent manner, such that the acute treatment downregulates BDNF expression, whereas chronic treatment upregulates it [
264,
265]. One possible explanation for this effect is that the bi-phasic shifts in BDNF regulation may be caused by differences in the expression pattern of individual BDNF exons. For example, 4 h after systemic injection of paroxetine, the expression of
BDNF exon IV was found to be selectively downregulated in the rat hippocampus [
266]. In rats, neuronal activity has been shown to induce
BDNF exon IV expression as an immediate-early gene response, meaning
BDNF exon IV mRNA levels can exhibit fast and transient changes, whereas
BDNF exon I levels exhibit slower responses [
266,
267]. Moreover, the therapeutic effects of paroxetine therapy are associated with polymorphism of the
BDNF gene, whereby carriers of the A allele of
BDNF G196A polymorphism respond better to the paroxetine therapy in AD-related depression [
268]. These findings support the notion that the ability of SSRIs to alleviate depression-related symptoms may be mechanistically linked to the BDNF/TrkB signaling. Nonetheless, although antidepressant drugs are a primary therapeutic approach currently used for the treatment of depression in AD patients, several systematic meta-analyses have suggested that SSRIs fair no better than a placebo in their ability to alleviate depressive symptoms in AD [
269‐
272]. Additional high-quality randomized controlled trials with different drug types, dosages, and treatment periods should be conducted to confirm the effectiveness and safety of antidepressants in AD patients.
Estrogens
Estrogen and its receptor-mediated signaling pathways play vital roles in brain function. Both estrogen and BDNF have been shown to exert highly potent effects in the hippocampus, and thus have been explored as potential pathological mediatory and therapeutic targets in psychiatric conditions characterized by memory loss [
273‐
275]. Estradiol (E
2) and BDNF have also been shown to help regulate many of the same biological functions, including modulating the activity of NMDARs (especially the NR
2B subunit), promoting neurogenesis in the dentate gyrus, and facilitating the formation of memories [
276,
277]. It has been reported that estrogen receptor α (ERα) and BDNF are colocalized in CA3 subregion of the developing hippocampus [
278,
279]. LPS-induced sickness behavior in mice shows that the role of BDNF in the response to neuroinflammatory challenge occurs in a sex-dependent manner [
280]. Notably, LTP was found to produce an elevated inflammatory response in the cortex and hippocampus of wild-type males, as well as in BDNF
+/− males. Alternatively, the elevated inflammatory response was found to occur only in BDNF
+/− females (not in wild-type females) and only in the hippocampus. These results either suggest that the BDNF/TrkB signaling may be significantly more sensitive to inflammatory insults in the female hippocampus, or that the basal levels of BDNF are significantly higher in the hippocampus of females than males.
Inherent differences in the role of BDNF as an inflammatory mediator between males and females may arise because the
BDNF gene contains a sequence homologous to the estrogen response element [
281], and the estrogen ligand-receptor complexes can bind to this sequence and rapidly increase BDNF transcription. Additionally, the aromatization of testosterone in male mice leads to high levels of E
2 in the brain [
282]. As a result, the expression of BDNF can still be regulated through estrogen-mediated mechanisms in male mice [
280]. However, the effects of exogenous E
2 treatment on various types of memory, and the estrogen-receptor pathways that are activated, have been shown to differ significantly in the hippocampus of male and female rodents [
283]. Thus, these differences may be more related to the inherent differences in the expression of estrogen-receptors and downstream signaling pathways between males and females than to E
2. In agreement, BDNF may act as a signaling molecule downstream of E
2 to mediate its structural and electrophysiological effects [
284]. E
2 and BDNF have been shown to share several signal transduction pathways and transcription factors, such as AKT, ERK, MAPK, PI3K, Src/Fyn, Ca
2+/calmodulin-dependent protein kinase II (CaMKII) and CREB [
285‐
288]. 17β-estradiol administration induces the phosphorylation of TrkB and the expression of mature BDNF. However, 17β-estradiol activates hippocampal TrkB signaling independently of enhanced mBDNF [
289]. Although many studies have highlighted the benefits of estrogen replacement therapy (ERT) among AD patients [
290‐
293], the impact of ERT on the risk of cognitive decline remains highly contentious [
294,
295].
Cannabinoids
Since the 1990s, the endocannabinoid system has received increasing interest due to its neuroprotective effect, and there is considerable evidence suggesting that targeting the cannabinoid system might be an effective strategy to protect against AD [
296‐
298]. Cannabinoid type 1 (CB1) receptors primarily localize at nerve terminals and regulate excitatory and inhibitory neurotransmission [
299]. In kainic acid (KA)-induced excitotoxicity, inactivation of CB1 receptors can decrease the KA-induced BDNF mRNA levels, indicating that CB1 receptor-mediated neuroprotection may be, at least partially, dependent on BDNF expression [
300]. The CB1 receptor is the main molecular target of endocannabinoids and phytocannabinoids, such as Δ
9-tetrahydrocannabinol, extracted from the
Cannabis sativa plant [
301]. To better understand CB1/BDNF interaction, healthy volunteers were intravenously injected with Δ
9-tetrahydrocannabinol, which increased serum BDNF levels [
302]. One possible explanation is that the CB1 receptor-mediated BDNF expression relies on the activation of the
BDNF gene promoter IV via the PI3K/Akt/mTORC1/BDNF pathway, which is capable of enabling rapid responses to promote BDNF production [
303]. A major drawback of using Δ
9-tetrahydrocannabinol as a therapeutic agent in AD is that it has been shown to produce deficits in cognitive behaviors that are impaired in AD, such as learning and memory [
304]. However, overexpressing BDNF in these regions protects against the cognitive deficits induced by adolescent cannabis exposure in mice [
304]. In turn, BDNF-TrkB-CB1R interactions promote the release of endocannabinoids at cortical excitatory synapses [
305]. Endogenous BDNF also plays a crucial role in cannabinoid-induced neurogenesis in the subventricular zone and hippocampal dentate gyrus [
306]. Although cannabinoids have demonstrated the potential to offer multifaceted protection against AD, further studies are warranted to determine whether chronic administration of cannabinoids can be considered a safe, effective, and low-cost therapy for AD.
Herbal extracts
Herbal extracts have been proposed as an alternative medicine to delay the progression of AD, and some extracts have been shown to work through regulating BDNF. For example, resveratrol (3, 5, 4’-trihydroxy-
trans-stilbene) treatment ameliorates oxidative stress and cognitive deficits in a rat model of vascular dementia by increasing hippocampal BDNF expression [
307]. Chronic administration of curcumin, the main active ingredient in turmeric, alleviates AD-associated cognitive impairments via upregulating BDNF/ERK and Akt/GSK3β signaling in the hippocampus [
308‐
311]. However, as the low bioavailability of curcumin limits its effect in humans, some modified curcumin formulations are being studied. Huperzine A is a novel lycopodium alkaloid extracted from the Chinese herb
Huperzia serrata (Qian Ceng Ta). It belongs to the class of non-competitive AChEIs, and has an antagonistic effect on NMDARs [
312]. Huperzine A improves oxidative glutamate toxicity by activating the BDNF/TrkB-dependent PI3K/Akt/mTOR signaling pathway [
313]. Moreover, oral administration of huperzine A remarkably alleviates the neuronal damage and memory deficits by increasing the expression and levels of BDNF, which it accomplishes by phosphorylating the MAPK/ERK pathway [
314]. However, in a recent phase II clinical trial in individuals with AD, huperzine A (200 μg) failed to demonstrate clinical efficacy [
315]. Other herbs, such as
Ginkgo biloba,
Panax ginseng,
Rehmannia glutinosa Libosch.,
Polygala tenuifolia Willd,
Salvia miltiorrhizae Bunge, and
Ficus erecta Thunb. leaves, have also been investigated for therapeutic efficacy in AD and are considered as potential agents that could endogenously increase BDNF [
316‐
323]. However, clinical evidence supporting the beneficial effect of herbal extracts on BDNF is still lacking.
Lithium and zinc
Lithium or zinc supplementation has been proposed as a novel AD therapeutic strategy due to their modulatory effects on multiple targets, including inflammation, autophagy, oxidative stress and mitochondrial dysfunction [
324‐
327]. Notably, lithium treatment in AD patients has been shown to increase BDNF serum values (~ 30%) and mitigate cognitive impairment [
328]. However, a negative correlation between lithium in drinking water and changes of AD mortality has been reported [
329]. It should be noted that limitations in the experimental design may have caused these conflicting results. While using “microdoses” of lithium in mild cognitive impairment has yielded encouraging results, prolonged exposure and high doses of lithium treatment induce toxicity [
330,
331]. For example, De-Paula et al. stimulated primary cortical and hippocampal neurons with therapeutic (2 mM) and subtherapeutic (0.02 and 0.2 mM) dosages of lithium [
332]. They found that administering low subtherapeutic doses of lithium (0.02 mM) had a more extensive and robust effect on enhancing neuronal BDNF in different brain regions than the higher doses typically considered to be therapeutic. Interestingly, the role of lithium on BBB integrity in rats is dependent on their state of mental health. Whereas lithium treatment repairs the stress-induced BBB hyperpermeability in the hippocampus, it has the opposite effect in normal controls [
333]. This suggests that lithium may interact with BDNF signaling pathways in a context-dependent manner.
Experimental research has shown that zinc interacts with multiple AD-related pathologies, some of which are directly mediated by BDNF. Zinc activates GPR39 metabotropic receptors in the CNS [
334,
335]. GPR39 knockout mice display decreased CREB and BDNF levels in the hippocampus, but not in the frontal cortex [
336]. This suggests that the expression of BDNF and CREB can only be modulated by zinc in certain brain regions. In zinc transporter-3 knockout mice, deficits in learning and memory were observed at 6 months of age, accompanied by decreased levels of TrkB, NMDAR2b, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)2a, BDNF, and pro-BDNF [
337]. Oral supplementation with zinc has been found to reduce Aβ and tau pathology in the hippocampus, ameliorate mitochondrial dysfunction, reduce inflammation, inhibit oxidative stress, and increase BDNF concentration [
338‐
343]. Importantly, zinc gluconate solution can cross the BBB to biosynthesize fluorescent zinc oxide nanoclusters, enabling high spatiotemporal bioimaging [
344]. Therefore, zinc supplementation has the potential to play a dual role in AD treatment, neuroprotection and bioimaging, with the latter function being beneficial for evaluating its own efficacy. Results from nuclear magnetic resonance spectroscopy, light scattering, and cryo-electron microscopy indicate that Zn
2+ binding to the BDNF Met66 prodomain and Val66 prodomain result in different conformational and macroscopic structures [
345]. The substitution of Met66 results in a higher affinity of prodomain to Zn
2+, owing to the His40-mediated stabilization of its multimeric structure. Moreover, the molecular mechanism of zinc deficiency-induced cognitive impairment is associated with hippocampal BDNF DNA methylation [
346]. In brief, this suggests that the upregulation of BDNF may contribute to the neuroprotective effects of lithium or zinc in AD treatment.
BDNF gene delivery
The primary obstacle for
BDNF gene delivery is the selection and optimization of vehicles. Gene-delivery vehicles are mainly divided into two categories: synthetic carriers and recombinant viruses. The former includes polymers and liposomes, and the latter includes AAV, poxvirus, retrovirus, adenovirus, lentivirus and herpes simplex virus [
347,
348]. Each delivery vector has its advantages and disadvantages. Polymer-based vectors used for
BDNF gene delivery include nanoparticles and hydrogels, among others [
349]. Liposomes, which are natural biodegradable lipid bilayers, have great advantage of being similar to natural cell membranes. These nonviral carriers are based on the electrostatic interactions of cationic compounds that spontaneously complex with the BDNF plasmid. Polymer-based vectors exhibit a number of desirable traits, including ease of manufacturing, good safety and stability, low immunogenicity, and simple methods to incorporate target ligands [
350,
351]. Unfortunately, the transfection efficiency of polymers as gene-delivery vectors is several orders of magnitude lower than that of recombinant viruses. Thus, using a recombinant virus is still the primary means for
BDNF gene delivery [
352,
353]. On the downside, viral vectors can induce inflammation and immune responses. Although the systemic immune response induced by systemic injection of viral vectors can be considered harmful in clinical trials, gene therapy of the brain is considered a relatively safe intervention strategy [
354,
355].
BDNF gene delivery exerts protective effects against Aβ- and tau-related pathologies in AD. However, this treatment has no direct action on Aβ deposition and tau hyperphosphorylation. Treating J20 APP transgenic mice with Lenti-BDNF gene delivery for 5 months alleviated learning and memory deficits, ameliorated synaptic degeneration, and reduced atrophy [
126]. However, this BDNF treatment did not change amyloid plaque density. Similarly, P301L mutant tau transgenic mice receiving recombinant human
BDNF gene using an AAV8 vector (AAV-BDNF) showed higher BDNF levels in the brain and improved memory deficits, although the AAV-
BDNF gene delivery had no direct effect on tau protein, GSK3β, and phosphatase PP2A [
189]. On the other hand, BDNF supplementation indeed did successfully alleviate tauopathy-induced memory impairments by inhibiting neuron loss, synaptic degeneration, and impaired neurogenesis [
189].
FDA-approved clinical trials of gene therapies have previously applied the AAV delivery strategy because it can target specific neurons in the brain regions, allowing widespread and stable expression of proteins with the safety of long-term treatment [
356‐
358]. MR-guided infusion of AAV2-BDNF has been used to accurately and consistently target BDNF into the non-human primate entorhinal cortex [
230]. Moreover, real-time MR imaging of AAV in the primate brain has been applied to accurately target intracranial structures and monitor the vector distribution in real-time during injection, thereby ensuring accurate targeting and spread of the vector [
359]. Mutant AAVs have also been studied intensively. Delivery of BDNF using the tyrosine triple mutant AAV (tm-scAAV2) showed that the RNA expression of BDNF was about 300 times higher than that of the AAV group, and produced significantly higher proteins [
360]. These methods enable more effective clinical translation to alleviate neuronal loss and prevent neuronal dysfunction in AD. In February 2021, a first-in-human Phase I clinical trial was launched to assess the safety and efficacy of modified AAV2-BDNF in the treatment of patients with AD or MCI [
361]. The modified method for delivering BDNF will be more conducive for the delivery and distribution of BDNF into the entorhinal cortex and hippocampus.
Another approach for extended delivery of BDNF is the use of cell-based vectors, such as neural stem cells (NSCs), mesenchymal stem cells (MSCs), Schwann cells, CD4 T cells, and fibroblasts [
362‐
365]. Direct
BDNF gene delivery using MSC can overcome BBB blocking [
366]. In previous research, BDNF-transduced bone marrow stromal cells (BMSCs) were transplanted by intravenous injection into irradiated female SJL/J mice for 8 weeks, resulting in a dramatic delay of experimental autoimmune encephalomyelitis onset and a reduction in overall severity [
367]. On the other hand, these BDNF-producing cells only allow prolonged delivery of BDNF. Unfortunately, this method is difficult to be controlled precisely because the delivered BDNF dosages are dependent on cell survival and the stability of transfection. Another concern is that bone marrow-derived cells can migrate and reside in various nonhematopoietic tissues, therefore producing undesired effects. Thus, encapsulation of these BDNF-producing cells has been proposed to achieve continuous and local release. Encapsulated BDNF-producing fibroblasts in alginate-poly-
L-ornithine survived for at least one month after being transplanted into the site of cervical spinal cord injury in rats without immunosuppression [
363]. Transfection of
BDNF gene recombinant MSCs via the adhesive peptide PPFLMLLKGSTR-modified scaffold improved cell survival and BDNF expression [
368]. Alginate-based compositions have also been used to transport NSCs-BDNF and BMSCs-BDNF, maintaining long-term survival and proliferation of cells, as well as controlled release of BDNF [
362]. However, when delivering the
BDNF gene to APP transgenic mice after “disease onset”, no protection against neuronal death was found following a 1.5-month therapeutic period [
88]. This suggests that
BDNF gene delivery might not be a suitable therapeutic strategy for AD at all stages of the disease. As such, both early and long-term treatments may be required.
Physical interventions
Numerous physical interventions have been used to slow down the progression of AD, such as laser therapy, repetitive transcranial magnetic stimulation (rTMS) and exercise [
369‐
372]. Low-level laser treatment has been shown to alleviate Aβ-induced neuronal loss and dendritic atrophy by enhancing BDNF via ERK/CREB pathway activation [
32]. In clinical trials, laser therapy has been successfully applied to treat prostate cancer, lung cancer, and acute pain [
373‐
375]. However, it has not been translated well to AD patients. Novel approaches and more clinical studies are needed to evaluate the efficacy of laser therapy for Alzheimer’s patients. rTMS is a non-invasive therapy for cognitive dysfunction in AD that acts by regulating neuronal excitability [
376]. Different frequencies of rTMS target different brain regions, making it theoretically possible to improve cognitive deficits that are highly localized to a particular brain region [
377]. Additionally, the cognitive benefits of rTMS have been associated with the induction of hippocampal BDNF expression. Low-frequency (1 Hz) rTMS increased hippocampal BDNF and NMDAR expression, and rescued deficits in LTP and spatial memory in an Aβ
1-42-induced toxicity rat model [
378]. While this approach seems promising, changes in BDNF expression following rTMS treatment are difficult to detect in human brain tissues. The role of transcranial direct current stimulation (tDCS) in memory improvement has also been investigated as a possible intervention strategy that could promote the BDNF signaling pathway [
379,
380]. Mice subjected to tDCS stimulation exhibit enhanced acetylation at
Bdnf promoter I that persists for one week, suggesting that remodeling of
Bdnf may mediate the long-lasting effects of tDCS treatment. The action of tDCS varies in Val/Val and Met/Met carriers [
381]. Compared with BDNF
Val/Val mice, BDNF
Met/Met show decreased levels of
BDNF exon IV- and VI-specific transcripts, higher trimethyl-histone-H3-Lys27 binding to
BDNF exon V, VI and VIII promoters, and impaired trafficking of
BDNF VI transcript to CA1 and CA3 regions. Moreover, tDCS promotes synaptic plasticity via activity-dependent BDNF secretion [
382].
Physical exercise, especially aerobic exercise, is beneficial for improving cognitive function. Studies have attributed many of the therapeutic benefits of exercise in AD to its effect on BDNF levels [
383,
384]. Exercise increased the levels of pCREB, CaMKIV and BDNF in the CA1 and dentate gyrus of rats with intracerebroventricular infusion of 250 pmol/day Aβ
1-42 peptides for two weeks [
385]. Four weeks of cardiovascular exercise in mice led to a remarkable increase in BDNF mRNA and protein levels, accompanied by an improved synaptic load in the dentate gyrus region [
386]. Moreover, six months of voluntary physical exercise in 5× FAD mice rescued cognitive deficits by increasing astrocytic BDNF in the hippocampus [
387]. Astrocyte-released BDNF plays a vital role in modifying the morphology and density of dendritic spines through a truncated form of the TrkB (TrkB T1) receptor [
388]. The TrkB T1 receptor specifically localizes at GFAP
+ astrocytes to increase the number of GFAP
+ astrocytes and improve Aβ plaque-associated astrocytic morphology via the BDNF/TrkB signaling pathway [
386]. A ten-week treadmill training program in APP/PS1 mice also restored hippocampal memory and dendritic arbor in the CA1 and CA3 regions via BDNF/TrkB signaling pathways [
389]. For obvious reasons, these results cannot be directly translated to humans. Exercise protocols used in animal studies are significantly different from those used in humans, and how exercise enhances BDNF levels during AD is still unknown. A meta-analysis by da Costa Daniele et al. found that exercise indeed promotes neurogenesis and reduces cerebral Aβ deposition in both healthy and dementia models [
390]. However, evidence on exercise-induced inflammation, oxidative stress, metabolism and insulin sensitivity was scarce. Few studies have compared the beneficial effects among acute exercise, chronic exercise and high-intensity training in AD. It has been demonstrated that aerobic exercise training is associated with increased polyunsaturated free fatty acids, decreased phospholipids, sphingolipids and ceramides, and alterations of gut microbiome metabolites–among which, approximate 30% of these metabolites are correlated with altered BDNF levels [
391]. Thus, more direct evidence should be obtained to confirm how to use exercise to prevent or treat AD.
Regulation of microbiota
A growing body of evidence has suggested that dysregulation of the human microbiome may contribute to the pathogenesis of AD. Poor dental status (i.e., loss of teeth) has been considered an early sign of AD, and irregular tooth brushing is a high risk factor for dementia [
392,
393].
P. gingivalis,
T. forsythia, and
T. denticola have been implicated as the main pathogens responsible for triggering inflammatory responses, and are associated with the pathogenesis of AD [
394]. Gut microbial diversity is altered in AD patients [
395]. Compared with healthy controls, AD individuals’ microbiome show a lower abundance of
Firmicutes and
Actinobacteria, and a higher abundance of
Bacteroidetes at the phylum level. Researchers have also identified 13 genera as potential CSF biomarkers of AD pathology. Among these, increased levels of
Dialister and
SMB53 are associated with less AD pathology. The abundance of
Bacteroides,
Turicibacter and
SMB53 (family
Clostridiaceae) is closely linked with CSF chitinase-3-like protein 1 in AD patients, supporting that the change of intestinal bacterial abundance may be correlated with glial activation in AD.
The BDNF level is closely related to the composition of gut microbiota. Compared to mice with normal gut microbiota, germ-free mice show lower mRNA and protein concentration of BDNF in the hippocampus, amygdala and cortex [
396‐
398]. After transferring fecal microbiota, the levels of cognitive behavior, inflammatory mediators, microglia activity, and BDNF in recipient mice are similar to those of donor mice [
399]. This mechanism is associated with the activation of AKT-GSK3β/β-catenin pathways. These results suggest that the CNS BDNF levels can be significantly disturbed due to the absence of gut microbiota and restored by microbiota transplantation. Furthermore, probiotic supplements are beneficial for up-regulating BDNF levels. VSL#3 is a probiotic mixture composed of 8 Gram-positive bacterial strains. In aged (20–22 months) male rats, VSL#3 treatment increases the abundance of
Actinobacteria and
Bacteroidetes, suppresses microglial activation, and enhances BDNF levels [
400]. How might gut microbiota regulate BDNF levels? Some neurochemicals such as neurotransmitters, butyrate, short-chain fatty acids, and secondary bile acids, can be synthesized and recognized by gut microbiota [
396,
401‐
406]. Accordingly, gut microbiota may influence CNS BDNF function by modulating the activity of these neurochemicals.