l-Fucose is a candidate monosaccharide neuromodulator and mitigates Alzheimer’s synaptic deficits

Abstract

Fucosylation, a major glycan modification, has been shown to influence neuronal and microglial mechanisms, but whether unconjugated free l-fucose can affect brain function is unknown. l-Fucose can be transported into cells and metabolized by fucokinase (FCSK) via the poorly understood salvage pathway. Using mouse hippocampal slices, we showed that l-fucose enhanced excitatory neurotransmission and long-term potentiation (LTP) through regulation of presynaptic release. Such effects required l-fucose be metabolized through the FCSK-driven salvage pathway, suggesting a metabolic-signaling mechanism. Human Alzheimer’s disease (AD) and 5xFAD mouse brains showed signs of fucose hypometabolism with impaired l-fucose signaling. Such abnormalities were corrected by exogenous l-fucose, exemplified by rectification of LTP deficits in 5xFAD hippocampus. A dietary l-fucose supplement, which increased cerebral free l-fucose levels and up-regulated FCSK to drive the salvage pathway, mitigated synaptic and behavioral deficits of 5xFAD mice. Our data suggest an unrecognized neuromodulatory function of free l-fucose and reveals its therapeutic potential for AD.

The sugar l-fucose can modulate synaptic function and may ameliorate synaptic deficits in Alzheimer’s disease.

INTRODUCTION

Among monosaccharides, l-fucose is unique in its l-configuration, whereas all other naturally occurring sugars in mammals have the d-configuration, and in its distinct structure lacking the hydroxyl group on the C-6 carbon. Its role in mammalian physiology has been exclusively attributed to fucosylation, its incorporation into the oligosaccharide chains of N- and O-linked glycoproteins or glycolipids that confer functional modifications of macromolecules (1). In the brain, fucosylation was reported to affect processes involved in learning and memory, such as long-term potentiation, neurite outgrowth and migration, and synapse formation, among others (1). Fucosylation of synapsin I was found to regulate the turnover and stability of this key synaptic vesicle membrane protein involved in neurotransmitter release (2). In addition, we reported that core fucosylation regulates mechanisms of microglia activation (3). However, the exact mechanisms of how l-fucose, especially unconjugated free l-fucose, affects brain function remain largely unknown.

Fucosylation in biological systems requires the generation of guanosine 5′-diphosphate–l-fucose (GDP-Fuc), a glycosyltransferase donor, through two pathways. The de novo pathway converts glucose-derived GDP-mannose to GDP-Fuc, and the salvage pathway utilizes free l-fucose to generate GDP-Fuc through an intermediate, fucose-1-phosphate (F-1P), via the rate-limiting enzyme fucokinase (FCSK) (1) (fig. S1). Greater than 90% of cellular GDP-Fuc is derived from the de novo pathway; the salvage pathway has long been considered to play a minor if not negligible role in replenishing cellular macromolecular fucose (4–6). The function of the salvage pathway, particularly in the brain, is poorly understood although it would allow free l-fucose to influence cell functions. Our recent mass spectrometry study showed that differentiated neural cells efficiently incorporated ¹³C-labeled l-fucose intact into glycosylated macromolecules, suggesting an active salvage pathway in the brain (7). In addition, hippocampal neurons released l-fucose in an activity- and store-dependent manner (see below), implying that free l-fucose may be involved in neural cell function via cellular or intercellular signaling and could be physiologically significant.

The current study tests the hypothesis that free l-fucose modulates synaptic plasticity via activating the salvage pathway. We further asked whether such a mechanism is impaired in Alzheimer’s disease (AD), an archetypal disease of synaptic plasticity imposing huge impact on human society. Our findings suggest that l-fucose is a signaling molecule able to regulate synaptic neurotransmission via metabolism through the salvage pathway and up-regulation of FCSK. Furthermore, our data suggest that a fucose hypometabolism state links the pathological drivers of AD to downstream synaptic deficits, which could be remedied by l-fucose supplementation that drives fucose metabolism via the salvage pathway.

RESULTS

l-Fucose improves short-term presynaptic plasticity and long-term potentiation in 5xFAD hippocampal slices

We perfused hippocampal slices obtained from 12-month-old wild-type (WT) and 5xFAD mice with l-fucose (200 μM) and recorded the field excitatory postsynaptic potential (fEPSP) in Schaffer collateral-CA1 synapses. In WT slices, the fEPSP increases were apparent within 1 to 2 min after initiation of perfusion, reached a plateau at ~10 min, and returned to baseline ~5 min after cessation of l-fucose (Fig. 1A). As controls for the specificity of l-fucose, three naturally occurring structurally close monosaccharides, d-fucose, the enantiomer of l-fucose, d-galactose that differs from d-fucose by having a hydroxyl group at C-6, and d-mannose that differs from d-galactose by inverting the stereochemistry of the hydroxyl groups at C-2 and C-4, did not alter fEPSP, suggesting l-fucose being a specific substrate for synaptic facilitation (Fig. 1A and fig. S2). By comparison, fEPSP facilitation in 5xFAD slices was not elicited, even after prolonged perfusion. However, increasing the concentrations of l-fucose appeared to be able to switch on the responses of 5xFAD slices, albeit with substantial delays and subdued rates of increase (Fig. 1B and fig. S2).

Fig. 1. l-Fucose modulates excitatory neurotransmission.

All data are presented as means ± SEM. (A) fEPSP recording in the stratum radiatum of hippocampal slices from WT mice perfused with 200 μM l-fucose or d-fucose in aCSF. Representative traces before (black) and after (red) perfusion is shown. Scatter plot and quantification of maximum fEPSP slope values show that fEPSP slope is enhanced by l-fucose (n = 14) but not d-fucose (n = 7). Black bar on scatter plot show when Fucose was present in recording chamber. ***P < 0.001, Student’s unpaired t test (t = 5.812, df = 18). (B) Stunted responses of 5xFAD hippocampus to l-fucose. Same data from WT in (A) is plotted here for comparison with 5xFAD + l-fucose (200 μM) (n = 6) and 5xFAD + l-fucose (500 μM) (n = 6). *P < 0.05, **P < 0.01, and ***P < 0.001; one-way ANOVA with Tukey’s post hoc test. (C and D) Preincubation with l-fucose for 2 or 5 hours enhances hLTP in WT and 5xFAD hippocampi. Shown are scatter plots of hLTP (C) and quantification of fEPSP slope during the early and late phases of hLTP (D). N = 4 to 5 per group. *P < 0.05, **P < 0.01, and ***P < 0.001 compared to WT; #P < 0.05, ##P < 0.01, and ###P < 0.001 compared to respective aCSF control; one-way ANOVA with Tukey’s post hoc test. (E) l-fucose induces a decrease in paired-pulse ratio (PPR). N = 8; ** P < 0.01; Student’s paired t test (t = 4.986, df = 7). (F) l-fucose increases the frequency, but not the amplitude, of mEPSCs in CA1 pyramidal neurons. Shown are representative tracings and the average amplitude and frequency values of mEPSCs over a 5-min period starting at 10 min of l-fucose perfusion. N = 16, **P < 0.05, Student’s paired t test (t = 3.246, df = 15).

We further tested the effects of l-fucose on hippocampal long-term potentiation (hLTP), a cellular correlate of learning and memory, deficits of which are considered a precursor or substrate of dementia (8). We and others previously showed hLTP deficits in 5xFAD mice in Schaffer collateral-CA1 synapses following high-frequency stimulation (9). l-Fucose, upon 2-hour preincubation, rectified hLTP deficits in 5xFAD hippocampus to approximating the WT level, although appearing not able to sustain the later phase (Fig. 1, C and D). Upon 5-hour preincubation, l-fucose further enhanced hLTP in WT hippocampus, but did not further enhance hLTP in 5xFAD hippocampus. This effect of l-fucose is specific as d-fucose showed no hLTP-enhancing effect (Fig. 1, C and D). Together, both short-term facilitation and LTP studies suggest a deficit in the l-fucose metabolic pathway in 5xFAD brains. The ability of higher concentrations of l-fucose to partially rectify such a deficit suggests that l-fucose or its metabolites could drive forward its metabolic pathway.

To elucidate how l-fucose may facilitate neurotransmission, we conducted paired-pulse facilitation, a form of short-term synaptic plasticity, in WT hippocampal slices. The l-fucose–enhanced evoked EPSP (eEPSP) amplitude was associated with a decrease in the paired-pulse ratio (PPR) (Fig. 1E). Furthermore, l-fucose perfusion increased the frequency, but not the amplitude, of the miniature excitatory postsynaptic current (mEPSC) (Fig. 1F). These results indicate that the l-fucose effects were mediated by changes in release probability of presynaptic glutamatergic synaptic vesicles, and not by postsynaptic mechanisms. To evaluate whether the above defective synaptic responses to l-fucose in 5xFAD mice can be generalized to other AD mouse strains, we conducted the short-term facilitation experiment using 12-month-old APP-PS1 mice (10). Similar to 5xFAD mice, hippocampal slices from APP-PS1 mice showed no response to 200 μM l-fucose but a subdued and delayed response to 500 μM l-fucose (fig. S3).

L-Fucose enhances synaptic plasticity via driving the salvage pathway

Regarding the metabolic mechanisms via which l-fucose enhances synaptic plasticity, the salvage pathway is particularly relevant as it uses free l-fucose derived from dietary sources, added to culture medium/perfusate (as in our experiments), or salvaged from degradation of glycoproteins (fig. S1) (11, 12). We further ruled out the involvement of the de novo pathway as 6-alkynyl-fucose (6-Alk-Fuc), a specific and potent inhibitor of GDP-4-keto-6-deoxymannose 3,5-epimerase-4-reductase (FX) (13) that catalyzes the last step of the de novo pathway, did not affect l-fucose–facilitated fEPSP (fig. S4). We therefore hypothesized that l-fucose affects synaptic plasticity via driving the salvage pathway. To evaluate this hypothesis, we tested the effects of l-fucose in Fcsk^em1(IMPC)Bay mice deficient in FCSK, the gatekeeping enzyme that converts l-fucose to F-1P in the salvage pathway (1) (fig. S1). FCSK knockout (FCSK-KO) abolished the short-term facilitation effect of l-fucose (Fig. 2A) and significantly attenuated hLTP (Fig. 2, B and C). In contrast to WT or 5xFAD mice, the application of l-fucose to the FCSK-KO mice did not augment hLTP, either by 2- or 5-hour preincubation (Fig. 2, B and C). Together, our data indicate that the synaptic effects of l-fucose are mediated by the salvage pathway, requiring the function of FCSK.

Fig. 2. l-Fucose effects depend on FCSK.

All data are presented as means ± SEM. (A) Scatter plot and quantification of maximum fEPSP slope values show that fEPSP slope is enhanced by l-fucose (200 μM) in WT (n = 3), but not in FCSK-KO hippocampus (n = 5). ***P < 0.001, Student’s unpaired t test (t = 6.911, df = 8). Black bar on scatter plot shows when Fucose was present in recording chamber. (B and C) Preincubation with l-fucose does not enhance hLTP in FCSK-KO hippocampi. Shown are scatter plots of hLTP (B) and quantification of fEPSP slope during the early and late phases of hLTP (C). *P < 0.05, **P < 0.01, and ***P < 0.001 compared to WT; #P < 0.05 and ###P < 0.001 compared to respective aCSF control; one-way ANOVA with Tukey’s post hoc test.

l-fucose affects gene sets related to synaptic functions

To comprehensively understand the molecular mechanisms of the effects of l-fucose, we conducted next-generation RNA sequencing (RNA-seq)–based transcriptome analysis on hippocampal slices following 5 hours treatment of l-fucose or control artificial cerebrospinal fluid (ACSF). Differentially expressed genes (DEGs) analysis showed that a total of 797 genes were differentially expressed in l-fucose–treated compared to control group, with 433 up-regulated genes and 364 down-regulated genes. To define transcriptomic changes associated with synaptic transmission after l-fucose incubation, we used SynGO, a comprehensive, evidence-based reference for synaptic gene annotations and ontologies (14). Applying high-stringency parameters to filter annotations by experimental evidence, we identified top biological process terms “process in the synapse,” “process in the postsynapse,” and “process in the presynapse” (fig. S5, A and B, and table S2). Specifically, the highest enrichment scores corresponded to the “synaptic vesicle cycle.” Among the up-regulated genes were genes encoding proteins that are biomarkers for presynaptic vesicle release: vamp1 (vesicle-associated membrane protein 1) and stx1b (syntaxin-1B), as well as genes related to synaptic activity and postsynapse: ncam2 (neural cell adhesion molecular 2), cux2 (cut-like homeobox 2), DLG3 (disks large homolog 3), and grin3b (NMDA receptor subunits 3B) (fig. S5C). Although l-fucose was previously linked to anti-inflammatory activities related to gut microbiota (15), gene ontology (GO) analysis did not reveal significant changes directly associated with inflammatory processes in up- (fig. S5D) and down-regulated (fig. S5E) gene sets. This result suggests that l-fucose metabolism affects expression of genes governing synaptic functions.

Reductions of l-fucose and defective salvage pathway in AD brain tissue

We next tested the hypothesis that the salvage pathway is defective in AD in view of the subdued responses to l-fucose in 5xFAD hippocampus. Quantitative polymerase chain reaction (qPCR) showed that transcripts of FCSK and fucose-1-phosphate guanylyltransferase (FPGT), the two enzymes constituting the salvage pathway (fig. S1), were reduced in 5xFAD (Fig. 3A) and pathologically confirmed human AD brains compared to age-matched controls (Fig. 3B and table S1). In contrast, GDP-mannose 4,6-dehydratase and FX, the two enzymes constituting the de novo pathways, were not significantly different between AD and control samples, except that FX was found increased in human AD brains. Western blotting of FCSK further confirmed reduced FCSK protein levels in both 5xFAD (Fig. 3C) and human AD samples (Fig. 3D). Furthermore, the levels of free l-fucose were significantly reduced in AD samples (Fig. 3E). Notably, in 5xFAD hippocampal slices, the FCSK protein level was enhanced in a time-dependent manner by l-fucose treatment, reaching the WT level following 5-hour treatment, suggesting a positive regulation of FCSK by l-fucose (Fig. 3F). This set of data indicates both a deficiency of available free l-fucose and a defective salvage pathway to metabolize free l-fucose in AD brains. The rescue effects of the applied l-fucose, albeit being subdued, could be due to its ability to supplement the deficient free l-fucose in 5xFAD brains and drive forward the salvage pathway in a limited capacity, in part via increasing the expression of FCSK.

Fig. 3. Abnormal l-fucose metabolism in 5xFAD and human AD brain tissue.

All data are presented as means ± SEM. (A and B) Quantitative PCR shows that transcript levels of the enzymes in the salvage pathway but not in the de novo pathway are decreased in (A) 5xFAD mice (n = 16 per group, ***P < 0.001 Student’s unpaired t test; for FCSK, t = 4.184 and df = 30; for FPGT, t = 3.887 and df = 22), and in (B) human AD brains (n = 20 per group, *P < 0.05 and **P < 0.01 Student’s unpaired t test; for FCSK, t = 3.156 and df = 46; for FPGT, t = 2.411and df = 38; for FX, t = 2.419 and df = 38). (C and D) Western blotting shows that protein levels of FCSK are decreased in (C) 5xFAD mice (n = 5, normalized to actin, *P < 0.05; Student’s unpaired t test, t = 2.696, df = 8) and in (D) middle temporal cortex of human AD brains (n = 6, normalized to α-tubulin, *P < 0.05 Student’s unpaired t test, t = 1.885, df = 12). (E) Free l-fucose levels are decreased in hippocampal slices of 5xFAD mice (n = 8 per group) and middle temporal cortex of human AD brains (n = 13 per group). *P < 0.05 and **P < 0.01, Student’s unpaired t test (for mouse, t = 3.766 and df = 14; for human, t = 2.211 and df = 24). (F) Western blotting shows that 200 μM l-fucose time-dependently increases FCSK levels in 5xFAD hippocampi. Shown are a representative Western blot and quantitative data compiled from five independent experiments; data are normalized to actin. *P < 0.05 and **P < 0.01, one-way ANOVA with Tukey’s post hoc test.

The synaptic effects of l-fucose on hippocampal slices are not likely fucosylation dependent

The dependence of the l-fucose effects on the salvage pathway may support the assumption that l-fucose affects synaptic transmission via fucosylation of glycoconjugates, as GDP-Fuc, the end product of the salvage pathway, is the substrate used by all fucosyltransferases. However, this is not compatible with the time frame of its actions as new glycosylation is a co- and post-translational event that takes long hours (16). As shown above, the short-term effects of l-fucose occurred within 1 to 2 min of application (Fig. 1A), prompting us to question the involvement of fucosylation. Moreover, preincubation with 6-Alk-Fuc, a potent inhibitor of cellular fucosylation (13), did not block the short-term facilitation of l-fucose (fig. S4). To further interrogate this notion, we evaluated changes of fucosylation by lectin blots. Following up to 5 hours incubation of WT or 5xFAD hippocampal slices with l-fucose, we did not find any significant changes of fucosylation detected by lectins Ulex europaeus agglutinin 1 (UEA-1), Aleuria aurantia lectin (AAL), and Lens culinaris agglutinin (LCA) that are specific for fucose linked to the terminal or internal carbohydrate units (fig. S6).

l-Fucose is a candidate endogenous signal molecule/neuromodulator

The observation that free l-fucose is present in both human and mouse brains in moderate abundance (Fig. 3E) and facilitates fEPSP within seconds of application suggests that it may function as a yet unrecognized neuromodulator and, together with its intracellular metabolites, serve signaling functions within or between cells. To further test this possibility, we asked the following questions pertaining to key properties of a neuromodulator: (i) whether l-fucose is released by neurons in an activity-dependent manner, (ii) whether the release can be regulated by intracellular stores of l-fucose, and (iii) whether l-fucose can induce immediate and early signaling changes in neural cells in a short time frame that is clearly not contingent upon fucosylation. Using differentiated primary hippocampal neurons, we found that following KCl-induced depolarization, l-fucose release was increased in a time-dependent manner (Fig. 4A), suggesting that endogenous l-fucose can be released following activity. We next hypothesized that l-fucose–containing glycoconjugates serves as a cellular store from which l-fucose could be released. To modulate cellular l-fucose levels, we preincubated the hippocampal neurons with deoxyfuconojirimycin (DFJ), a potent inhibitor of α-l-fucosidase known to enrich cellular fucose store (17, 18). Incubation with DFJ before KCl-induced depolarization enhanced the release of l-fucose (Fig. 4A). On the other hand, pretreatment with 6-Alk-Fuc for 3 days to deplete the cellular fucose store reduced KCl-induced l-fucose release and also abolished DFJ-induced increase in l-fucose release (Fig. 4A) (13). These results are consistent with the notion that the level of activity-dependent l-fucose release from neurons is related to the cellular fucose store. We next determined if such an enhanced release of endogenous l-fucose would facilitate neurotransmission in hippocampal slices similar to externally applied l-fucose. Preincubation with DFJ indeed induced short-term facilitation in WT hippocampus comparable to external application of l-fucose. As a negative control, the DFJ effect was not observed in 5xFAD slices, consistent with deficient 5xFAD responses to endogenous l-fucose (Fig. 4B). This result implies that cellular l-fucose present in glycoconjugate forms could be released extracellularly or into the synapse by a yet unknown mechanism that does not involve fucosidases. Notably, a classic quantitative radioisotope study in HeLa cells identified a small internal glycoprotein fucose pool that eventually released 75 to 80% of its fucose extracellularly as free fucose (19), supporting our notion that regulated release of fucose from glycoconjugate could be a means of intercellular communication.

Fig. 4. l-Fucose shows activity- and store-dependent release and elicits rapid signaling changes.

All data are presented as means ± SEM. (A) Neuronal depolarization by KCl (300 mM) induces l-fucose release, which is further enhanced by preincubation with DFJ (10 μM) for 10 min before KCl. This increase is abolished in neurons pretreated with 6-Alk-Fuc (100 μM) for three days. Shown are l-fucose levels in the conditioned media of cultured well-differentiated mouse hippocampal neurons. N = 4 to 5 per group. (B) DFJ enhances the slope of fEPSP recorded in the stratum radiatum of WT hippocampal slices (n = 4 to 6) but not that of 5xFAD slices (n = 2 to 5). The scatter plot shows fEPSP slopes in a 10-min baseline recording which was discontinued during the 10-min preincubation with DFJ (10 μM), and resumed for the following 10 min. Bar graphs present average fEPSP slopes in the last 10 min of recording. **P < 0.01, Student’s unpaired t test (t = 3.413, df = 8). (C) l-fucose (but not d-fucose) increases the transcript levels of fos and bdnf within 5 min of treatment in WT hippocampal slices but not in 5xFAD slices. N = 3 to 4 per group; *P < 0.05; one-way ANOVA with Tukey’s post hoc test.

We next conducted experiments over short time frames to measure several key activity-related mediators of neuronal plasticity. Treating WT hippocampal slices with l-fucose rapidly (within 5 min) increased the RNA levels of the c-fos and bdnf, two neuronal activity-induced immediate early genes implicated in synaptic plasticity (20, 21) (Fig. 4C). This transcriptional effect is specific as c-Jun or Jun-B was not activated (fig. S7). In contrast, d-fucose treatment of WT slices or l-fucose treatment of 5xFAD slices failed to induce c-fos or bdnf expression (Fig. 4C). Because our data indicate that l-fucose enhances presynaptic release, we next examined activation of Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) and phosphorylation of its substrate synapsin I, key events leading to presynaptic release (22). l-Fucose induced phosphorylation of both CaMII and synapsin I, which was apparent within 5 min of application (fig. S8, A to C). Other signaling molecules that were quickly activated by l-fucose include CREB, but not ERK (fig. S8D and E). In contrast, 5xFAD hippocampus showed significantly subdued responses (Fig. 4C and fig. S8). Together, we showed that l-fucose is endogenously released in an activity- and store-dependent manner, and rapidly activates signaling pathways to modulate synaptic plasticity, making it a candidate neuromodulator.

Dietary supplement of l-fucose mitigates AD-like synaptic deficits in 5xFAD mice

Our results suggest that the administration of l-fucose could supplement brain l-fucose to drive the fucose salvage pathway in 5xFAD mice and mitigate AD-like defects in synaptic plasticity and cognition. To test this possibility, we allowed 11-month-old 5xFAD and WT littermate mice to consume 0.5% l-fucose–supplemented diet (FD) or control diet (CD, same diet composition without l-fucose supplement) ad lib for 30 days. Mice receiving FD were well groomed and displayed no observable differences, including body weight and food consumption habit, compared to those receiving CD (fig. S9). l-Fucose supplement significantly increased the level of free l-fucose in the 5xFAD brains to a level comparable to WT (Fig. 5A). It also improved their exploratory behavior, evidenced by increased traveling distance in the open field (Fig. 5B), and mitigated their memory deficits, evidenced by improved performance in novel object recognition (Fig. 5C) and T-maze tests (Fig. 5D). Furthermore, we tested the abnormal sensorimotor gating mechanisms noted in prodromal stages of AD (23) using acoustic response tests. These tests examine the response to sudden, loud noises, as measured by acoustic startle response (ASR) and prepulse inhibition (PPI). We found that CD-fed 5xFAD mice had reduced magnitude and prolonged latency of ASR to loud stimuli (120 dB) compared to CD- or FD-fed WT mice (Fig. 5E). While WT animals showed inhibition of the startle response with soft but increasingly loud prepulse tones, the inhibition was significantly reduced in CD-fed 5xFAD mice (Fig. 5F). This PPI impairment suggests inability to filter out unnecessary information due to abnormal sensorimotor gating mainly from entorhinal lesions (24, 25), previously demonstrated in humans with early stage of AD (26). FD consumption significantly alleviated the ASR and PPI deficits of 5xFAD mice (Fig. 5, E and F).

Fig. 5. l-Fucose supplement mitigates behavioral and synaptic deficits of 5xFAD mice.

Eleven-month-old WT and 5xFAD mice were assigned to FD or CD for 30 days. All data are presented as means ± SEM. (A) FD increases the l-fucose levels in 5xFAD forebrain to the WT levels. N = 10 per group. Two-way ANOVA with Tukey’s post hoc test shows a significant difference in diet treatment [F(1,36) = 14.60, P < 0.001] and genotype [F(1,36) = 8.672, P = 0.006]. (B to F) FD improves exploratory, memory, and sensorimotor gating deficits of 5xFAD mice: (B) distance traveled in cm in open field test; (C) percentage of discrimination index in novel object recognition test {two-way ANOVA with Tukey’s post hoc test shows a significant difference in genotype [F(1,44) = 4.823, P = 0.033]}; (D) percentage of alternation in T-maze test {two-way ANOVA with Tukey’s post hoc test shows a significant difference in genotype [F(1,45) = 7.491, P = 0.009]}; (E) mean startle response and latency (in ms) in response to a 120 dB startle stimulus {two-way ANOVA with Tukey’s post hoc test shows a significant difference in genotype in the mean startle response [F(1,44) = 8.465, P = 0.006] and in latency [F(1,44) = 6.705, P = 0.013]}; and (F) percentage of PPI using a 70 dB prepulse followed by a 120 dB startle stimulus. N = 11 for WT-CD, 12 for 5xFAD-CD, 13 for WT-FD, and 12 for 5xFAD-FD {two-way ANOVA with Tukey’s post hoc test shows a significant difference in diet treatment [F(1,44) = 6.563, P = 0.014] and genotype [F(1,44) = 22.26, P < 0.001]}. (G) Western blotting shows that FD increases the FCSK protein levels in 5xFAD forebrain to the levels of WT and further increases its levels in WT mice. N = 6 per group. *P < 0.05, **P < 0.01, and ***P < 0.001, two-way ANOVA with Tukey’s post hoc test.

Corresponding to the memory improvement, dietary supplementation of l-fucose also enhanced hLTP in 5xFAD mice, particularly in the late phase (fig. S10). This effect of l-fucose in hLTP appeared to require FCSK. First, FCSK-KO mice fed CD showed substantially reduced magnitude of hLTP compared to WT mice in both induction and maintenance. Second, such stunted hLTP was not improved by FD consumption, indicating that FD-dependent hLTP enhancement was mediated by the FCSK-catalyzed salvage pathway (fig. S11). In congruence with in vitro data, FD consumption by 5xFAD mice also resulted in increased levels of FCSK (Fig. 5G), supporting the notion that administered l-fucose could supplement the deficient free l-fucose in 5xFAD brains and drive the salvage pathway to mitigate AD-like synaptic deficits. We further showed that FD consumption also rectified several key molecules in synaptic plasticity, such as activation of synapsin I, CaMKII, and CREB (fig. S12). Such effects occurred in the absence of detectable up-regulation of fucosylation, analyzed by lectin blots (fig. S13A). Flow cytometry analyses using fucosylation-specific lectins as well as an antibody to Lewis X antigens [CD15, a series of carbohydrate epitopes with terminal fucosylation and abundantly expressed in neural tissue (27, 28)] further supported this assessment (fig. S13, B to D), indicating that FD did not significantly increase fucosylation of brain macromolecules and supporting the notion that the overall change in fucosylation does not correlate with the synaptic effects of l-fucose. This finding is also consistent with the notion that free l-fucose increases its fractional contribution to fucosylation by suppressing the contribution of the de novo pathway, thus maintaining a tightly regulated cellular level of GDP-Fuc and consequent fucosylation (29, 30). We further found that the effects of FD were not related to a reduction of AD-like amyloid pathology as it did not reduce the levels of cerebral amyloid deposition or Aβ peptides (of both soluble and insoluble forms) in 5xFAD mice (fig. S14), supporting a direct action of l-fucose on synaptic plasticity.

We conducted RNA-seq analysis using hippocampal tissue from three groups of mice: WT/CD, 5xFAD/CD, and 5xFAD/FD. SynGO enrichment analysis revealed that 100% of the 154 curated synaptic genes were down-regulated in in 5xFAD/CD mice compared to WT/CD mice. The main categories affected were “organization of synapse” (54 genes), “presynapse,” (37 genes) and “signaling” (39 genes). (Fig. 6, A and B). 5xFAD/FD showed an opposite trend of 5xFAD/CD showing up-regulations of 70% (109 of 154) of the down-regulated genes in 5xFAD/CD mice (Fig. 6, B and C, and tables S3 and S4), indicating a reinstitution of the WT synaptic gene profile. Among the highly up-regulated genes by the FD, lrrk2 has been shown to regulate synaptic vesicle endocytosis and neurotransmitter release (31, 32). Several genes, such as tiam1, raph1, and rims3, are involved in the regulation of the actin cytoskeleton and dendritic spine morphology. Other genes, like cdk5r1, cdk5r2, and camk4, are involved in intracellular signaling pathways that modulate synaptic transmission and plasticity (33, 34). In addition, some genes, such as chrm2, chrna6, chrna7, gabra4, gabrb2, gabrd, and gabrg3, are directly involved in the regulation of neurotransmitter receptors and ion channels crucial for synaptic transmission. Deficits in synaptic proteins such as syngr3, syt12, and syt2 can have significant effects on synaptic function and neuronal communication. These proteins are involved in various stages of synaptic vesicle trafficking, neurotransmitter release, and synaptic plasticity. Furthermore, GO analysis found that protein phosphorylation pathway, an important process for activation of synaptic protein, was increased in 5xFAD/FD compared to 5xFAD/CD (Fig. 6D), further supporting that the FD improves synaptic function in 5xFAD mice. This analysis also revealed up-regulations of gene sets related to anti-inflammatory processes such as “positive regulation of inflammatory response” and “autophagy” and down-regulations of gene sets related to pro-inflammatory processes such as “positive regulation of interleukin-6 production” and “positive regulation of tumor necrosis factor production” (Fig. 6, D and E), suggesting an anti-inflammatory effect of FD, which may contribute to improved synaptic function.

Fig. 6. RNA-seq analysis reveal l-fucose supplement affects genes set related to synaptic function.

(A) Sunburst plot showing gene counts for respective synaptic genes annotated against SynGO Biological Process term in 5xFAD/CD versus WT/CD. (B) Volcano plot of synaptic genes in SynGO analysis between 5xFAD/CD versus WT/CD (top) and 5xFAD/FD versus 5xFAD/CD (bottom). Red dots represent down-regulated genes in 5xFAD/CD versus WT/CD and green dots represent up-regulated genes in 5xFAD/FD versus 5xFAD/CD. Black dots represent genes not significantly different between the 5xFAD/FD and 5xFAD/CD groups. Genes are considered significant with a −log(P value) > 1.3. The names of the top five DEGs are listed next to the dots. (C) Heatmap of top 20 significant DEGs related to synaptic function from SynGO between WT, 5xFAD/CD, and 5xFAD/FD. Each column represents an independent biological replica obtained from a mouse. (D and E) Bubble plot of up-regulated (D) and down-regulated (E) biological process in GO terms in 5xFAD/FD versus 5xFAD/CD.

To corroborate the above anti-inflammatory activity of l-fucose, we quantified the immunoreactivities of glial fibrillary acidic protein (GFAP; an astrocytic marker) and TMEM119 (a microglial marker) in hippocampal sections and found that FD alleviated the enhanced GFAP and TMEM119 immunostaining in 5xFAD mice to levels comparable to WT mice (Fig. 7). We further costained for c-Fos with GFAP, TMEM119, and NeuN and found that a majority of c-Fos–positive cells are NeuN-positive neurons. FD substantially increased the immunoreactivity of c-Fos in both WT and 5xFAD mice (Fig. 7, A to C). Together, we showed that FD enhanced synaptic functions in 5xFAD mice associated with increased c-Fos expression and reduced astro- and microgliosis.

Fig. 7. FD consumption increases c-Fos expression in neurons and decreases astrogliosis and microgliosis.

(A to C) Co-immunofluorescence staining of brain section including the hippocampal region of CA1 with antibody against c-Fos and NeuN (A), GFAP (B), and TMEM119 (C). Scale bar, 50 μm. (D) Bar graph showing mean fluorescence intensity of GFAP. N = 4 per group. Two-way ANOVA with Tukey’s post hoc test shows a significant difference in genotype [F(1,12) = 11.58, P = 0.05]. (E) Bar graph showing mean fluorescence intensity of TMEM119. N = 4 per group. Two-way ANOVA with Tukey’s post hoc test shows a significant difference in genotype [F(1,12) = 15.63, P = 0.02]. (F) Bar graph showing mean fluorescence intensity of c-Fos. N = 4 per group. Two-way ANOVA with Tukey’s post hoc test shows a significant difference in diet treatment [F(1,12) = 29.55, P < 0.001].

DISCUSSION

We identified a signaling function of l-fucose that is not likely dependent on fucosylation of macromolecules. We further showed that this function depends on its metabolism through FCSK, suggesting a metabolic signaling mechanism mediated by the understudied and poorly understood salvage pathway. On the basis of its rapid release by neurons in an activity- and store-dependent manner and its rapid induction of signaling changes enabling presynaptic glutamate release, l-fucose is qualified as a candidate neuromodulator, the first described in the chemical category of sugars.

Prior reports from two groups may have implicated the signaling function of l-fucose. Matthies et al. (35, 36) in their pioneering but limited studies showed that exogenously applied l-fucose–enhanced hippocampal LTP in vitro, which was replicated by us using both WT and 5xFAD mouse models. They speculated that l-fucose interfered with the interactions between endogenous lectins and synaptic receptors or ion channels, many of which are glycoproteins. Recently Hong et al. (37) showed that exogenously applied l-fucose rapidly activated Wnt signaling in Chinese hamster ovary cells. However, the study only used free l-fucose as a competitive probe to support a putative receptor that would bind fucosylated LRP-6 (a Wnt coreceptor) and regulate Wnt complex endocytosis. The possible signaling function of free l-fucose was not stipulated. Notably, neither group described l-fucose effects as physiological, and both attributed the effects to “competitive interference” of glycoprotein function. Although the competitive interference model remains possible, it is not compatible with our observation that the effects of l-fucose depend on its intracellular salvage pathway metabolism. Furthermore, l-fucose is rapidly released by neurons in an activity- and store-dependent manner and rapidly induces signaling pathways to affect synaptic plasticity, suggesting it being an endogenous physiological neuromodulator. Our findings call for further studies of its mechanisms of action. One may consider that l-fucose, like many neuromodulators, acts as a ligand recognized by cellular receptors or sensors. It is known that l-fucose can serve as an energy substrate for the microbiota, which have developed well-regulated l-fucose sensing or receptor systems that control their growth and virulence (38). So far, there is no evidence supporting the existence of such l-fucose receptors or sensors in mammalian cells. Alignment studies for fucose permease in Campylobacter jejuni (39) and FusK and FusR in Escherichia coli (40), all well-studied bacterial fucose sensors, did not yield homologous genes or proteins in mammalian cells.

On the other hand, a cell surface l-fucose transporter that internalizes l-fucose by facilitated diffusion was partially purified from the mouse brain (41, 42). Recently, Ng et al. (30) identified GLUT-1 as a l-fucose transporter, which also transports d-glucose in a noncompetitive manner. They also provided evidence that the cell can distinguish the origin of l-fucose and segregate those sources into multiple nonhomogenous pools of GDP-Fuc (29, 30). Considering the requirement of FCSK for l-fucose effects in our study, we speculate that the uptake of exogenous l-fucose initiates a pathway into a unique pool of GDP-Fuc, which may activate a metabolic signaling mechanism, similar to the action of glucose. Glucose, following intracellular transport, is rapidly phosphorylated by hexokinases into glucose-6-phosphate (G-6P) (fig. S1). Hexokinases are rate-liming enzymes of glucose metabolism, while its downstream metabolites, including G-6P, are signaling molecules (43, 44). In a parallel fashion, dietary or cell-released l-fucose can be rapidly internalized and phosphorylated by the rate-liming enzyme FCSK into F-1P, which we speculate could be a signaling metabolite. However, how F-1P and perhaps its metabolites lead to downstream signaling (to the effects of facilitation of excitatory neurotransmission and regulation of genes governing synaptic functions) remains to be studied. The ability of l-fucose to quickly up-regulate FCSK is supportive of this notion.

Emerging evidence supports that synaptic hypometabolism, manifested by imaging evidence of reduced glucose uptake, precedes synaptic dysfunction and imposes a risk of future cognitive decline and AD (45). While most attention has been focused on glucose, our data show signs suggesting fucose hypometabolism in AD. We found that free l-fucose level and its salvage pathway enzymes were significantly reduced in human AD and mouse 5xFAD brains, a condition that limits efficient usage of l-fucose. We further found that increasing cellular fucose store by fucosidase inhibition resulted in enhanced activity-induced l-fucose release and more robust fEPSP (Fig. 4), suggesting that releasable and functional l-fucose could be drawn from internal macromolecular fucose. As macromolecular fucose is largely derived from glucose via the de novo pathway (fig. S1), it is possible that abnormal glucose metabolism in AD may indirectly limit the availability of free l-fucose and the capacity of activity-dependent l-fucose release, particularly at the synapse, contributing to fucose hypometabolism and synaptic dysfunction.

The present study is limited by our inability to quantify F-1P and GDP-Fuc. Such methods are currently not available, hampering the understanding of the downstream pathways of FCSK and the identification of the key signaling metabolites mediating l-fucose effects. We have attempted to develop enzymatic assays for F-1P and GDP-Fuc and this effort is still in progress. We also failed to detect them by mass spectrometry; it could be due to their levels being below our limit of detection or due to an ion suppression effect from other molecules such as lipids that are more apt to ionize in negative ion mode. Future work is needed to meet these technical challenges. More work is also warranted to identify the pool or pools of l-fucose that provide the released l-fucose upon depolarization. Our data suggest that the released l-fucose is not from the pool released by α-l-fucosidase activity as its inhibitor DFJ actually enhanced the depolarization-induced release. Rather, it could arise from an internal glycoprotein fucose pool that was identified as a major supplier of extracellular free fucose (19), through an unknown mechanism.

In this study, the synaptic effects of l-fucose were shown in short time frames (minutes to hours in in vitro hippocampal slice studies of synaptic facilitation and hLTP) as well as in a longer time frame (1 month in the in vivo studies of FD effects). Both settings demonstrated synaptic enhancement; however, it is not clear whether the long-term effects were built upon the cumulative short-term effects or whether they have the same or divergent mechanisms. l-Fucose appears to have dual impacts on the synapse: short-term signaling independent of fucosylation as described here and long-term modifications of glycoproteins via fucosylation that affect turnover of synaptic proteins, neurite outgrowth, and neuronal architecture, as described by Murrey et al. (2). Both modes of action appear to converge on presynaptic mechanisms involving synapsin I; its phosphorylation can be quickly induced by free l-fucose, and its fucosylation in the form of fucose-α(1-2)-galactose protects it from degradation and thereby preserves neuronal architecture (2). It is conceivable that the long-term effects of l-fucose likely involve alterations of cellular fucose pools and changes of fucosylated glycans on many macromolecules in addition to synapsin I. Oral fucose supplementation normalized fucosylation of proteins through the salvage pathway within 4 weeks in a patient with congenital disorders of glycosylation (CDG) harboring biallelic variants in GFUS (encoding GDP-l-fucose synthase, abbreviated as FX in fig. S1) (46). It is therefore warranted to investigate whether the FD results in a similar fucosylation-normalizing effect in AD mouse models. It is likely that in addition to the short-term synaptic modulatory mechanism which is likely fucosylation-independent, there are multiple sequential mechanisms, and some of the mechanisms may rely on fucosylation of selected proteins. Our observation that FD-treated mice did not show overall changes in bulk macromolecule fucosylation does not exclude the possibility of altered fucosylation in specific proteins or the possibility of rapidly turned-over fucosylated proteins mediating l-fucose effects. These possibilities need to be investigated by more sensitive methods such as MS-based glycoproteomics analysis. In addition, the 1-month treatment with FD may have effects on non-neuronal cells that indirectly affect the synapse. For example, we showed that the FD mitigated neuroinflammation in 5xFAD hippocampi, evidenced by up-regulations of gene sets related to anti-inflammatory processes and reductions of astrogliosis and microgliosis. Such actions would have non–cell autonomous effects on synapses. Furthermore, gut l-fucose may change gut microbiota to generate metabolites that affect brain functions and neuroimmune interactions (47, 48). Thus, the long-term effects of l-fucose on the synapse may involve broader and more diverse mechanisms compared to the short-term effects. This notion is exemplified by the observation that, in 5xFAD tissue, l-fucose up-regulated c-Fos and some proteins regulating presynaptic release in the long-term setting (Fig. 7F and fig. S12) but was not able to do so in the short-term setting (Fig. 4C and fig. S8). It may also explain some discordances in the DEGs identified by RNA-seq between the two settings (fig. S5 and Fig. 6). Furthermore, while both settings require FCSK metabolism of l-fucose, the l-fucose–induced up-regulation of FCSK in 5xFAD samples occurred after 2-hour treatment (Fig. 3F), which would not contribute to the short-term facilitation that occurred within minutes, but would provide a mechanism of FCSK recovery, further driving hLTP enhancement and other longer-term effects.

Although there are few sources of bioavailable l-fucose in the human diet (30), exogenous l-fucose in free or conjugated forms could have health benefits (49). Fucoidans, which are polysaccharides rich in seaweeds and characterized by high l-fucose and sulfate contents, are suggested to have therapeutic potentials in multiple diseases, including AD (50, 51). Our previous study found that in human milk oligosaccharides (HMOs), l-fucose is the abundant component, a major distinguishing feature between human and bovine milk (52). The superior health benefits of HMOs have been attributed in part to their ability to regulate the gut microbiota (53). While the microbiota-gut-brain axis could provide an indirect mechanism of how l-fucose could affect brain function, a more direct mechanism would be that free l-fucose, in the diet or generated from catabolism of fucosylated glycans (in HMOs or fucoidans) by intestinal microflora and subsequently up-taken by the colon and passing the blood-brain barrier (4, 54), could directly affect the brain. Our FD study lends support to this mechanism as FD consumption remedied the low levels of cerebral free l-fucose in 5xFAD mice, enhanced hLTP, and improved performance in tasks enabled by multiple functional synaptic systems. Furthermore, parallel to up-regulation of FCSK by l-fucose incubation of hippocampal slices, FD consumption enhanced the cerebral FCSK level to drive the salvage pathway. Our data lead to a therapeutic hypothesis that oral consumption of l-fucose can improve synaptic function in AD via activating the salvage pathway and mitigating fucose hypometabolism. Because l-fucose consumption is likely safe and easy to comply with, it is a candidate therapy that can be readily tested in human clinical trials.

MATERIALS AND METHODS

Animal studies

Tg6799 5xFAD [B6SJl-Tg(APPSxFILon,PSEN1*M146L*L286V)6799Vas/Mmjax] mice on the C57Bl/6 background were originally obtained from R. Vassar, Ph.D., Northwestern University (55). APP-PS1 [B6.Cg-Tg(APPswe,PSEN1 d E9)85Dbo/Mmjax] mice on the C57Bl/6 background were originally purchased from the Jackson Laboratory. WT littermates were used as WT controls. Fcsk ^em1(IMPC)Bay mice were obtained from A. Beaudet, Ph.D., Baylor College of Medicine, and generated for the Knockout Mouse Production and Phenotyping Project (KOMP2). The strain, C57BL/6N-Fcsk^em1(IMPC)Bay/Mmmh, is available from the MMRRC, RRID:MMRRC_068158-MU. Phenotypic details can be found on the International Mouse Phenotyping Consortium site at www.mousephenotype.org/data/genes/MGI:1916071. Mice were randomly assigned to treatment groups and phenotyping parameters, including neurobehavioral assessments, were evaluated in a blinded manner. Roughly equal numbers of male and female mice were used. All protocols involving mouse models were approved by the Institutional Animal Care and Use Committee (IACUC study protocol no. 22220).

Human brain samples

All tissues were collected and provided by the University of California Davis Alzheimer’s Disease Research Center (P30 AG072972). Written informed consent, including consent for autopsy, was obtained from study participants or, for those with substantial cognitive impairment, a caregiver, legal guardian or other proxy. Study protocols were reviewed and approved by the Institutional Review Board (IRB). For postmortem diagnosis, we followed the National Institute on Aging-Alzheimer’s Association guideline for the neuropathologic assessment of AD. The brain tissues used for the current study were snap-frozen during autopsy and was stored in −80°C before use. Patient characteristics are summarized in table S1.

Chemicals and designed diets

l-Fucose, d-fucose, d-galactose, and d-mannose were purchased from Sigma-Aldrich (Sigma, St. Louis, MO). DFJ was purchased from Sigma-Aldrich. 6-Alkynyl-l-fucose was purchased from Fisher Scientific (Thermo Fisher Scientific, Waltham, MA). For diet treatment, CD (TD. 140368) and 0.5% l-fucose supplemental diet [TD. 200724, which has the same composition as the CD with the addition of l-fucose (5 g/kg) and red food color (0.1 g/kg) for identification] were designed by a certified Teklad nutritionist at Envigo (Indianapolis, IN).

Brain slice preparation

Brain slices containing the hippocampus were obtained as previously described (56). The mice were anesthetized with isoflurane and decapitated. The brain was rapidly removed and submerged in ice-cold high-sucrose artificial cerebrospinal fluid containing: 220 mM sucrose, 2 mM KCl, 0.2 mM CaCl₂, 6 mM MgSO₄, 26 mM NaHCO₃, 1.3 mM NaH₂PO₄, and 10 mM d-glucose (pH 7.4, set by aeration with 95% O₂/5%CO₂). Coronal brain sections (300 μm thick) containing the dorsal hippocampus were cut in ice-cold modified ACSF using a DTK-1000 D.S.K Microslicer (Ted Pella Inc., Redding, CA, USA). Slices were then immediately transferred to a nylon net submerged in normal ACSF containing: 126 mM NaCl, 3 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 26 mM NaHCO₃, 1.25 mM NaH₂PO₄, and 10 mM d-glucose (pH 7.4, set by aeration with 95% O₂/5% CO₂), for at least 40 min at 35°C controlled temperature. After subsequent incubation for at least 1 hour at room temperature, hemi-slices were transferred to a recording chamber, submerged in normal ACSF with a constant flow of ~2 ml/min.

Extracellular recording of fEPSPs, pair pulse facilitation, and long-term potentiation

Field EPSPs were evoked after stimulation of the Schaffer collateral afferents by a bipolar stimulating electrode (WPI, Sarasota, FL), and recorded from the stratum radiatum of the CA1 region using a borosilicate capillaries filled with 3 M NaCl (resistance, 1 to 3 megohms). Baseline stimulation rate was 0.05 Hz. fEPSPs were filtered at 2 kHz and digitized at 10 kHz with a Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA). Data were collected and analyzed with pClamp 10.3 software (Molecular Devices). Slope values of fEPSPs were obtained for quantitation of the responses. Paired-pulse facilitation (PPF) of synaptic transmission was induced by electrical stimulation of Schaffer collaterals with paired pulses at an interstimulus interval of 50 ms. The PPF was quantified as the ratio of the second fEPSP slope over that of the first response. For LTP, after 10 min of stable baseline recording of fEPSPs evoked every 20 s at a constant current pulse of 0.2 to 0.4 mA with a duration of 60 μs, high-frequency stimulation, consisting of two trains of 100-Hz (1 s) stimulation with the same intensity, was applied. Recording was then continued for 60 min with stimulation of fEPSPs every 20 s.

Whole-cell patch-clamp recordings

Whole-cell patch-clamp recording was conducted as previously described (57). To record mEPSCs, patch pipettes (5 to 8 milliohms) were filled with 270 mOsm internal solution containing the following: 120 mM K-gluconate, 10 mM KOH, 9 mM KCl, 3.48 mM MgCl2, 4 mM NaCl, and 10 mM HEPES. AMPA receptor currents were recorded from cells voltage clamped at −70 mV, in bath solution containing 10 μM bicuculline (to block GABAA receptors) and 1 μM tetrodotoxin (to block presynaptic action potentials). Bath application of 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide disodium salt (10 μM) was used as a control to confirm that currents were attributable to AMPA receptor activity. Analysis of mEPSCs was performed manually using Mini analysis software (Synaptosoft Inc., version 6.0.2) with a noise amplitude threshold of 2 pA.

l-Fucose quantification

Measurement of l-fucose levels was conducted using the l-Fucose Assay kit from Megazyme (Megazyme, Wicklow, Ireland). This kit quantifies the level of l-fucose via this reaction catalyzed by l-fucose dehydrogenase (l-FDH): l-Fucose + NADP⁺ → l-fucono-1,5-lactone + NADPH + H⁺. For brain homogenates, samples were deproteinized by using the Deproteinizing Samples Preparation Kit (BioVision, Milpitas, CA). For conditioned media from hippocampal neuronal cultures, aliquots were incubated at 95°C for 10 min to inactivate enzyme activity. After centrifugation, supernatant was use for the assay. Thirty microliter of samples were added to a 96-well plate with distilled water, solution I (buffer), and solution II (NADP+), provided by the manufacturer. A full calibration curve was used for this assay. Absorbances of the solution were read at 340 nm after approximately 4 min and the reaction was started by the addition of 5 μl of l-FDH solution. Absorbance was read again for 10 min in kinetic mode.

Tissue homogenate preparation and Western blotting

Brain sections containing hippocampus were homogenized in lysis buffer (150 mM NaCl, 50 mM tris-HCl, 2 mM EDTA, 1% Triton X-100, 0.5% SDS, and 0.1% Na-deoxycholate) with protease inhibitor cocktail and phosphatase inhibitor (Sigma-Aldrich). Equivalent amounts of protein were analyzed by 4 to 20% tris-glycine gel electrophoresis (Invitrogen). Proteins were transferred to polyvinylidene difluoride membranes and probed with antibodies. Visualization was enabled using enhanced chemiluminescence (GE Healthcare Pharmacia). The following primary antibodies (dilutions) were used: anti-FUK (1:1000, Invitrogen), anti–phospho-synapsin I (1:1000, Cell Signaling), anti–synapsin I (1:1000, Cell Signaling), anti–phospho-CaMKII (1:1000, Cell Signaling), anti-CaMKII (1:1000, Cell Signaling), anti–phospho-CREB (1:1000, Cell Signaling), ant-CREB (1:1000, Cell Signaling), and actin (1:5000, Cell Signaling). Secondary antibodies were horseradish peroxidase–conjugated anti-rabbit or anti-mouse antibody (1:1000, Cell Signaling).

Quantitative real-time PCR

Total RNA from brain sections containing hippocampus were extracted using RNeasy Plus Universal Mini Kit (Qiagen, Valencia, CA) according to manufacturer’s protocol. cDNA was synthesized using iScript Reverse Transcription Supermix (Bio-Rad, Hercules, CA). RNA purity and concentrations were assessed by measuring the absorbance at 260 nm, and 280 nm through a NanoDrop 2000C Spectrophotometer (Thermo Scientific, Waltham, MA). qPCR was performed using the Sso Fast EvaGreen Supermix (Bio-Rad) in the CFX96 Touch Real-Time PCR Detection System (Bio-Rad). Mouse and human primers were purchased from Bio-Rad. Gene expression was normalized to an endogenous reference gene, β-actin. Data were analyzed by the 2^−ΔΔCt method. All experiments were performed in duplicate.

Primary hippocampal neuron culture

Primary hippocampal neuron cultures derived from newborn C57BL/6J mice were prepared as previously described (58). Cells were plated on poly-d-lysine–coated six-well plate at a density of 5 × 10⁵ cells per well in serum-free Neurobasal medium plus B27 supplement and maintained at 37°C in 5% CO₂. Cultured neurons were depolarized using 300 mM KCl. Conditioned media were collected at 0, 1, 10, 30, and 60 min postdepolarization for the quantification of free l-fucose level. In some experiments, neurons were preincubated with 100 μM 6-alkynyl-l-fucose for 3 day and/or 10 μM DFJ for 10 min for inhibition of α-l-fucosidase before KCl depolarization. The neurons remained viable all through the experiments.

RNA-seq and analysis

For RNA-seq from incubated slices, total RNA was extracted from brain slices using RNeasy Plus Universal Mini Kit (Qiagen). RNA quality was assessed on the basis of RNA integrity number (RIN). For RNA-seq conducted on brain section from mice on FD, total RNA was extracted from frozen sections using the RNeasy Plus Universal Mini Kit (Qiagen). RNA quality was assessed on the basis of RNA RIN. RNA library and paired-end RNA sequencing was performed using NovaSeq X Plus system (Illumina) with 6 Gb throughout by NovoGene (Sacramento, CA, USA). All RNA-seq reads were aligned to the mouse reference genome (mm10) using Hisat2 version 2.1.0. FeatureCounts version 1.4.6 was used to quantify reads counts. DEGs were analyzed using DESeq2 package using the open source, web-based platform Galaxy (usegalaxy.org). To control false-positive rates, we applied the Benjamini-Hochberg false discovery rate approach. To investigate genes involved in synaptic function and process we analyzed DEGS using SynGo (https://syngoportal.org/). GO pathway enrichment analysis was performed using DAVID Bioinformatics Resources.

Immunofluorescence staining

Cryostat brain sections (16 μm) were fixed with 4% paraformaldehyde and blocked with 10% normal goat serum with 0.3% Triton X-100. Fixed brain sections were incubated with primary antibody, anti–c-Fos (1:400, Invitrogen, no. PA1-830), with either anti-NeuN (clone A60) (1:400, Millipore, MAB3771), GFAP (1:400, Cell Signaling, no. 3670), or TMEM119 (1E7A2) (1:400, Invitrogen, no. 66948-1-IG) overnight at 4°C. Sections were then incubated with Alexa Fluor–conjugated secondary antibody (1:700, Invitrogen). Immunofluorescence images were acquired using an Olympus FV3000 confocal microscope (Olympus Corporation, Tokyo, Japan) with identical laser power, gain, and offset settings for all samples. Images within hippocampal CA1 were randomly taken at 200× magnification in a blinded fashion. Quantification of c-Fos, GFAP, and TREM119 immunoreactivity was performed using ImageJ [National Institutes of Health (NIH)]. Images were converted to 8-bit grayscale. After subtracting the background with a rolling ball radius of 50 pixels, each integrated density and area were measured using the Measure function in ImageJ. The integrated density was then normalized to the selected area to calculate the fluorescence intensity per unit area using the formula mean fluorescence intensity (MFI) = integrated density / area. For each section, values from four different areas of hippocampal CA1 were averaged to obtain a representative value.

Behavioral tests

1) Open-field test. The SmartCage system (29.8 by 18 by 12.8 cm) (AfaSci Inc., Redwood City, CA) was used to assess spontaneous activity in homecages. Automated data analysis was performed using CageScore software (AfaSci, Inc.). The home cage activity was determined by infrared beam breaks (x, y, and z photo-beam break counts). Mice were allowed to explore freely for 10 min. Movements of mice were recorded by Cage Center software and analyzed by Cage Score 2 software as described (59, 60).

2) Novel object recognition test. Novel object recognition protocol was similar to previously described (61). In brief, the apparatus consisted of a white plastic arena (40 by 40 cm) placed in the experimental room under low lighting conditions (~20 lux). The mouse was habituated to the arena without objects for 10 min for two consecutive days. On the third day, after habituation, the mouse was placed back to the home cage and the arena was cleaned with ethanol. Two identical objects, which were predetermined to have no natural significance or association with the control mice, were secured to the bottom of the arena. The mouse was allowed to explore the objects for 10 min, and then placed in the home cage again. The arena and objects were clean with ethanol. One same object and one different object were placed into the arena. The mouse was returned to the arena for a 10-min exploration of the same and novel objects, which was recorded and the discrimination index percentage was calculated.

3) T-Maze spontaneous alternation test. Each mouse was subjected to five trials. The T-maze test apparatus had three arms (30 × 10 cm²) and was elevated 30 cm above the ground. One trial consisted of a choice and a sample run. During the choice run one of the two target arms was blocked by a barrier according to a pseudorandom sequence, with equal numbers of left and right turns per session and with no more than two consecutive turns in the same direction. The mouse was allowed to explore the accessible arm. Before the sample run (intertrial interval of ∼10 s), the barrier was removed enabling accessibility to both arms. On the sample run, the mouse was replaced back into the start arm facing the experimenter. The mouse was allowed to choose one of the two target arms. The trial was classified as success if the animal chose the previously blocked arm.

4) Acoustic startle reflex and PPI. An automated Startle Reflex System (SR-LAB; San Diego Instruments Inc., San Diego, CA, USA) was used to measure acoustic startle reflex and PPI. Before the experiments, detailed piezoelectric sensitivity tests and dB calibration were performed to ensure reproducible startle responses. The system consisted of a startle chamber housed in a sound attenuated isolation cabinet equipped with an internal fan and light. A cylindrical transparent acrylic holding apparatus resting within the isolation chamber was used to hold each subject throughout the testing session. Background noise and acoustic stimuli were controlled through the SR-LAB microcomputer and interface assembly and kept constant at 65 dB. Mice were placed in cylinders 10 min before the initial startle stimuli and only background noise was offered during this acclimation period. To measure ASR and PPI, the five trials were: no stimulus, two types of startle stimulus only (100- and 120-dB, 20-ms broad band burst), and two types of startle stimulus preceded by a prepulse (a 20-ms broad band burst). The onset of the prepulse was separated from the startle onset by a 100-ms prepulse-startle interval, and the prepulse intensity used was 70 dB. Each was repeated nine times in random order. The trials were separated by an average interval of 30 s. SR-LAB software was used to calculate latency and startle response. PPI values were calculated as a percentage score for each prepulse using the following formula:

$$\begin{array}{c}\begin{array}{c}\text{PPI}(\%)=100\hfill \\ -{[(\text{ASR}\text{for prepulse}+\text{pulse trial})/\left(\text{ASR}\text{for pulse alone trial}\right)]}^{\ast }100\hfill \end{array}\hfill \end{array}$$

Flow cytometry

To assess brain cell fucosylation after FD, hippocampi were dissociated enzymatically using the Neural Tissue Dissociation Kit (Miltenyi Biotec). Cells suspension was resuspended in phosphate-buffered saline (PBS). AAL-fluorescein, LCA-Fluorescein (2 μg/ml; Vector Laboratories, Newark, CA) was added, or monoclonal mouse anti-mouse SSEA-1 (against Lewis X/CD15, clone MC-480, at 1:200 dilution, LSBio, Seattle, WA), and samples were incubated at 37°C for 1 hour in the dark. After incubation, the cells were pelleted at 300g for 5 min at 4°C and washed with PBS. The cells were then resuspended in 3 ml of PBS. Cell staining was excited at 488 nm and the emission was detected via the FL1 channel by a flow cytometer, BD Accuri C6 Plus (BD Sciences). Data were analyzed using FlowJo 10.8 (BD Sciences).

Human Aβ42 ELISA

Brain tissue samples from mice treated with FD or CD were fractionated into tris-buffered saline (TBS)–soluble and SDS-soluble fractions to assess the solubility of amyloid-β species as previously described (61). In brief, for the TBS-soluble fraction, tissues were homogenized in TBS with protease inhibitors, followed by centrifugation at 100,000g for 1 hour at 4°C. The pellets were then homogenized in SDS buffer containing 2% SDS in TBS with protease inhibitors, incubated on ice for 10 min, followed by centrifugation at 100,000g for 1 hour at 4°C. Aβ42 levels were measured using Human Aβ42 ELISA Kits (Thermo Fisher Scientific), following the manufacturer’s instructions.

Immunofluorescence staining of amyloid

Freshly dissected brains were fixed in 10% formalin and 1-mm coronal slices containing hippocampus and cerebral cortex were embedded in paraffin. Six-micrometer paraffin sections were obtained by microtome and used for immunofluorescence. Sections were dewaxed in xylene and hydrated through alcohol gradient. For antigen recovery, slides were boiled in Na citrate (10 mM, pH 6.0) for 15 min. After 1-hour incubation with 10% normal goat serum (Invitrogen) to block nonspecific binding, sections were stained with the amyloid dye (E,E)-1-fluoro-2,5-bis(3-hydroxycarbonyl-4-hydroxy) styrylbenzene (Sigma-Aldrich) (500 nM) and counterstained with propidium iodide (500 nM). Images were taken using a SPOT RT sCMOS camera (SPOT Imaging, Sterling Heights, MI) and analyzed using the NIH ImageJ software.

Statistics

Statistical analysis was performed using GraphPad Prism 9 software. All data are presented as means ± SEM. Comparison of the mean values was performed using unpaired Student’s two-tailed t test, one-way analysis of variance (ANOVA), or two-way ANOVA with Tukey’s post hoc test. Exact sample sizes and statistical test used for each comparison were provided in corresponding figure legends. P < 0.05 was considered to be statistically significant.

Supplementary Materials

The PDF file includes:

Figs. S1 to S14

Legends for tables S1 to S4

Other Supplementary Material for this manuscript includes the following:

Tables S1 to S4