Abstract

   Compelling evidence suggests that cognitive decline in Alzheimer’s
   disease is associated with the accumulation and aggregation of tau
   protein, with the most toxic aggregates being in the form of oligomers.
   This underscores the necessity for direct isolation and analysis of
   brain-derived tau oligomers from patients with Alzheimer’s disease,
   potentially offering novel perspectives into tau toxicity. Alzheimer’s
   brain-derived tau oligomers are potent inhibitors of synaptic
   plasticity; however, the involved mechanism is still not fully
   understood. We previously reported a significantly reduced incidence of
   Alzheimer’s disease in ageing humans chronically treated with a Food
   and Drug Administration–approved calcineurin inhibitor, FK506
   (tacrolimus), used as an immunosuppressant after solid organ
   transplant. Using a combination of electrophysiological and
   RNA-sequencing techniques, we provide here evidence that FK506 has the
   potential to block the acute toxic effect of brain-derived tau
   oligomers on synaptic plasticity, as well as to restore the levels of
   some key synaptic mRNAs. These results further support FK506 as a
   promising novel therapeutic strategy for the treatment of Alzheimer’s
   disease.

   Keywords: tacrolimus, FK506, tau, LTP, Alzheimer’s disease
     __________________________________________________________________

   Scaduto et al. report that tau oligomers, directly extracted from
   Alzheimer’s brain, cause synaptic impairment and consequent impairment
   of transcription induced after long-term potentiation. This toxicity is
   neutralized by FK506 (tacrolimus), a Food and Drug
   Administration–approved calcineurin inhibitor. These results suggest
   the central role of calcineurin in mediating tau toxicity.

Graphical Abstract

Graphical Abstract.

   [32]Graphical Abstract
   [33]Open in a new tab

Introduction

   Alzheimer’s disease is the most prevalent form of neurodegenerative
   dementia. It manifests as memory deficits and impairments in various
   cognitive domains, including learning, language comprehension and
   visuospatial abilities.^[34]1 Alzheimer’s disease is a growing global
   health concern. Current treatments do not offer a resolving cure as
   their efficacy is limited by our incomplete understanding of the
   underlying pathophysiology.^[35]2 Hallmarks of Alzheimer’s disease
   include brain atrophy beginning in the entorhinal cortex and
   hippocampus, dysfunction and ultimately loss of neurons and synapses,
   and the deposition of β-amyloid (Aβ) and hyperphosphorylated tau
   proteins, which are observable as extracellular plaques and
   intracellular neurofibrillary tangles (NFTs), respectively.^[36]3-6 The
   severity of the pathology is categorized using Braak stages^[37]7,[38]8
   that quantify the distribution of hyperphosphorylated tau protein
   throughout the brain. This assessment is based on the reactivity of the
   AT8 antibody, which specifically binds to tau protein when
   phosphorylated at serine 202 and threonine 205. Together with
   dysfunctional synapses, Braak stage correlates best with cognitive
   decline.^[39]9-12 Present scientific data indicate that toxic tau
   species predominantly propagate in the brain in the form of soluble tau
   oligomers.^[40]13 These oligomers seem to exhibit greater toxicity
   compared with larger tau fibrils found in NFTs or tau
   monomers.^[41]13,[42]14 Although many rigorous investigations have
   utilized recombinant tau (rTau), showing that it causes synaptic
   toxicity,^[43]15-17 direct analysis of brain-derived tau oligomers
   could offer novel insights. Brain-derived tau oligomers from
   Alzheimer’s disease individuals (hereafter referred to as BDTOs), and
   recently also tau in cerebrospinal fluid, have been found to impede
   long-term potentiation (LTP), an essential calcium-dependent process in
   synaptic plasticity, underlying learning and memory.^[44]18,[45]19 Tau
   pathology seems to be directly responsible of calcium
   dysregulation,^[46]20 and in addition, we previously reported that the
   calcineurin inhibitor Tacrolimus (FK506), a US Food and Drug
   Administration–approved immunosuppressant drug, is associated with
   reduced incidence of Alzheimer’s disease.^[47]21,[48]22 Calcineurin is
   a calcium- and calmodulin-dependent protein phosphatase that plays a
   regulatory role in various cellular processes, including neuronal
   signalling and synaptic plasticity.^[49]23 Calcineurin is involved in
   the dephosphorylation of tau protein, aiding in maintaining the balance
   between tau phosphorylation and dephosphorylation.^[50]24,[51]25 This
   deep integration in crucial neuronal processes suggests its potential
   implication in a variety of neurological disorders.^[52]26,[53]27 Based
   on these premises, we here hypothesized that inhibition of calcineurin
   prevents synaptic impairment caused by BDTO. We evaluated the potential
   neuroprotective properties of FK506, using electrophysiological and
   RNA-sequencing (RNA-seq) combined approach. Our study revealed that
   FK506 effectively counteracts synaptic impairment caused by BDTO
   effects, preserving transcription of key synaptic RNAs post-LTP, which
   is pivotal for the conversion of short- to long-term memory.

Materials and methods

BDTO isolation from human tissue

   BDTO isolation was performed as previously described.^[54]28 This
   preparation generates tau enriched for BDTO, which we will term as
   BDTO. Briefly, 100 mg of fresh-frozen hippocampus tissues of a single
   Alzheimer’s disease case (female, 90 years old, Braak stage 6 and PMI
   15 h) provided by The University of Michigan Brain Bank,
   neuropathological assessment conformed to the National Institute on
   Aging/Reagan Institute consensus criteria. The brain sample was
   homogenized in phosphate-buffered saline (PBS; Corning
   #46-013-CM)–containing protease (Roche, #11836153001) and phosphatase
   inhibitors (Roche, #490683700). Following homogenization, the tissues
   were centrifuged using a microcentrifuge to remove large debris. Tau
   oligomers were immunoprecipitated from this PBS soluble fraction by
   using the Pierce co-immunoprecipitation kit (Thermo Fisher #26149),
   with 75 µg of the T18 antibody, which has been demonstrated to
   selectively bind tau oligomers in the previous studies.^[55]29-31 To
   ensure the specificity of the T18 antibody, a mock pull-down using
   rabbit IgG isotype control was performed as a control, and the PBS
   soluble fraction was used as input for immunoprecipitation
   ([56]Supplementary Fig. 1). The antibody–antigen complex was eluted
   using 0.1 M glycine buffer (pH 2.8) and neutralized with 1.0 M Tris-Cl
   (pH 8.0). The co-immunoprecipitation process was carried out under a
   fume hood, and samples were periodically agitated using an orbital
   shaker to maximize antibody–antigen interaction. Western blot analysis
   was performed to validate BDTO immunoprecipitation upon protein
   quantification with the Micro BCA kit (Thermo Fisher #23235). In
   particular, 5 µg of input (PBS soluble fraction), 10 µl of
   immunoprecipitated material was loaded on precast NuPAGE 4–12% Bis–Tris
   gels (Invitrogen) for sodium dodecyl sulphate–polyacrylamide gel
   electrophoresis analysis. Gel was subsequently transferred onto
   nitrocellulose membranes and blocked for 1 h at room temperature using
   Odyssey Blocking Buffer. The membrane was probed overnight with Tau 13
   (1:20 000) at 4°C, as previously described.^[57]31 The following day,
   the membranes were washed and probed with IRdye secondaries antibodies
   (LI-COR) at 1:10 000 for 1 h at room temperature. The images were
   acquired by LI-COR Odyssey imager. Post-immunoprecipitation, the BDTOs
   from the patient with Alzheimer’s disease were amplified by using
   recombinant Tau 4R monomers (TauM), as previously
   described.^[58]28,[59]29,[60]32,[61]33 Briefly, BDTOs were seeded with
   recombinant TauMs at a ratio of 1:100 (BDTOs:TauM, w/w), with a minimum
   seeding amount of 100 µg TauM per seeding event. The seeding process
   involved the use of low-retention pipette tips and tubes, with samples
   stored in 50 ml falcon tubes and regularly agitated using an orbital
   shaker for 2 days at room temperature. The obtained amplified BDTOs
   were used for further experiments.

Animals, slice preparation and solutions

   Male C57Bl/6 mice (n = 17, age = 8–12 weeks) were procured from Jackson
   Laboratories to conduct the experimental procedures. The animals were
   accommodated in filter-topped cages, with a maximum of four mice per
   cage, in a facility maintaining a stable environment at 22°C and 40%
   humidity, under a 12:12 h light–dark cycle. Standard chow and water
   were available ad libitum. Specifically, male mice of an age known to
   yield stable and robust synaptic plasticity experiments to manage sex
   and age variables were chosen. This acknowledges the variability in
   responses to Alzheimer’s disease proteins and treatments across genders
   and ages, narrowing the generalizability of our results but
   establishing a foundation for broader, more inclusive future research.
   Experimental groups comprised a minimum of three mice each, and a total
   of at least 12 hippocampal slices were analysed per group. However,
   slices exhibiting suboptimal field excitatory post-synaptic potential
   (fEPSP) recordings (characterized by unclear or indistinguishable
   signals with amplitudes <50µV) were systematically excluded from the
   analysis (CTRL n = 11; BDTO n = 7; BDTO + FK506 n = 8; FK506 n = 11).
   The expected effect size was estimated from preliminary data obtained
   from similar studies. For randomization, brain slices from each animal
   were allocated across different experimental groups using a random
   number generator to ensure balanced representation and minimize
   intra-animal variability. This process ensured unbiased allocation not
   influenced by prior knowledge or expectations. To further minimize
   potential confounders, several strategies were employed: the order of
   treatments and measurements for each animal was randomized to prevent
   systematic bias; environmental conditions such as temperature, humidity
   and lighting were kept constant; and the experimenter conducting the
   measurements and analyses was blinded to the treatment groups. These
   measures were taken to ensure that the results obtained are as reliable
   and replicable as possible. This study received approval from the
   Institutional Animal Care and Use Committee at the University of Texas
   Medical Branch (UTMB, Galveston, TX, USA) and was conducted in strict
   adherence to the National Institutes of Health (NIH) guidelines on the
   ethical use of laboratory animals.^[62]34 All methodologies and
   experimental procedures were performed in compliance with UTMB and
   NIH-approved guidelines and regulations. All the solution used in these
   experiments were constantly bubbled with 5% CO[2] and 95% O[2]. The
   mice were euthanized under deep isoflurane anaesthesia and perfused
   using ice-cold NMDG-aCSF (N-Methyl-D-Glucamine artificial cerebrospinal
   fluid) solution (in mM: KCl 2.5; NaH[2]PO[4] 1.2; NaHCO[3] 30; HEPES
   (4-(2-Hydroxyethyl)-1-peperazineethanesulfonic acid) 20; glucose 25;
   sodium ascorbate 5; thiourea 2; sodium pyruvate 3; N-acetyl-cysteine
   12; NMDG 92; CaCl[2] 2; MgSO[4] 2). Brains were surgically removed from
   the skulls and sliced using the Compresstome VF-300 (Precisionary
   Instruments, Greenville, NC, USA) in ice-cold NMDG-aCSF to obtain
   350 µm thick brain slices. Slices were allowed to recover for 20 min in
   NMDG-aCSF at 35°C. Slices were then stored for at least 2 h in a
   recovery chamber containing holding-aCSF (in mM: KCl 2.5; NaH[2]PO[4]
   1.2; NaHCO[3] 30; HEPES 20; glucose 25; sodium ascorbate 5; thiourea 2;
   sodium pyruvate 3; N-acetyl-cysteine 12; NaCl 92; CaCl[2] 2; MgSO[4] 2)
   solution at room temperature. Prior to recording, all slices were
   incubated an additional hour with regular holding solution, or holding
   solution + BDTO 100 nM. During electrophysiological recordings, the
   slices were perfused at 2.5 ml/min with regular aCSF (in mM: NaCl 124;
   KCl 2.5; CaCl[2] 2; MgSO[4] 2; NaH[2]PO[4] 1.2; NaHCO[3] 24; HEPES 5;
   glucose 13) or aCSF + FK506 10 µM, at 21 ± 1°C. FK506 (Prograf®) was
   obtained from UTMB pharmacy and diluted before each experiment in aCSF
   to reach final concentration of 10 µM.

Electrophysiological recording

   The fEPSP was evoked stimulating Schaffer collateral pathway via a
   concentric bipolar electrode (FHC Inc., Bowdoin, ME, USA—Cat# 30200),
   and recording electrode was located at the stratum radiatum (CA1).
   Recordings were digitized with Digidata 1550B (Molecular Devices,
   Sunnyvale, CA, USA), collected using an Axon MultiClamp 700B
   differential amplifier (Molecular Devices) and analysed using Clampex
   10.6 software (Molecular Devices). Current stimulation was delivered
   through a digital stimulus isolation amplifier (AMPI, Israel). We
   performed an input–output protocol with stimulus ranging from 100 to
   1000 mA ([63]Supplementary Fig. 2), and for synaptic potentiation
   experiments, the stimulus was set to elicit an fEPSP ∼30% of maximum
   response. The online data acquisition software was used to monitor
   fEPSP slope every 20 s and average every three responses. After a
   stable baseline of 20 min, high-frequency stimulation (HFS), consisting
   of three trains of 100 pulses at 100 Hz with a 10-s interval, was used
   to induce LTP. LTP magnitude was measured as the average of fEPSP slope
   between the 50th and 60th minutes after HFS. The values were expressed
   as mean ± standard error of the mean (SEM) percentage change relative
   to their mean baseline fEPSP slope. For testing synaptic facilitation,
   two paired pulses (PPs) were administered within a 50-ms interval,
   which has been identified as optimal for eliciting substantial synaptic
   facilitation under experimental conditions similar to ours.^[64]35-38
   PP facilitation was quantified by calculating the ratio of the synaptic
   response amplitude elicited by the second stimulus to that of the first
   stimulus, with the result expressed as a percentage. The order in which
   the protocols were performed is the following: (i) I/O; (ii) PP; (iii)
   baseline; (iv) HFS; and (v) post-HFS.

RNA-sequencing

   After completion of the electrophysiological experiments, hippocampi
   were meticulously isolated from the rest of the brain slice. The
   isolated tissues were immediately snap frozen in liquid nitrogen to
   preserve RNA integrity. RNA extraction was subsequently performed using
   the Monarch Total RNA Miniprep Kit (New England Biolabs). To ensure
   complete removal of potential DNA contamination, the optional DNAse1
   digestion step was incorporated, as per the manufacturer’s
   instructions. The RNA samples were then used for sequencing library
   preparation following the Smart-3SEQ protocol. Briefly, 3′-end RNA-seq
   libraries were generated to specifically capture and sequence the 3′
   ends of polyadenylated RNA. This approach enhances the quantification
   accuracy of gene expression levels by focusing on a consistent region
   across different RNA molecules. All sequencing experiments and
   subsequent quality checks were conducted at the Next Generation
   Sequencing core facility of the UTMB, ensuring standardized protocols
   and optimal conditions were met throughout the process. Libraries were
   prepared via the Smart-3SEQ method. This entailed annealing the first
   strand primer to the sample RNA and extending it using SMARTScribe
   reverse transcriptase from Clontech. Subsequent synthesis of the second
   strand was done post the addition of the primer, followed by polymerase
   chain reaction (PCR)-mediated integration of adapter sequences with
   unique indices utilizing NEBNext single-index adapters from New England
   BioLabs. These PCR products were then purified using AMPure XP SPRI
   beads and sequenced on the Illumina NextSeq 550 platform. In the
   sequencing reads, the distinctive five-base molecular identifier and
   three guanine bases incorporated by the Smart-3SEQ were eliminated, and
   the identifier was appended to the read name using the
   umi_homopolymer.py tool. These reads were mapped onto the Mesocricetus
   auratus NCBI genome assembly, and gene read counts were calculated
   using FeatureCounts, which fed into DESeq2 for differential gene
   expression analysis. Analysis of differentially expressed genes (DEGs)
   was done using JMP.software and Metascape for gene enrichment pathway
   analysis.

Statistical analysis

   The sample size for each experimental group was determined based on a
   power analysis, aiming to ensure sufficient statistical power to detect
   a significant difference between groups with an alpha level of 0.05 and
   a power of 80%. Six slices in each group were considered the minimum
   for having sample size = 0.6. All data are expressed as mean ± SEM. In
   [65]Fig. 1 and [66]Supplementary Fig. 3, we used one-way ANOVA followed
   by Tukey’s post hoc test. In [67]Fig. 2, we employed a one-way ANOVA to
   assess statistical differences. To address the issue of multiple
   comparisons and control the false discovery rate (FDR), we applied an
   FDR-adjusted P-value approach. This method was utilized to discern
   practical differences among the groups. We discard one BDTO + FK506
   slice because the sequencing did not pass the quality standard of a
   minimum read depth and uniformity across the targeted regions, as well
   as having a high percentage of low-quality base calls.
   [68]Supplementary Fig. 2 was used two-way ANOVA multiple comparison.
   The stimulus-response curve was fit using third order polynomial model
   equation: Y = B0 + B1 × X + B2 × X^2 + B3 × X^3.

Figure 1.

   [69]Figure 1
   [70]Open in a new tab

   FK506 prevents hippocampal synaptic plasticity impairment induced by
   BDTO. Hippocampal slices from C57BL/6J mice (age: 8–12 weeks) underwent
   a 1-h pre-treatment with either standard holding solution or holding
   solution + BDTO (100 µM) prior to electrophysiological recording.
   Slices exposed to BDTO were then recorded using regular aCSF or aCSF +
   FK506 (10 µM). Field recordings were conducted with the stimulation
   electrode in the Schaffer collateral and the recording electrode in the
   Stratum Radiata of CA1 pyramidal neurons. Stimulation intensity was set
   to evoke 30% of maximal response. (A) Representative current traces of
   LTP protocol, average of 30 sweeps for baseline (10 minutes) and 180
   sweeps post-HFS (60 min). (B) Slope percentage of fEPSP. (C) Slope
   average over the final 10 min showing FK506’s prevention of LTP
   impairment by BDTO (half-brain slices each group: CTRL n = 11; BDTO n =
   7; BDTO + FK506 n = 8). (D) Representative current traces of PP
   stimulation protocol, average on 10 sweeps over 3 min, interstimulus
   interval (ISI) was 50 ms. (E) PP ratio indicating a trend towards
   impairment in the BDTO group compared with control and a significant
   reduction versus BDTO + FK506. Minimum of three different mice each
   group, each point represents data from a single half-brain slice (CTRL
   n = 11; BDTO n = 7; BDTO + FK506 n = 8, see the Statistical analysis
   section for criteria of exclusion). Statistical analysis involved
   one-way ANOVA followed by Tukey’s post hoc test.

Figure 2.

   [71]Figure 2
   [72]Open in a new tab

   Heat map illustrating PEA of DEGs in CTRL, BDTO and BDTO + FK506
   groups. DEGs were identified using an FDR P-value threshold in pairwise
   comparisons (e.g. CTRL > BDTO refers to genes upregulated in CTRL
   compared with BDTO; BDTO > CTRL refers to genes upregulated in BDTO
   compared with CTRL). Enrichment analysis involved accumulative
   hypergeometric P-values and enrichment factors (gene ontology terms). A
   hierarchical clustering approach was employed to categorize
   experimental groups. The heat map’s value scale ranges from 0
   (indicating non-significance) to 20 (representing statistical
   significance in gene expression variations between groups). The
   analysis highlights specific pathways and gene expressions modulated
   differently across treatment conditions (CTRL n = 11; BDTO n = 7; BDTO
   + FK506 n = 7, see the Statistical analysis section for criteria of
   exclusion). U1 snRNP, U1 small nuclear ribonucleoprotein; SWI/SNF,
   switch/sucrose non-fermentable; AP-1, activator protein 1.

Results

Electrophysiology

   To test the hypothesis that calcineurin inhibition prevents synaptic
   impairment caused by BDTO in regions of the brain that are vulnerable
   to Alzheimer’s disease pathology, we performed field recording
   experiments in the hippocampal Shaffer collateral—CA1 synaptic pathway
   of wild-type C67BL/6J mice. Prior to conducting the
   electrophysiological recordings, we incubated hippocampal slices in a
   small chamber with 450 µl of oxygenated holding solution either with or
   without BDTO for 1 h. Previous research has demonstrated that this
   duration of incubation is adequate for primary neurons to internalize
   BDTO.^[73]33 Then, slices incubated with BDTO were recorded in the
   presence or absence of FK506. Basal synaptic transmission did not show
   differences within the experimental groups ([74]Supplementary Fig. 2).
   We observed that the treatment with BDTO repressed LTP by decreasing
   fEPSP slopes ([75]Fig. 1A–C), and this toxic effect was blocked if the
   recording was done in the presence of FK506 [one-way ANOVA, F(2, 23) =
   6.376, P = 0.006 followed by Tukey’s multiple comparisons]. To
   determine dynamic properties of synaptic transmission, specifically
   short-term plasticity, we performed PP stimulation ([76]Fig. 1D and E).
   BDTO caused a trend of decrease in the second response of the PP
   compared with CTRL group, that ultimately was re-established in the
   presence of FK506 [one-way ANOVA, F(2, 23) = 4.793, P = 0.018, followed
   by Tukey’s multiple comparisons]. To exclude any effect of FK506 alone,
   we compared control slices either in the presence or in the absence of
   FK506 ([77]Supplementary Fig. 3), and we did not find any differences
   in fEPSP (unpaired t-test P = 0.356) and PP (unpaired t-test P =
   0.813).

RNA-sequencing

   To further explore whether BDTO and FK506 treatments together with HFS
   produced specific transcriptomic changes, we investigated transcripts
   levels on the same hippocampal slices subjected to HFS and field
   recording (LTP protocol). After the electrophysiological experiments,
   we isolated the hippocampus and snap froze the slices (∼4 h after
   incubation and 3 h after HFS). We first evaluated the effect of HFS on
   CTRL slices by comparing stimulated with no stimulated slices. We
   performed multiple comparison tests specifically applying an FDR
   P-value for the practical difference test. We found 387 DEG between
   CTRL stimulated and CTRL no stimulated. Notably, most of such genes
   were related with regulation of gene expression and chromatin
   remodelling ([78]Supplementary Fig. 4 and [79]Table 1), suggesting
   massive changes in transcription regulation after HFS. We then screened
   differences in mRNAs levels among the experimental groups that received
   HFS: CTRL (stimulated), BDTO and BDTO + FK506 ([80]Supplementary Table
   2). We found 275 DEG between CTRL and BDTO groups (158 higher in CTRL +
   117 higher in BDTO), 288 DEG between BDTO and BDTO + FK506 groups (136
   higher in BDTO + 152 higher in BDTO + FK506) and 179 DEG between CTRL
   and BDTO + FK506 (95 higher in CTRL + 84 higher in BDTO + FK506). We
   then performed pathway enrichment analysis (PEA) of each of the DEG
   groups ([81]Fig. 2). In order to improve our comprehension of the
   affected pathways and to facilitate the comparison of the six gene
   lists, we conducted a hierarchical cluster analysis. The results of
   this analysis are presented as a clustered heat map accompanied by a
   dendrogram containing multiple branches ([82]Fig. 2). The proximity of
   these branches indicates a higher degree of shared pathways among the
   lists. Two of them showed PEA in both CTRL and BDTO + FK506 versus
   BDTO, and they contained pathways related to mitochondria, energy,
   regulation of DNA and transcription. The other two branches showed
   differences between CTRL and BDTO + FK506 groups and they contained
   pathways related to transcription factor activator protein 1 (AP-1)
   complex, protein tyrosine/serine/threonine phosphatase activity and DNA
   regulation. This result showed that BDTO + FK506 was better associated
   with CTRL then BDTO, suggesting that FK506 treatment was able to
   restore some of the key mRNAs that were impaired by BDTO.

   To test the potential modulation of inflammatory genes by BDTO and
   FK506 treatment, we selected genes that positively (GO:0050729) or
   negatively (GO:0050728) influence the inflammatory cascade. DEG
   analysis was conducted through pairwise comparisons, mirroring the
   approach used in [83]Fig. 2. Specifically, we compared: CTRL versus
   BDTO, with CTRL upregulated genes (CTRL > BDTO) and downregulated genes
   (CTRL < BDTO); BDTO versus BDTO + FK506, with BDTO upregulated genes
   (BDTO > BDTO + FK506) and downregulated genes (BDTO < BDTO + FK506);
   and BDTO + FK506 versus CTRL, with upregulated genes (BDTO + FK506 >
   CTRL) and downregulated genes (BDTO + FK506 < CTRL). The quantity of
   genes modulating inflammation did not show marked variations within the
   groups studied. Nonetheless, a comprehensive list of DEG was compiled
   for each pairwise comparison in [84]Supplementary Table 3.
   Specifically, in comparison with the BDTO + FK506 group, the CTRL group
   exhibited an upregulation of 10 pro-inflammatory genes, including
   H2-M5, known for its role in antigen presentation [1], [2] and Pla2g3,
   implicated in the biosynthesis of pro-inflammatory eicosanoids [3].
   Additionally, CTRL presented an upregulation of 14 anti-inflammatory
   genes such as Foxp3, a transcription factor critical for the regulatory
   T-cell function [4], [5] and Nod2, known for its role in the negative
   regulation of immune responses [6] and [7].

   For the BDTO + FK506 group compared with CTRL, genes such as Pdcd4,
   involved in apoptosis and inflammation [8], and Clock, associated with
   circadian regulation of immune responses [9], were upregulated. In the
   comparison between BDTO and BDTO + FK506, genes such as Ripk1, a key
   regulator of inflammation and cell death [10], and Tnf, a major
   pro-inflammatory cytokine [11], were more expressed in BDTO.

   In the anti-inflammatory gene set, BDTO + FK506 showed higher
   expression levels of Cx3cr1, which encodes a receptor involved in the
   negative regulation of leucocyte adhesion and migration [12], [13] and
   Cd276, which has been implicated in the suppression of T-cell
   activation [14]. Conversely, BDTO displayed elevated levels of Ash1 l,
   know to its role in chromatin modification and suppression of the IL-6
   [15] and Bcr crucial for the maturation, antigen-specific growth and
   survival of B cells [16], compared with CTRL.

LTP correlates with synaptic transcripts

   To elucidate the impact of BDTO on synaptic mRNAs, and whether FK506
   was able to counteract their effect, using the same hippocampal slices,
   we focused our attention specifically on 1112 genes relevant to
   synaptic functioning as identified by the SYNGO database.^[85]39 Our
   approach employed a multiple comparison test (with FDR P-value for the
   practical difference test), mirroring the method in [86]Fig. 2.
   Synaptic DEGs were represented using a Venn diagram ([87]Fig. 3B). We
   observed that 40% of the genes that were upregulated (Atp6ap1, Tuba1a,
   Dlg4, Eef1a2, Ap2m1, Vps35, Cltc and Ywhaq) or downregulated (Sirt2,
   Kcnb1, Psenen, Dnajc5, Slc1a2, Prkn, Chd4 and Plg) in the CTRL group
   compared with the BDTO group also showed a similar trend in the BDTO +
   FK506 group when compared with the BDTO group ([88]Fig. 3A and B).
   These genes seem to be influenced by BDTO, but their expression is
   rectified by FK506. In addition, we found nine genes modulated uniquely
   by FK506 compared with BDTO (upregulated: Elavl4, Chrm3, Rpl23a and
   Snap91; downregulated: Adgrb1, Prkar2a, Rps23, Rpl38 and Rps27), and 16
   genes modulated specifically by BDTO compared with CTRL (upregulated:
   C1ql3, Grin2b, Bcr, Psen1, Sipa1l1, Pde4a, Arhgef7 and Syt11;
   downregulated: Rab8a, Cdc42, Snapin, Rab5b, Vamp4, Eif2s1, Eps15 and
   Cacng7). Interestingly, most of the genes dysregulated by BDTO and
   restored by FK506 were correlated with LTP, as shown by Pearson’s
   correlations between gene expressions and both fEPSP and PP ([89]Fig.
   3C and D). This result suggests that FK506 treatment can restore not
   only some of the basic plasticity mechanism in hippocampus, but also
   some of the early LTP downstream synaptic transcripts.

Figure 3.

   [90]Figure 3
   [91]Open in a new tab

   Synaptic gene expression and correlation with electrophysiological
   parameters. (A) Venn diagram illustrating the overlap and uniqueness of
   DEGs across different treatment groups. (B, C) Pearson’s correlation
   analysis between synaptic DEGs and electrophysiological measurements
   from [92]Fig. 1 (LTP and PP ratios). Genes upregulated in CTRL and BDTO
   + FK506 groups compared with BDTO (C) show a positive correlation with
   electrophysiological measurements, whereas genes upregulated in BDTO
   compared with CTRL and BDTO + FK506 (D) demonstrate a negative
   correlation. The data suggest a distinct gene expression profile in the
   BDTO group that inversely correlates with synaptic plasticity measures.
   The list of the genes in [93]Supplementary Table 4.

Discussion

   By employing a combination of electrophysiological and mRNA-seq
   techniques, our investigation provides evidence of the central role of
   calcineurin inhibition in mitigating the synaptic dysfunctions caused
   by BDTO, particularly in the hippocampus, a highly vulnerable brain
   region to Alzheimer’s disease pathology. This outcome aligns with our
   prior research, underscoring the diminished prevalence of Alzheimer’s
   disease and dementia in individuals prescribed the calcineurin
   inhibitor tacrolimus (FK506), thereby suggesting its protective
   properties.^[94]21,[95]22

   Using wild-type mice, we performed extracellular field recording of
   hippocampal Shaffer collateral—CA1 synaptic pathway confirming prior
   findings by others that BDTO hampers LTP by diminishing fEPSP
   slopes.^[96]18 This detrimental effect was effectively neutralized in
   the presence of FK506. Similarly, when we tested short-term
   potentiation, we found that BDTO decreased the ability of the synaptic
   transmission facilitation, re-established when in the presence of
   FK506. PP facilitation primarily indicates an augmented
   neurotransmitter release probability following the second pulse, likely
   attributable to residual Ca^2+ in the pre-synaptic terminal from the
   first pulse. However, synaptic transmission involves complex
   interactions between pre- and post-synaptic elements, and in some
   contexts, changes in post-synaptic sensitivity can also influence the
   overall synaptic response.^[97]40,[98]41 To examine whether basal EPSPs
   varied across groups, we generated input–output curves. These curves
   revealed no significant differences among the three groups
   ([99]Supplementary Fig. 2). Consequently, the diminished PP
   facilitation observed in BDTO does not appear to result from an
   inherently higher neurotransmitter release probability during the first
   pulse. We also tested FK506 alone that showed no differences in fEPSP
   slope and PP when compared with the CTRL group ([100]Supplementary Fig.
   3) excluding any effect on synaptic plasticity of FK506 alone at the
   concentration used (10 µM). Guided by our latest findings, we tailored
   the subsequent transcriptomic analysis to primarily assess FK506’s
   capacity to neutralize the harmful impact of BDTO, focusing on three
   experimental groups: CTRL, BDTO and BDTO + FK506. Although our
   electrophysiological data did not indicate significant effects of FK506
   alone on synaptic plasticity and existing literature lacks evidence
   this does not comprehensively rule out the drug’s potential
   transcriptomic impacts when administered independently.

   Considering that toxic tau induces rapid structural changes in neuronal
   nuclei,^[101]42,[102]43 which in turn can lead to chromatin
   rearrangements affecting transcription,^[103]44-46 and that memory
   consolidation necessitates the synthesis of de novo
   mRNA,^[104]47,[105]48 we conducted mRNA-seq on the identical
   hippocampal slices employed in our electrophysiology experiments. Our
   initial focus was to determine the sole impact of HFS by comparing the
   gene expression in CTRL stimulated with HFS versus CTRL no stimulated.
   In line with expectations for LTP responses, genes, such as Arc and
   Sgk1, were significantly elevated in the HFS-exposed CTRL group
   compared with the non-stimulated CTRL group. Arc is known to influence
   synaptic strength and reorganize the actin cytoskeleton
   post-LTP,^[106]49 while Sgk1 plays a role in the modulation of ion
   channels, neurotransmitter receptors and various signalling
   pathways.^[107]50 Interestingly, other canonical LTP-responsive genes,
   such as CREB, BDNF and c-fos, did not show differential expression
   following HFS, which might be due to the normalization of their
   expression levels within the 3-h post-induction timeframe and the fact
   that our samples comprise a heterogeneous mix of cells, not exclusively
   neurons. Broadening our analysis, enrichment pathway analysis of all
   DEG unravelled pronounced alterations between CTRL stimulated with HFS
   versus CTRL no stimulated, predominantly linked to gene regulation and
   chromatin remodelling. Furthermore, we compared the three
   HFS-stimulated experimental groups (CTRL, BDTO and BDTO + FK506), our
   transcriptomic analysis highlighted a closer similarity between BDTO +
   FK506 with CTRL, diverging notably from the BDTO-alone group. The
   hierarchical clustering shows this distinction, where two of the major
   branches of the dendrogram highlighted differences in mitochondrial
   genes, energy pathways and DNA transcription regulation. This evidence
   suggests that FK506 aids the preservation of mitochondrial integrity
   and function, helping in counteracting the energy deficits and
   oxidative stress caused by the impaired mitochondrial activity, common
   in tauopathies.^[108]51 Additionally, FK506 showed efficacy in
   modulating the general cellular energy metabolism, further contributing
   to the restoration of cellular homeostasis initially perturbed by the
   BDTO. To address the issue of sample cell type heterogeneity, we
   narrowed our focus to the impact of BDTO on synaptic mRNAs (neuronal)
   and its possible rectification by FK506. From the examination of 1112
   genes pivotal for synaptic functioning, as delineated by the SYNGO
   database, we identified 49 DEG across the 3 groups, observing that
   FK506 demonstrated promising restorative effects on ∼40% of the genes
   dysregulated by BDTO. This group includes key players, such as Atp6ap1,
   Dlg4 (PSD-95), Ywhaq (14-3-3 protein theta), Slc1a2 (GLT-1), Psen1 and
   Prkn. Delving into these specific genes provides critical molecular
   insights into FK506’s protective mechanisms on synaptic plasticity
   within the intricate framework of Alzheimer’s disease pathophysiology.
   Atp6ap1 is implicated in the endosomal–lysosomal system, a pathway
   increasingly recognized for its role in Alzheimer’s disease–associated
   neurodegenerative processes.^[109]52  Dlg4, also referred to as PSD-95,
   is integral to post-synaptic density structures and is essential for
   maintaining synaptic plasticity.^[110]53,[111]54 Previous research has
   linked aberrant Dlg4 activity to synaptic deficits characteristic of
   Alzheimer’s disease.^[112]55 Concurrently, Ywhaq, known as the 14-3-3
   protein theta, interacts directly with phosphorylated
   tau,^[113]56,[114]57 and it is downregulated in Alzheimer’s
   disease.^[115]58 The Slc1a2 gene, denoted as GLT-1, is central to the
   modulation of glutamate uptake, and alteration in its expression is
   related with sporadic Alzheimer’s disease.^[116]59,[117]60 Presenilin
   genes, included Psen1, are central to certain familial Alzheimer’s
   disease forms and are integral for amyloid precursor protein processing
   and the regulation of intracellular calcium dynamics.^[118]61,[119]62
   The Prkn gene is known for its role in Parkinson’s disease due to its
   involvement in mitophagy and mitochondrial quality control.^[120]63
   Although not a classic genetic marker for Alzheimer’s disease, Prkn’s
   role in cellular maintenance, including mitochondrial regulation and
   autophagy, may intersect with Alzheimer’s disease mechanisms,
   indicating potential relevance to the mitochondrial dysregulation
   observed in Alzheimer’s disease pathology.

   This intricate genomic modulation appears consistent with the observed
   electrophysiological changes. Many genes disturbed by BDTO, and
   subsequently normalized by FK506, exhibit significant correlations with
   PP and fEPSP. This interplay suggests that changes in mRNA expression
   related to synaptic functions are closely associated with alterations
   in synaptic plasticity. Furthermore, these observations may reveal
   novel molecular targets and mechanisms by which FK506 preserves
   synaptic function against the detrimental effects of BDTO.

   Some of the genes were modulated by BDTO but not restored by FK506,
   including Chrm3, which encodes the muscarinic acetylcholine receptor
   M3, known for its involvement in cognitive functions.^[121]64
   Additionally, a subset of genes was uniquely modulated by the BDTO +
   FK506 compared with BDTO alone, indicating a unique modulation of the
   FK506 on the toxic environment induced by BDTO treatment. Among these,
   Grin2b and Psen1 stand out for their direct relevance to this study.
   Grin2b is crucial for synaptic plasticity.^[122]65 Psen1 mutations are
   a known genetic cause of Alzheimer’s disease^[123]61 and reduction of
   calcineurin activity in presenilin 1 M146V mutant rescue physiological
   synaptic plasticity.^[124]66

   Application of BDTO or BDTO in combination with FK506 in our study did
   not result in a massive variation in the expression of genes modulating
   inflammation across the different groups. Notably, in the CTRL group
   compared with the BDTO + FK506 group, there was an upregulation of
   small number of pro-inflammatory and anti-inflammatory genes,
   indicating a nuanced effect of FK506 on inflammatory pathways.
   Moreover, our findings suggest that while FK506 influences the
   expression of some genes related to inflammation and immune response,
   the overall impact on inflammatory gene expression appears to be
   limited in the context of our experimental conditions may due to the
   short time between treatments and storage of the brain slice at −80°C.
   However, prolonged administration of calcineurin inhibitor, as observed
   in previous studies,^[125]21,[126]22 could potentially yield
   anti-inflammatory effects of FK506, leading to neuroprotection. This
   effect may be mediated by inhibiting the calcineurin/Nuclear Factor of
   Activated T-cells pathway implicated in the regulation of glutamate
   homeostasis through the downregulation of Glt1/EAAT2.^[127]67,[128]68
   Loss of Glt1/EAAT2 can lead to neuronal hyperexcitability,^[129]69
   which is one of the alteration found in Alzheimer’s disease.

   In summary, our findings converge to strengthen the therapeutic promise
   of FK506 in addressing synaptic and transcriptional abnormalities that
   arise in the context of Alzheimer’s disease. These revelations serve as
   a guiding light, directing further in-depth investigations into the
   precise mechanisms of action of FK506 and its potential clinical
   implications.

Supplementary Material

   fcae277_Supplementary_Data
   [130]fcae277_supplementary_data.zip^ (575.1KB, zip)

Acknowledgements