Abstract

   MicroRNA (miR)-132 brain-to-body messages suppress inflammation by
   targeting acetylcholinesterase (AChE), but the target specificity of
   3’-AChE splice variants and the signaling pathways involved remain
   unknown. Using surface plasmon resonance (SPR), we identified
   preferential miR-132 targeting of soluble AChE-R over synaptic-bound
   AChE-S, potentiating miR-132-mediated brain and body cholinergic
   suppression of pro-inflammatory cytokines. Inversely, bacterial
   lipopolysaccharide (LPS) reduced multiple miR-132 targets, suppressed
   AChE-S more than AChE-R and elevated inflammatory hallmarks.
   Furthermore, blockade of peripheral miR-132 by chemically protected
   AM132 antisense oligonucleotide elevated muscle AChE-R 10-fold over
   AChE-S, and cortical miRNA-sequencing demonstrated inverse brain
   changes by AM132 and LPS in immune-related miRs and neurotransmission
   and cholinergic signaling pathways. In neuromuscular junctions, AM132
   co-elevated the nicotinic acetylcholine receptor and AChE, re-balancing
   neurotransmission and reaching mild muscle incoordination. Our findings
   demonstrate preferential miR-132-induced modulation of AChE-R which
   ignites bidirectional brain and body anti-inflammatory regulation,
   underscoring splice-variant miR-132 specificity as a new complexity
   level in inflammatory surveillance.
     __________________________________________________________________

   Most coding RNA transcripts undergo alternative splicing[38]^1, and the
   resultant splice variants can interact with non-coding microRNAs (miRs)
   which suppress their function[39]^2. Functional miR recognition sites
   mainly reside within and near both ends of the 3′-untranslated region
   (3′-UTR)[40]^3. In over half of the human genes, alternative splicing
   events modify the 3′-UTR[41]^4,[42]^5, indicating functional
   association of transcript variants with miRs that target them[43]^6.
   Such 3′-UTR alternative splicing may further produce protein isoforms
   with modified properties, as is the case for the secreted, soluble
   monomeric acetylcholine hydrolyzing protein AChE-R, which differs in
   location from the synaptic membrane-associated AChE-S tetramers[44]^7.
   However, whether miRs differentially suppress these splice variant
   products of the AChE gene, and if so what are the consequences of such
   selectivity remain unknown. For example, miRs modulate
   neurotransmission and immune pathways in body and brain in response to
   acute inflammation and stressful experiences[45]^8,[46]^9,[47]^10. This
   induces a rapid anti-inflammatory response which is initiated by
   potentiated cholinergic tone via brain-to-body signaling of the vagus
   nerve[48]^11. We, and others have shown that miRs as well as natural
   antisense transcripts[49]^12 are pivotal for controlling these
   neuronal-immune processes[50]^13,[51]^14,[52]^15 and provide continuous
   surveillance over peripheral cholinergic anti-inflammatory
   signaling[53]^16,[54]^17,[55]^18,[56]^19. Nevertheless, whether the
   body orchestrates this brain-induced miR-mediated response via splice
   variant preference was not yet explored.

   Notably, miRs operate in a complex combinatorial mode, driven by
   simultaneous cell type and tissue-specific, context-dependent
   suppression of multiple target transcripts[57]^18,[58]^20. Also,
   numerous transcripts are commonly suppressed by multiple
   miRs[59]^18,[60]^19, and recent work supports the concept that miRs
   migrate between tissues when packaged in exosomal lipoprotein
   compartments[61]^21. Yet, some miRs show more potent function than
   others; in particular, miR-132 plays multiple roles in neuronal
   development, anxiety, and metabolism. It suppresses many
   pro-inflammatory target transcripts, including the transcriptional
   co-activator P300[62]^22, the growth factor HB-EGF[63]^23, the
   dopaminergic regulator Nurr1[64]^24 and the epigenetic modulator SIRT1,
   further potentiating anti-inflammatory processes[65]^25. We set out to
   explore the hypothesis that miR-132 is causally involved in the
   bidirectional brain-body signaling that provides continuous
   surveillance over inflammatory reactions. Since anti-inflammatory
   responses involve elevation of the brain’s miR-132[66]^26,[67]^27, and
   as luciferase tests and lentivirus suppression validated that miR-132
   targets AChE[68]^26,[69]^27, we predicted that miR-132 conveys
   brain-to-body anti-inflammatory signals, by modulating acetylcholine
   (ACh) levels via targeting, and suppressing both central and peripheral
   AChE[70]^26. This inhibition limits ACh hydrolysis, and elevates the
   cholinergic tone (known to increase substantially under acute
   stress)[71]^28. Moreover, miR-132 may induce, rather than suppress
   particular pathways; for example, by decreasing the expression of
   negative transcriptional regulators known to exist in the cholinergic
   system[72]^29. Therefore, increasing levels of brain miR-132 can
   attenuate inflammation through up- and down-regulation of multiple
   genes and pathways.

   Cholinergic perturbation-induced alternative splicing promotes exon
   inclusion, which shifts production of the major ‘synaptic’
   membrane-associated AChE-S tetramers to the soluble AChE-R
   monomer[73]^28,[74]^30,[75]^31. The 3′-untranslated regions of both
   AChE variants include the ‘seed’ motif for miR-132[76]^32, indicating
   that stress-induced miR-132 increases can suppress both AChE-S and
   AChE-R. However, AChE-S suppression would elevate synaptic ACh levels
   and modulate cholinergic neurotransmission, whereas AChE-R suppression
   would elevate extracellular ACh, potentiating the cholinergic
   anti-inflammatory effect[77]^11. In comparison, acute organophosphate
   poisoning blocks AChE irreversibly, and induces cholinergic
   hyper-excitation while leading to massive inflammation[78]^33. To
   determine if the AChE-S or AChE-R variants are preferentially targeted
   by miR-132, we used surface plasmon resonance (SPR) to predict variant
   binding preference; and employed an antisense oligonucleotide
   suppressor of miR-132, AM132, which cannot cross the blood brain
   barrier and whose effect is limited to the periphery[79]^34, compatible
   with numerous reports[80]^35. To find if such suppression involves
   brain-body inflammatory regulating messages[81]^36, we characterized
   the brain and body reaction of mice injected with AM132 or LPS (known
   to increase miR-132 expression[82]^26) by RNA-sequencing, qRT-PCR
   validations, AChE activity tests and bioinformatics interrogation (see
   [83]Fig. 1 for scheme of workflow).

Figure 1. Schematic work flow.

   [84]Figure 1
   [85]Open in a new tab

   We injected LPS (i.p.) and AM132 (i.v.) on two alternate days and 24 h
   later extracted brain, muscle, and serum. We used SPR tests to assess
   variant specificities; RNA-sequencing to derive frontal cortices miR
   profiles; and bioinformatics processing and qPCR for quantifying mRNA
   targets of miR-132 and pro-inflammatory cytokines. AChE activity was
   measured in serum, tissue homogenates and NMJs, and NMJ AChE and nAChR
   were labeled with histochemical Karnovsky’s staining;
   immunohistochemistry and NMJ anatomy were explored by light and
   transmission electron microscopy (TEM), respectively.

Results

SPR-based analysis predicts tighter miR-132 interaction with the soluble
AChE-R splice variant

   Like many other miRs, miR-132 has an exhaustive set of potential
   targets (see miRNAwalk:
   [86]http://www.umm.uni-heidelberg.de/apps/zmf/miRNAwalk). Of those, we
   were particularly interested in the validated miR-132 targets AChE-S
   and AChE-R[87]^32 as well as SIRT1[88]^37, all of which may be involved
   in anxiety and anti-inflammatory signaling. To experimentally measure
   association of miR-132 with its targets, we used a SPR binding assay.
   Given that miRNA-target interactions may involve longer regions than
   the seed itself[89]^38, we immobilized biotinylated 30-mer RNA
   sequences corresponding to the 3′-UTR regions of the major AChE-S
   splice variant[90]^7 or of the soluble AChE-R splice variant to SPR
   chips and injected a 22-mer RNA oligonucleotide containing the miR-132
   sequence (see [91]Supplementary Fig. S1 and [92]Supplementary Table
   S1). Results demonstrated a ~2.5-fold higher affinity of miR-132 to the
   stress-inducible minor splice variant AChE-R compared to the major
   synaptic AChE-S sequence (K[D] of 8.13 vs 18.75 nM, [93]Fig. 2a,b) and
   an even higher affinity to SIRT1 (0.42 nM, [94]Fig. 2b), predicting a
   hierarchical binding preference of miR-132 to (in increasing
   affinities) SIRT1, AChE-R, and AChE-S. Interestingly, all three showed
   similar affinity to that of the other AChE-targeting miR-608[95]^39.

Figure 2. AM132 reduces miR-132 and elevates AChE in periphery and brain.

   [96]Figure 2
   [97]Open in a new tab

   (a,b) SPR quantifications predict Ka (M-1 sec-1), Kd (sec) and KD (nM)
   binding affinities of miR-132 to AChE-S, AChE-R and SIRT1. (c)
   Schematic work flow of AM132 treatment. (d) MiR-132 levels are
   unchanged in the brain and significantly decreased in muscle (p < 0.01,
   t-test). (e) Elevated levels of the AChE-R (p < 0.0001) and AChE-S
   (p < 0.01) alternative splice variants in AM132-treated gastrocnemius
   muscle as compared to control. (f) Increased serum AChE activity
   post-AM132 treatment at 6, 12 and 24 h (p < 0.05, T-test); and at 24 h
   in frontal cortex and in muscle (p < 0.05). (g) AM132 treated mice show
   early latency to fall off the bars in a third trial of a fixed speed
   Rotarod test (*p < 0.05, ANOVA).

AM132 suppresses peripheral miR-132 and preferentially elevates AChE-R
activities

   Inflammation commonly induces ‘sickness behavior’[98]^36,[99]^40,
   suggesting that transcriptional changes in the Central Nervous System
   (CNS) of miR and mRNA expression during inflammation may be causally
   involved. To address this issue, we treated mice with AM132, a 15-mer
   LNA-protected antisense oligonucleotide ([100]Figs 1, [101]2c) that
   selectively binds and inactivates miR-132, or with a control
   oligonucleotide with no murine transcript targets ([102]Fig. 1,
   [103]Supplementary Fig. S1). Muscle, but not cortical miR-132 levels
   were reduced ([104]Fig. 2d) within 24 hours in AM132-treated mice,
   compatible with the predicted failure of AM132 to cross the blood-brain
   barrier[105]^34. Also, q-PCR measurements revealed elevated muscle
   AChE-R, much more than AChE-S mRNA levels in AM132-treated mice (n = 4
   [106]Fig. 2e; p < 0.0001 compared to p < 0.05), and the effect was even
   greater than what was predicted by the SPR measurements, suggesting
   involvement of additional elements in the in vivo reaction. Given that
   alternative splicing regulators may control this change in AChE splice
   variants, we further quantified the muscle mRNA encoding the SC35
   splicing factor known to shift the 3′-UTR of AChE-S to AChE-R[107]^41.
   Muscle SC35 mRNA levels increased by approximately two fold
   ([108]Supplementary Fig. S2a), which might only explain part of the
   increase in AChE-R[109]^41. In the frontal cortex, SC35 levels remained
   unchanged under AM132, but were significantly elevated under LPS
   exposure. In the frontal cortex, SC35 levels remained unchanged under
   AM132, but were significantly elevated under LPS exposure
   ([110]Supplementary Fig. S2b). Additionally, SIRT1 expression was
   modestly increased ([111]Supplementary Fig. S2c), further highlights
   the pronounced AM132 impact on AChE-R. Compatible with the elevated
   AChE-R mRNA levels, the ACh hydrolytic activity of AChE was increased
   in the serum, brain and muscle of AM132-treated mice at 6, 12 and
   24 hours post-injection ([112]Fig. 2f; n = 4, p < 0.05), in spite of
   the brain-inaccessible nature of AM132[113]^35. However, in spite of
   this massive elevation in ACh hydrolysis, AM132-treated mice only
   showed mild muscle incoordination, evident in their shorter latency to
   falling off a beam (n = 6, p < 0.05, ANOVA) on the third trial in the
   Rotarod test. This observation reflects somewhat exacerbated muscle
   fatigue compared to PBS-treated and naïve mice ([114]Fig. 2g).
   Together, these findings validated a preferential impact of
   peripherally modified miR-132 signaling via the soluble AChE-R splice
   variant and consequently weakened cholinergic signaling, indicating
   that selective AM132 modulation of muscle AChE-R expression leads to
   mild muscle incoordination.

AM132 and LPS exert inverse impacts on cortical cholinergic anti-inflammatory
regulators

   Peripheral elevations and decreases in miR-132 targets may modify
   efferent vagus signaling reaching the brain[115]^11,[116]^26, which
   would affect the brain’s cholinergic balance and inflammatory state. To
   quantify miR-132 target transcripts, we compared frontal cortices of
   AM132-treated and LPS-exposed mice (shown to elevate brain miR-132
   levels, [117]Supplementary Figs S3, S4). Cortices from LPS-exposed mice
   presented down-regulation of the ‘synaptic’ variant AChE-S (n = 8,
   p < 0.01), the transcription controller P300 (n = 8, p < 0.01), the
   epigenetic modulator SIRT1 (n = 8, p < 0.05), and the growth factor
   HB-EGF (n = 8, p < 0.05), which are all validated miR-132 targets
   ([118]Fig. 3, left hand side), compatible with our previous findings
   that LPS exposure modulated cortical cholinergic neurotransmission via
   potentiating miR-132 suppression[119]^26,[120]^42. In contrast, the
   stress-induced soluble AChE-R variant and the cholinergic-regulated
   pro-inflammatory cytokines IL18 (n = 8, p < 0.01), IL6 (n = 8,
   p < 0.01), IL1β (n = 8, p < 0.05), and TNFα (n = 8,
   p < 0.05)[121]^11,[122]^26 were all elevated, reflecting an accentuated
   cortical inflammatory state following LPS exposure. These findings, and
   that the splice factor SC35, which promotes the alternative splicing
   shift to AChE-R was also increased under LPS exposure
   ([123]Supplementary Fig. S2b), together indicated co-exacerbated brain
   anxiety and inflammation under LPS exposure.

Figure 3. Inversely modulated cortical targets and cytokines under LPS and
AM132 treatment.

   [124]Figure 3
   [125]Open in a new tab

   (a) qRT-PCR of targets (top) and pro-inflammatory cytokines (bottom)
   under LPS (left) and AM132 treatment (right). Note inverse regulation
   under LPS exposure (pink background) and AM132 treatment (blue
   background). P values reflect 0.05 (*) or 0.01 (**) (Student’s T test).
   Note that LPS exposure elevates (red arrows) miR-132, reducing its
   targets (blue lines) and enhancing the consequent inflammation
   biomarkers. (b) Schematic representation of the tested brain and muscle
   tissues, the alternative splicing of AChE mRNA and the observed changes
   in modified miR-132 targets and pro-inflammatory cytokines. AChE-S
   downregulation may mitigate cholinergic neurotransmission whereas
   AChE-R down regulation should enhance the cholinergic anti-inflammatory
   reflex, demonstrating bidirectional rebalancing. In comparison, AM132
   treatment (blue background scheme) upregulates miR-132 targets which
   downregulates pro-inflammatory cytokines, with elevated AChE-S and
   AChR-R inhibiting the cholinergic anti-inflammatory effect.

   Unlike the LPS effect, cortices from AM132-treated mice showed
   significant up-regulation of both the membrane-bound AChE-S (n = 4,
   p < 0.01) and the soluble extracellular AChE-R transcripts, predicting
   suppressed cholinergic signals and modified structure and function of
   cholinergic synapses[126]^43. In parallel, up regulation of P300
   (n = 4, p < 0.05), SIRT1 (n = 4, p < 0.05), and HB-EGF (n = 4,
   p < 0.05) indicated a synergized brain anti-inflammatory impact of
   AM132, along with potentiated individual functions of these target
   genes[127]^13,[128]^16,[129]^17. This complex outcome was accompanied
   by downregulation of the pro-inflammatory MCP-1 (n = 4, p < 0.01) and
   TNFα (n = 4, p < 0.05) cytokines ([130]Fig. 3, top right). Thus, in
   spite of its brain-impermeable feature, peripheral miR-132 manipulation
   affected both the miR-132 targets and their downstream cortical
   transcripts, in a largely inverse manner to the LPS effect ([131]Fig.
   3, bottom schemes).

AM132 and LPS induce inversely oriented changes in cortical miR profiles

   To compare the outcome of peripheral miR-132 increases and decreases on
   the brain and body inflammatory status, we performed cortical miRNA
   profiling of LPS-induced and AM132 treated samples. RNA-sequencing
   yielded over 14.5 M reads per cortical sample; with mean 32% alignment
   to miRBase and 1,199 distinct miRs that were expressed in all samples.
   In LPS-exposed mice compared to controls, expression analysis using
   DeSeq2 revealed a small number of 20 differentially expressed cortical
   miRs (see [132]Supplementary Table S2 for detailed lists of modified
   brain miRs) ([133]Fig. 4a, n = 3). Although AM132 cannot pass the
   blood-brain barrier[134]^34,[135]^35, peripheral AM132 injection
   induced differential expression of 129 cortical miRs compared to
   control brains ([136]Fig. 4b. n = 4, FDR < 0.05, >30 CPM expression;
   see [137]Supplementary Table S3 for detailed lists of modified brain
   miRs and [138]Supplementary Fig. S3 for qPCR results of predicted AChE
   targeting miRs). In both analyses, unsupervised classification
   correctly separated experimental from control cortices (PCA plot,
   [139]Supplementary Fig. S3). In spite of the relatively high
   inter-animal variability in frontal cortex-expressed miRs from
   LPS-exposed mice, 12 out of the 20 miRs that were differentially
   expressed between LPS-injected mice and their controls were also
   differentially expressed in between AM132-injected mice and their
   controls ([140]Fig. 4b). Six of those 12 miRs (mmu-miR-21a-5p, -27a-3p,
   -137-3p, -341-3p, -344b-5p and -let-7b-5p) were inversely regulated
   between AM132 and LPS treatments, and the other six (mmu-miR-7b-5p,
   -181a-5p, -218-5p, -488-3 and -490-3p) were regulated in the same
   direction (either up or down) under both treatments ([141]Fig. 4c).
   Thus, AM132 exposure induced profound and distinct cortical miR changes
   compared to LPS exposure.

Figure 4. Cortical miR profiling reveals inverse changes under LPS and AM132
treatment.

   [142]Figure 4
   [143]Open in a new tab

   (a) Hierarchical clustering dendrograms showing unsupervised clustering
   and differentially changed miRs under LPS exposure compared with naïve
   control (Red), and under AM132 treatment compared with random
   oligonucleotide treated control (Blue). (b) Venn diagram showing the
   number of miRs differentially expressed with LPS/ AM132 and the overlap
   between them. (c) List of miRs overlapping between LPS/AM132 and their
   changing directions (up or down). Note the largely inverse direction of
   changes.

   Next, we ran a miRpath v. 3.0-based pathway analysis (using tarbase v.7
   validated target database) updated to current annotations.
   Specifically, we searched for overlapping miRs and their regulated
   pathways between the LPS-exposed and AM132-treated groups, with
   p < 0.05 after FDR correction on both sets ([144]Fig. 5a). This
   analysis revealed unique enrichment of upregulated cholinergic
   transcripts under both LPS and AM132, suggesting a potentiated
   cholinergic response to both of these perturbations. This was
   accompanied by an overall enrichment for neuronal pathways such as
   synaptic transmission, but also inflammatory signaling pathways in both
   groups. Transcripts in 13 of these co-enriched pathways (after
   filtering 5 pathways relating to systems that do not operate in the
   brain), could either be up-regulated and/or down-regulated under LPS or
   AM132. Thus, both AM132 and LPS induced bi-directionally regulated
   cellular and metabolic pathways that are involved in several
   neuro-signaling and inflammatory processes, which include dopaminergic,
   glutamatergic, and GABAergic pathways[145]^16,[146]^17.

Figure 5. Pathway enrichment analysis of overlapping miRs.

   [147]Figure 5
   [148]Open in a new tab

   (a) Inversely and similarly regulated miR-related enriched pathways
   arranged by the significance of change. (b) Overlapping pathways
   clustered by percent of overlapping miRs in each. Note:
   neurotransmitter signaling pathways and inflammatory pathways cluster
   together. (c) Scheme of mediation analysis linking miR and pathway
   changes with modified target genes mediating these pathways.

   The modified brain miRs are each targeted to many different coding
   transcripts. Therefore, we next questioned how many of those targets
   are shared between the brain pathways that are regulated by both LPS
   and AM132. Re-clustering of these pathways according to the percent
   overlap of their validated miR targets ([149]Fig. 5b,c) pointed at both
   inflammation-controlling and neurotransmission-related targets, such as
   VEGF, axon guidance, MAPK and neurotrophin receptor as modified under
   both AM132 and LPS. The neurotransmission controllers were modified
   more often than inflammation-related ones (binomial test, p < 0.05). To
   exclude the possibility that size variations in the number of genes per
   pathway skewed the analysis, we applied a permutation test to miR
   targets composing these different pathways. This test re-affirmed that
   brain-modified miRs which change under both AM132 and LPS are enriched
   for neurotransmitter signaling and inflammatory genes (n = 10,000
   permutations, p value < 0.05 for each test). Repeating this permutation
   test after removal of inflammation and neuronal signaling genes from
   the analyzed lists deleted the significant difference (p > 0.05)
   between LPS and AM132-regulated pathways, validating these subgroups as
   pivotal to the observed differences. Taken together, these findings
   indicate that LPS and AM132 induced partially similar miR changes
   involved in both inflammation and neurotransmission, but the
   LPS-induced changes promoted inflammation whereas AM132-mediated
   cortical changes suppressed it.

AM132 co-increases the nAChR and AChE proteins in diaphragm NMJs

   One tissue where blocking miR-132 by AM132 may exert a significant
   impact is the diaphragm, where balanced response of neuromuscular
   junctions (NMJs) to cholinergic signals is essential for breathing.
   Specifically, AM132 would decrease miR-132 levels, elevating the
   soluble AChE-R protein more than the membrane-bound AChE-S variant.
   This may functionally diminish the response of nicotinic acetylcholine
   receptors (nAChR) in NMJs to ACh signals ([150]Fig. 6a). However, the
   mild Rotarod effect (see [151]Fig. 2g above) indicated that this
   outcome was minimized, indicating a compensating process. To
   interrogate the effect of AM132 on the NMJs cholinergic tone, we used
   preparations of whole diaphragm muscle, known to be extensively
   innervated by the vagus nerve. Diaphragms were extracted from mice
   injected with AM132, scrambled (SCR) oligonucleotide control or
   phosphate-buffered saline (PBS), 6 h and 24 h post-injection. They were
   then labeled with antibodies directed to AChE or with fluorescein
   labeled α-Bungarotoxin (αBTX) for tagging the NMJ’s post-synaptic α
   subunit of the nicotinic ACh receptors (nAChR)[152]^44. AM132, but not
   the control oligonucleotide enhanced both AChE and αBTX staining of
   NMJs at 6 h (ANOVA p = 7.38E-13; n = 105 fields, and p = 0.00091;
   n = 67 fields, 3 mice/group) and 24 h post-injection (ANOVA
   p = 0.38E-13; n = 105 fields, and p = 0.00256; n = 71 fields, 3
   mice/group), most likely reflecting membrane-bound AChE-S and nAChR. Of
   note, nAChR is not a target of miR-132; therefore, that AM132 treatment
   induced massive long-lasting increases in nAChR levels at NMJs may
   indicate an indirect feedback response to the modified muscle
   cholinergic tone at this time window, which may minimize the damage to
   muscle functioning. Supporting this notion, NMJs from oligonucleotide
   control-injected mice showed faintly increased αBTX staining but
   non-altered AChE compared to NMJs from control mice ([153]Fig. 6b,c).

Figure 6. AM132 enhances peripheral cholinergic transmission.

   [154]Figure 6
   [155]Open in a new tab

   (a) Schematic presentation of the working hypothesis, where AM132
   functionally masks miR-132 inhibition, leading to AChE-R
   over-expression, blockade of ACh and consequent increase of nAChR.
   (b,c) Double staining of AChE and nAChR in diaphragm neuromuscular
   junctions (NMJs) 6 h and 24 h after IV injection of 3.3 mg/kg AM132,
   control oligo or PBS. Note AM132-induced enhancement of NMJs nAChR and
   AChE levels as detected by immunohistochemistry and α-bungarotoxin
   binding. (a) Histochemical activity by Karnovsky’s staining for AChE
   shows more stained NMJ/Field following 3.3 mg/kg AM132 IV injection.
   (e) Representative TEM micrograph showing characteristic diaphragm NMJ
   ultrastructure following AM132 injection. One out of 20 electron
   micrographs, all comparable.

   Increased AChE protein levels upon AM132 treatment could also reflect
   increased ACh hydrolytic activity. To test this hypothesis, we
   dissected intact diaphragms 24 h post-treatment, stained them with
   acetylthiocholine for catalytically active AChE, and quantified labeled
   NMJ density ([156]Fig. 6d). AM132-treated mice showed 10.9 ± 4.2
   compared to 6.3 ± 2.7 enzyme-stained NMJ/field in each diaphragm of
   control mice (Student’s t-test P < 0.01, n = 3 mice/group), suggesting
   selectively enhanced ACh hydrolytic activity under AM132, albeit at a
   considerably more limited level than the muscle AChE-R mRNA changes.
   There was, however no sign of structural NMJ changes in transmission
   electron microscopy, where NMJs from AM132 and PBS-treated diaphragms
   showed typical and indistinguishable healthy characteristics, and
   demonstrated sustained membrane fold structures and apparently healthy
   mitochondria in AM132-treated mice ([157]Fig. 6e), validating the
   selectivity of the observed differences. The impact of AM132
   suppression of AChE-R thus spanned cortical, muscle and circulation
   effects above and beyond its direct targets.

Discussion

   Injecting the brain-impenetrable AM132 oligonucleotide reduced
   peripheral miR-132 levels, causing massive, splice variant-specific
   AChE-R elevation in peripheral tissues while modifying the brain’s
   miR-132 targets and signaling pathways and suppressing both brain and
   body inflammation, together reflecting reciprocal body-brain signals
   that induced more substantial cortical changes in brain miRs and their
   target transcripts than LPS exposure. Bioinformatics interrogation
   identified AM132-modified neurotransmission and immune-related circuits
   culminating with interrupted cholinergic signaling and suppressed
   pro-inflammatory cytokine transcripts. Elevated AChE-R in brain, serum,
   and muscle was accompanied by re-balanced cholinergic tone in NMJs,
   where co-elevated nAChR could expedite the cholinergic response that
   was weakened by AChE-R elevation. Other examples of splice variant
   differences that modify the protein product location and properties
   include the 3′-longer alternative splicing variants of the
   brain-derived neurotrophic factor (BDNF) and calmodulin-activated
   protein kinase 2 (Camk2a), which like AChE-R present modified cellular
   location[158]^45,[159]^46,[160]^47. By immune-labeling neuromuscular
   junctions, we found AM132 co-elevation of the nicotinic acetylcholine
   receptor and AChE, re-balancing neurotransmission; which led to mild
   muscle incoordination. That the peripheral suppression of miR-132
   globally reduces inflammation suggests a surprising role of miR-132 as
   an AChE splice variant-specific bidirectional body-brain regulator.

   Our SPR measurement demonstrated 3-fold larger effectiveness of AChE-R
   suppression by AM132 compared to AChE-S, yet the in vivo AM132-mediated
   difference between the increase in muscle AChE-R and AChE-S was much
   higher than the SPR measured difference. The SPR measurement however
   was performed in a synthetic buffer, devoid of the additional molecular
   components that exist in vivo. This context-dependent difference and
   the apparent increases in the SC35 splice factor and Sirt1 may together
   explain this contradistinction. Notably, LPS-induced inflammation
   signals reach the brain, induce ‘sickness
   behavior’[161]^13,[162]^16,[163]^17,[164]^40, and change the
   cholinergic status[165]^48. In contrast, oligonucleotides are too large
   to cross the blood-brain barrier[166]^34,[167]^35. Nevertheless,
   unbiased cortical RNA-sequencing revealed larger and inverse impact on
   cortical miR profiles in AM132-treated as compared to differentially
   expressed cortical miRs in LPS-exposed mice. This is compatible with
   miR genes being small and rapidly reacting[168]^15, and functioning as
   multi-target dimmers of entire pathways[169]^8, which may lead to
   anti-inflammatory regulation by sending smaller molecules to the brain
   (e.g. ACh).

   MiR-132 knockout induces modest brain-related functional
   phenotypes[170]^49, but the global cortical reactions to partial and
   temporary changes in miR-132 levels remained largely unknown. To the
   best of our knowledge, this is the first example of a differential
   effect of a miR on specific splice variants and the consequent changes
   in its validated targets and their downstream neuro-modulatory and
   inflammatory cytokines. Specifically, the AChE-S transcript was
   significantly decreased under LPS exposure, indicating limited synaptic
   ACh hydrolysis and therefore exacerbated cholinergic signaling and
   increased anti-inflammatory protection; in contrast, cortical AChE-S
   was increased under AM132, possibly reflecting peripheral body-brain
   cholinergic messages known to accompany stress insults[171]^31. That
   the stress-induced soluble AChE-R variant[172]^7,[173]^50 was
   significantly increased both under LPS and following AM132 may possibly
   reflect general differences in alternative splicing and increases in
   the splicing factor SC35 in these systems[174]^41. At the biochemical
   level, elevated AChE-S activity may suppress the stress-characteristic
   hyper-activation of synaptic neurotransmission[175]^26, whereas AChE-R
   elevation might reduce the cholinergic blockade over brain NFkB. A
   parallel bidirectional change was also observed in NMJs (where both
   nAChR and AChE were increased), suggesting inversely directed checks
   and balances aimed to achieve homeostasis.

   Other quantified miR-132 target transcripts include the transcriptional
   co-activator P300-CBP[176]^22, growth factor HB-EGF, and SIRT1.
   P300-CBP controls both miR-132 expression and innate immune response to
   viral infection. HB-EGF regulates the NFkB pathway in a context-
   dependent manner[177]^23, and may thus interfere with cytokine
   production[178]^51. In cellular injury, miR-132 is upregulated and
   HB-EGF decreases, reducing leukocytes capability to attract chemokines
   and divert the healing process from inflammation towards
   proliferation[179]^52. Similarly, miR-132 levels increased and HB-EGF
   decreased under LPS, but AM132 yielded the opposite outcome, mimicking
   healing conditions. SIRT-1 is another more upstream epigenetic
   modulator[180]^53; its decrease is associated with both liver[181]^25
   and brain[182]^54 inflammation[183]^53, impaired energy
   metabolism[184]^25, and ageing[185]^54. SIRT-1 showed LPS-induced
   decreases and AM132-induced increases, which might protect from
   inflammation[186]^55. Several cytokines, including IL-6, IL-18, IL-1β,
   TNFα and MCP1 were also increased under LPS and decreased under AM132.
   This observation is compatible with the cortical decline of
   anti-inflammatory miRs under LPS and their increase under AM132.

   The profound impact of AM132 on miRNA and mRNA cortical transcripts
   likely reflects the hierarchically high position of miR-132 and its
   downstream targets as regulators of cortical gene expression. Examples
   of the AM132-modulated miRs include let-7b, which regulates miR
   metabolism pathways[187]^56 and was up-regulated under LPS and
   suppressed in AM132-treated brains, supporting the recently reported
   functional link between miR regulation of anxiety and
   metabolism[188]^57. Also, miR-21 and miR-27a are both suppressors of
   inflammation[189]^56 that were down-regulated under LPS and
   up-regulated by AM132, compatible with the exacerbated and suppressed
   inflammation in these two conditions. Overlapping pathways that came up
   as enriched for differentially expressed miRs between AM132 and LPS
   treatments included synaptic transmission and inflammatory signaling.
   We found inflammatory signaling pathways to be primarily affected by
   inversely expressed miRs between LPS and AM132 treatments. However,
   despite the limited LPS-mediated changes in cortical miR expression
   profiles, miRs targeting pathways of the cholinergic system were
   similarly regulated in both the LPS and AM132 treatment groups. Further
   studies will be required to find out if these distinctions reflect
   cholinergic-related compensation processes aimed to re-achieve
   homeostasis, and if this occurs under diverse perturbations.

   AM132 cannot cross the blood-brain barrier[190]^35, which implies that
   its brain impact mainly reflects the outcome of its peripheral
   body-to-brain signals. For example, by targeting peripheral
   AChE[191]^26,[192]^27, miR-132 might elevate ACh levels and activate
   efferent vagal messages. AM132 would reduce miR-132 levels, possibly
   limiting exosomal transport of miR-132 to the brain. Although AM132
   treatment caused no change in brain miR-132 levels, it led to
   upregulated AChE expression and hydrolytic activity in both brain and
   muscle. Immunohistochemistry and histo-chemical staining further
   confirmed elevated AChE labeling in otherwise healthy-looking NMJs from
   AM132-treated mice within the short 24 hours span of our tests. The
   numbers of NMJs stained for AChE were increased, perhaps reflecting
   somewhat more densely packed AChE-S in these synapses; and nAChRs was
   likewise upregulated in NMJs, possibly balancing the massive increase
   in muscle AChE-R, which competes with NMJ activation. We also
   identified a mild AM132-induced fatigue, i.e. mice falling off earlier
   in the third trial of Rotarod tests, but no overt changes were observed
   in TEM muscle morphology and NMJ structures. The imbalanced nAChR/AChE
   state might hence explain this mild muscular weakness[193]^58. In
   conclusion, we report inverse cortical response to LPS and the
   inhibition of miR-132 in the periphery, pointing to a body-to-brain
   anti-inflammatory que that was reciprocal to miR-132 induction by
   peripheral LPS. Together, that AM132 treatment not only affected
   peripheral tissues such as NMJs, but also the brain, underlines the
   high- hierarchy regulation by miR-132 in the context of
   neuro-inflammation[194]^13,[195]^16,[196]^17, and lead to the
   identification of selectivity of miR-132 to AChE-R over AChE -S.

Methods

Experimental design

Design and synthesis of AM132

   The 2′-O-methyl antisense oligonucleotides were prepared on an Expedite
   8909 Nucleic Acid Synthesizer (Applied Biosystems) on a
   phosphorothioate backbone, where the non-bridging oxygen was replaced
   with a sulfur atom. The AM132 Sequence was: 5′ CGA CCA UGA CUG UAG ACU
   GUU A 3′. A scrambled oligonucleotide with no apparent match in the
   mouse genome was composed of a mixture of the same bases
   ([197]Supplementary Fig. S1).

Ethics Statement

   All methods were carried out in accordance with the Animal Care and Use
   Guidelines of the Hebrew University (NS-10-12745-5 for injection
   experiments) which follow the National Research Council (US) Guide for
   Care and Use of Laboratory Animals. All experimental protocols were
   approved by the University Ethics Committee for Maintenance and
   Experimentation on Laboratory Animals, The Hebrew University,
   Jerusalem, Israel.

Animals

   Male 4 months old Balb/C mice were intravenously (i.v.) injected with
   3.3 mg/kg AM132 or a scrambled oligonucleotide or PBS as controls.
   After 24 hours all mice were euthanized and their brains, gastrocnemius
   muscle, diaphragm muscle and serum were collected and used for the
   following experiments: RNA extraction, small RNA-sequencing, ACh
   hydrolytic AChE activity tests, electron microscopy and qRT-PCR
   measurements. Protein content was measured by the Lowry assay.
   LPS-induced inflammation experiments were as described before[198]^48
   and frontal cortex RNA was used for miRs profiling, and for quantifying
   mRNA transcript levels.

Surface plasmon resonance

   Surface plasmon resonance experiments[199]^59 were conducted using a
   Biacore 3000 instrument (Biacore AB, Uppsala, Sweden). Standard buffer
   HBS (10 mM HEPES, 150 mM NaCl, 3.4 mM EDTA, 0.005% surfactant P20, pH
   7.4) was used for the analyses carried out at 25 °C. Biotinylated
   oligonucleotides were dissolved in 100% 1,1,1,3,3,3-
   hexafluoro-2-propanol to 1 mM, diluted (1:5000) into 10 mM sodium
   acetate pH 5.0 and injected at 10 ml/min. The AChE-R and AChE-S, SIRT1,
   and miR-132 binding site oligonucleotides were diluted in buffer
   (0.3125, 0.625, 1.25, 2.5, 5 and 10 mM), captured on the chip or
   injected over the flow cells for 2 min at 10 ml/min, with 5-min
   association and 5-min dissociation, except for the highest
   concentration that was allowed to dissociate for 1 h. The sensograms
   were double-referenced and were fit using a mathematical model of a
   simple 1:1 interaction (Scrubber 2 soft- ware).

RNA extraction

   RNA extraction from tissue samples was carried out using miRNeasy
   (Qiagen, Valencia, CA, USA) as per the manufacturer’s instructions.
   DNAse treatment was applied, RNA concentration and integrity confirmed
   by Nanodrop and gel electrophoresis respectively.

Sequencing data analysis and normalization

Construction of small RNA libraries and Sequencing

   Construction of libraries was done using the NEB-Next multiplex Small
   RNA Preparation Kit (New England BioLabs, USA) according to the
   manufacturer’s instructions using 150 ng RNA per sample. Next,
   individual libraries were adjusted to 2 nM and pooled before
   denaturation and dilution according to Illumina’s instructions. cDNA
   libraries were prepared from gel-extracted size-selected RNA with a
   median length of 140 nucleotides. Library quality for miRs was
   determined by TapeStation and BioAnalyzer. APhiX control library was
   spiked in, to determine the sequencing error rate (<1%). The diluted
   libraries (20 pM) were loaded onto a 4-lane flow cell for cluster
   formation using the TruSeq™ SR Cluster Kit v3 (Illumina). Sequencing
   was performed on an Illumina NextSeq 500 system. Libraries were
   sequenced from the 5′- end using a read length of 30 nt and results
   converted into fastq files with Casava 1.8.2 (Illumina) using BaseSpace
   software and downloaded as Fastq files for alignment and further
   analysis.

Alignment and normalization

   Raw sequence counts were processed in a bioinformatics pipeline
   including a quality control check (FastQC), adapter trimming,
   consolidation of reads across samples, and alignment of reads against a
   reference database (miRBase v. 21) using miRExpress 2.0. Sequences were
   normalized using the RLE method (DeSeq2) as well as by assessing
   counts-per-million (CPM) values to correct for library size. Quality
   check for genomic homology across groups was analyzed by PCA binomial
   plot and unbiased hierarchical clustering ([200]Supplementary Fig. S3).

Expression analysis and statistics

   Differential expression was analyzed using DeSeq2[201]^60 using
   likelihood ratio tests for both LPS and AM132 datasets. Differences
   were deemed significant for p < 0.05 after FDR correction and filtered
   for a mean greater than 15 CPM across all samples. Additionally, a
   non-parametric permutation test (10,000 permutations) was run
   independently for all pairwise samples in each group. Differentially
   expressed miRs of the LPS and AM132 groups were analyzed.

Pathway enrichment Analysis

   Pathway enrichment analysis was done using miRpath v. 3.0 (tarbase
   v.7)[202]^61. We identified overlapping miRs and enriched pathways
   between the LPS-exposed and AM132-treated groups, with p < 0.05 after
   FDR correction. Similarly and inversely regulated overlapping pathways
   targeted by the differentially expressed miRs were clustered via an
   unsupervised procedure using the fraction of overlapping genes in these
   pathways (number of genes shared in a pathway between inversely and
   similarly regulated miR groups divided by the supremum analysis of miR
   targets in the pathway). The clustered pathways were then deemed to
   significantly separate between inflammatory and signaling genes using a
   two-way binomial test. Mediation type analysis was done using
   permutation tests (n = 10,000) for inflammatory and neurotransmission
   signaling targets.

qRT PCR

   RNA samples were used for synthesis of cDNA, using Quanta cDNA
   synthesis kit for mRNA and qScript microRNA cDNA Synthesis Kit for miRs
   as per manufacturer’s (Quanta Biosciences) protocol. Quantitative real
   time- PCR (qRT-PCR) was performed using ABI prism 7900 HT and SYBR
   green master mix (Quanta Biosciences) with or without ROX for mRNA and
   miR transcripts, respectively. The mRNA primer sequences are detailed
   in [203]Supplementary Table S4. Serial dilution of samples was used to
   evaluate primers efficiency. Results were normalized to the expression
   level of actin. For miRs, PerfeCTa microRNA assay primers (Quanta
   Biosciences) were used. Results were normalized to the expression of
   snoRD47. Fold change values for both miRs and mRNAs were calculated
   using the ΔΔCt method.

AChE activity

   AChE activity in serum, brain and gastrocnemius muscle was determined
   by microtiter plate assay by measuring ATCh (1 mmol/L; Sigma-Aldrich,
   St. Louis, MO, USA) hydrolysis rate[204]^62 after pre-incubation with a
   specific BChE inhibitor (iso-OMPA (Sigma-Aldrich), in the dark at a
   concentration of 500 μmol/L. Enzyme activities were calculated using
   405 nm absorbance value for 5-thio-2- nitrobenzoate produced from
   dithionitrobenzoate reaction with the hydrolysis product, thiocholine
   (13,600 [mol·L]/cm). The non-enzymatic breakdown of substrate was
   subtracted from the total rate of hydrolysis.

Rotarod test for measuring muscle fatigue

   An accelerating Rotarod (Ugo Basile, Comerio VA, Italy) was used for
   testing muscle fatigue. The Rotarod was set to accelerate from 4 to
   40 rpm over 5 min. Mice were then tested for 3 successive times with a
   minimum of 2 min elapsed between trials.

Immunocytochemistry and α-bungarotoxin double staining

   Diaphragms were isolated as described before[205]^63 and were fixed in
   4% paraformaldehyde in PBS for 1 h., permeabilized in 0.1%Triton X-100
   for 20 min, blocked in 5% horse serum for 4 h., and then incubated with
   primary antibody anti-AChE (1:200, Santa Cruz, CA, USA) overnight at
   4 °C. Cy2-conjugated secondary antibodies were then added for 4 h at
   R.T. During the last wash 10 nM α-bungarotoxin-Cy3 conjugate (Sigma,
   St. Louis, MO, USA) was added for 30 min to mark α-subunits of the
   neuromuscular nicotinic receptors. Coverslips were rinsed twice with
   PBS and once with DDW and were mounted for electron microscopy
   (FluoroMount-G, Electron Microscopy Sciences, Washington, PA, USA).
   Slides were scanned using the Olympus FV-1000 confocal microscope as
   above. Image Analysis was performed using the ImageJ-software (National
   Institutes of Health, Maryland, USA).

Histochemical AChE activity staining of diaphragms

   AChE activity can be visualized in situ by staining with
   acetylthiocholine in the presence of copper and ferrous ions, which
   together form a brown precipitate[206]^64. The mice were sacrificed and
   dissected to expose the diaphragm muscle. After two washes with PBS,
   the diaphragms were incubated in Karnovsky’s staining solution
   including acetylthiocholine at R.T. Within 20 min, we could visualize
   NMJ numbers, size and AChE activity. AChE staining intensity for each
   NMJ was measured by applying a visual scale of 1-5 for 15 different
   fields of 100 um square each per diaphragm, and average ± SD intensity
   values were calculated for each diaphragm. The number of NMJs, their
   size and the intensity of staining were compared.

Electron Microscopy

   Diaphragms were removed with the surrounding ribs to prevent folding,
   washed twice with PBS and were fixed with 2.5% glutaraldehyde and 2%
   formaldehyde in 0.1 M cacodylate buffer (containing 1% OsO4 and 1.5%
   potassium ferrycyanide) solution, pH 7.4, 2 h at R.T., dehydrated in a
   graded ethyl alcohol series followed by propylene oxide, and embedded
   in Agar 100 resin. Polymerization was carried out at 60 °C for 48 h.
   Ultrathin sections were cut on an LKB III ultratome, mounted on copper
   200 mesh thin bar grids and stained with uranyl acetate and lead
   citrate. Pictures were captured on a Tecnai-12 electron microscope
   (Phillips, The Netherlands) equipped with Megaview II CCD camera and
   3.0 analysis software (Soft Imaging System, GmbH, Germany).

Statistical analysis

   Ellman’s assay and qRT PCR results were analyzed by T-test followed by
   Tukey-test and compared with controls. Microscopy and behavior tests
   were analyzed by ANOVA. Sequencing statistics analysis details are
   mentioned in the corresponding Methods section. Results are expressed
   as mean ± S.E.M, with p < 0.05 considered statistically significant.

Additional Information

   Accession codes: All of the sequencing data has been uploaded to Gene
   Expression Omnibus under accession code SUB1926319.

   How to cite this article: Mishra, N. et al. Antisense miR-132 blockade
   via the AChE-R splice variant mitigates cortical inflammation. Sci.
   Rep. 7, 42755; doi: 10.1038/srep42755 (2017).

   Publisher's note: Springer Nature remains neutral with regard to
   jurisdictional claims in published maps and institutional affiliations.

Supplementary Material

   Supplementary Information
   [207]srep42755-s1.pdf^ (520.5KB, pdf)

Acknowledgments