Targeting Galectin 3 to Counteract Spike-Phase Uncoupling of Fast-Spiking Interneurons to Gamma Oscillations in Alzheimer’s Disease


 Background: Alzheimer’s disease (AD) is a progressive multifaceted neurodegenerative disorder for which no disease-modifying treatment exists. Neuroinflammation is central to the pathology progression, with evidence suggesting that microglia-released galectin 3 (gal3) plays a pivotal role by amplifying neuroinflammation in AD. However, possible involvement of gal3 in the disruption of cognition-relevant neuronal network oscillations typical of AD remains unknown. Methods: Here, we investigate the functional implications of gal3 signaling on cognition-relevant gamma oscillations (30-80 Hz) by performing electrophysiological recordings in hippocampal area CA3 of wild-type (WT) and 5xFAD mice in vitro. Results: Gal3 application decreases gamma oscillation power and rhythmicity in an activity-dependent manner and is accompanied by impairment of cellular dynamics in fast-spiking interneurons (FSN) and pyramidal cells (PCs). We found that gal3-induced disruption is mediated by the gal3-carbohydrate-recognition domain and prevented by the gal3 inhibitor TD139, which also prevents Aβ42-induced degradation of gamma oscillations. Furthermore, we demonstrate that 5xFAD mice lacking gal3 (5xFAD-Gal3KO) exhibit WT-like gamma network dynamics.Conclusions: We report for the first time that gal3 impairs cognition-relevant neuronal network dynamics by spike-phase uncoupling of FSN inducing a network performance collapse. Moreover, our findings suggest gal3 inhibition as a potential therapeutic target to counteract the neuronal network instability typical of AD and other neurological disorders encompassing neuroinflammation and cognitive decline.

1.25 NaH 2 PO 4 , 3.5 KCl, 1.5 MgCl 2 , 1.5 CaCl 2 . The holding chamber was continuously supplied with humidi ed carbogen gas (5% CO 2 , 95% O 2 ), held at 37°C during slicing and then allowed to cool down to room temperature for at least 1 hr prior to start of any experiment.
Electrophysiological recordings LFP recordings were performed either in interface-or submerged-type recording chambers in the hippocampal CA3 area with borosilicate glass microelectrodes lled with standard ACSF and placed in the stratum pyramidale. For LFP recordings in an interface con guration, slices were continuously supplied with aerated ACSF at a rate of 4.5 ml/min at 34°C and the chamber was continuously supplied with humidi ed carbogen gas. Single-cell recordings were carried out either concomitantly with LFP or on their own in a submerged recording chamber continuously supplied with aerated ACSF at a perfusion rate of 3 ml/min at 34°C. Depending on the con guration of patch-recordings (voltage-or current-clamp), different intracellular solutions were used. Action potentials (AP) and excitatory postsynaptic currents (EPSCs) from PCs and FSN were recorded in whole-cell mode with an internal recording solution for current-clamp con guration containing (in mM): 122.5 K + -gluconate, 8 KCl, 2 Mg 2+ ATP, 0.3 Na + GTP, 10 HEPES, 0.2 EGTA, 2 MgCl. An internal recording solution for voltage-clamp containing (in mM) 140 CsMetSO 4 , 2 Mg 2+ ATP, 0.3 Na + GTP, 10 HEPES, and 0.6 EGTA was used for PC-inhibitory postsynaptic currents (IPSC) recordings. For each internal solution pH was set to 7.2-7.3 and osmolarity to 270-280 mosmol/l. PC-and FSN-EPSCs, and PC-IPSCs were recorded in voltageclamp con guration holding the membrane potential at -70 mV or 0 mV, respectively. FSN-membrane potential (Em) and ring threshold were recorded in current-clamp con guration at resting membrane potential in gapfree for Em, or by applying current steps of 10 pA starting at -70 mV holding membrane potential for ring threshold.
PCs and FSN were visually identi ed under an upright microscope using IR-DIC microscopy (Axioskop, Carl Zeis AG, Göttingen, Germany) based on their morphology and location in the hippocampal str. pyramidale and str. radiatum, respectively. PCs and the interneuron subtype (FSN) were further con rmed as previously described by applying different stimulation protocols to verify the cellular population type by their unique electrophysiological characteristics [24,44,46,47]. For recordings of PC or FSN in whole-cell con guration access resistance was monitored throughout the experiment. Cells were excluded from the study if more than 20% of change was observed.
Gamma oscillations were induced by applying 100 nM kainic acid (KA, Tocris) [43,44] and allowed to stabilize prior to recording of any cellular/LFP parameter. LFP recordings in the interface-type recording chambers were acquired with a 4-channel M102 ampli er (University of Cologne, Germany). Data was sampled at 10 kHz, conditioned using a HumBug 50 Hz noise eliminator (Quest Scienti c), low-pass ltered at 1 kHz, digitized using a Digidata 1440A (Molecular Devices, CA, USA) and stored using pCLAMP 9.6 software (Molecular Devices, CA, USA). LFP and patch-clamp recordings in submerged-type recording chamber were acquired with a patch-clamp ampli er (Multiclamp 700B) using pCLAMP 10.4 software (Molecular Devices, CA, USA). LFPs recorded in the submerged-type recording chamber were also conditioned using a HumBug 50 Hz noise eliminator (Quest Scienti c). All signals recorded in submerged con guration were low-pass ltered at 1 kHz, acquired at 5 kHz, digitized and stored using Digidata 1322A and pCLAMP 10.4 software (Molecular Devices, CA, USA).

Data analysis
For power spectral density plots, fast Fourier transformations (segment length 8192 points) were calculated from 60 s-long LFP recordings using Axograph X (Kagi, Berkeley, CA, USA). Gamma oscillation power was calculated by integrating power spectral density between 20 and 80 Hz using KaleidaGraph. Peak frequency of gamma oscillations was obtained using Axograph X. For further analysis of LFP recordings, the signals were ltered in both directions using a band-pass Butterworth lter set to 20-50 Hz for autocorrelation analysis of gamma oscillations. Normalized autocorrelations were performed using Matlab custom-written routines. The coe cient of rhythmicity (Cr) was calculated from the autocorrelograms as a measure of the quality of gamma oscillations [43,48] or EPSCs and IPSCs rhythmicity [48]. It was de ned as Cr=(α-β)/(α + β) where (1 + α) and (1 + β) corrections were applied. The value of α corresponds to the value of the height of the second peak and β to the rst trough in the autocorrelogram counting the rst peak at zero lag. Cr ranges between 0 and 1 with higher coe cient values denoting more rhythmic activity and thus better rhythm quality. Only recordings with Cr ≥ 0.01 were included in the study.
FSN-and PC-spike-phase coupling with corresponding gamma-LFP (60 s-long) were analyzed using a custommade Matlab routine as previously described [43,44,48]. LFP traces were ltered using a band-pass Butterworth lter (20-40 Hz, both directions) and AP were detected using an amplitude threshold set to 30-50% of the total AP amplitude. The instantaneous phase-angle of gamma oscillations for each AP was determined using a Hilbert transform. Firing window was analyzed by distributing phase-angles of all AP and gamma oscillation-phases (in radians) in polar-plots with the peak of the oscillation cycle corresponding to 0 π and the trough to π in the polar plots. For each AP, a vector of length=1 was assigned at the angle corresponding to the phase of the LFP. An averaged resultant phase-density vector was used to describe the preferred phase of ring (phase-angle) and how recurrent the ring in that angle was (vector length). Thus, a larger vector denotes a stronger spike-phase coupling/more synchronized AP ring relative to the network gamma oscillation. Vector length is shown normalized by the total number of AP for each cell recorded and the summary is plotted for each condition. All ring distributions were assessed for uniformity setting the inclusion criteria at p<0.05 when performing a Rayleigh's test.

Drugs and chemicals
All chemical compounds used in intracellular and extracellular solutions were obtained from Sigma-Aldrich Sweden AB (Stockholm, Sweden). KA was dissolved in miliQ water. Human gal3 protein was produced by the protein production unit LP3 at Lund University. Brie y, the gal3 protein was produced in strain Escherichia coli TUNER(DE3) / pET3c-hum-gal3 grown in LB medium, 18°C, 250 rpm with 1 mM IPTG overnight. After cell lysis and ultracentrifugation, gal3 was puri ed on a 20 ml lactocyl-sepharose column. Peak fractions containing gal3 were pooled and dialyzed against PBS, pH 7.4. Human gal3 was analyzed on a Criterion TGX AnykD precast SDS-PAGE gel (Bio-Rad) stained with Bio-Safe Coomassie (Bio-Rad). The purity was estimated to >95%.

Statistical analysis
All statistical analysis was performed using GraphPad Prism 8.0 either in absolute values or normalized data (to 5 min of stable baseline) when appropriate. To minimize the variation between slices, a paired statistical design was used when compounds were washed-in sequentially and unpaired when treatments were performed independently of each other. Two-or one-sided designs were used as appropriate. Prior to statistical analysis all data were subjected to detection of outliers that were removed using the ROUT method. This was followed by tests for normality distribution and variance similarity between groups. Wilcoxon's signed rank sum test was used when reporting statistical differences between control conditions and compound effects in the nonparametrically distributed data. A paired Student's t test was used when the data were distributed normally. For multiple comparisons, ordinary one-way ANOVA followed by Holm-Sidak's multiple comparisons test or Kruskal-Wallis test followed by Dunn's multiple comparisons were carried out depending on the parametric or nonparametric nature of the data, respectively. Results are reported as mean ± SEM and signi cance levels were set as follows: * p<0.05, ** p<0.01, ***p<0.005, ****p<0.0001.

Gal3 induces degradation of gamma oscillations via its carbohydrate recognition-domain
We rst evaluated the effect of gal3 on gamma oscillation power and rhythmicity. For this purpose, hippocampal slices of WT mice were incubated for 15 min with 1 µM gal3 and transferred to an interface-type recording chamber. Gamma oscillations were elicited with 100 nM KA and allowed to stabilize for 30 min following which gamma-local eld potential (gamma-LFP) was recorded and analyzed [43,45]. Incubation with gal3 prior to gamma oscillation induction resulted in a drastic decrease of gamma oscillation power (Fig. 1C, Prominent cellular actions of gal3 have been reported to be mediated by its characteristic carbohydrate recognition-domain (CRD) [26,32]. Thus, we investigated whether the deleterious effects of gal3 on network oscillations were driven by its CRD. For this purpose, hippocampal slices of WT mice were incubated for 15 min with 1 µM gal3-isolated CRD as well as with 1 µM R186S-Gal3 and then transferred to an interface-type recording chamber. R186S-Gal3 is a mutated version of gal3 with reduced a nity for many glycoproteins and for the disaccharide LacNAc, which is the most common minimal galectin-binding moiety in glycoproteins [50,51]. Recordings performed 30 min after the induction of gamma oscillations revealed that the effects of gal3-CRD parallel those of intact gal3: Decrease of gamma power (Fig. 1E) and impairment of gamma oscillation rhythmicity (Fig. 1F). Notably, R186S-Gal3 did not affect either gamma power (Fig. 1E) or rhythmicity (Fig. 1F).
Thus, gal3 induces a drastic degradation of gamma oscillations in the hippocampal network through its CRD and this effect is prevented with highest e cacy by 10 µM TD139.
Gal3 disrupts action potential phase-lock of fast-spiking interneurons to gamma oscillations in a concentrationand exposure time-dependent manner We evaluated the effect of gal3 on fast-spiking interneuron (FSN) phase-lock to ongoing gamma oscillations.
To this end, we performed concomitant recordings of LFP and whole-cell patch-clamp recordings of FSN activity in a submerged-type recording chamber during the activated network state (gamma oscillations). Thirty minutes after gamma oscillation induction, 5 min of control gamma and FSN ring were recorded and subsequently gal3 was bath-applied. Application of 1 µM gal3 for 30 min did not induce a decrease of gamma oscillation power ( Fig. 2A, B) indicating that its effect may depend on the activation state of the network.
Likewise, FSN ring phase-lock and AP ring rate were not affected by the application of 1 µM gal3 to the Gal3 impairs inhibitory and excitatory synaptic transmission during ongoing gamma oscillations Proper coordination of excitatory and inhibitory synaptic activities within neuronal networks is crucial for circuit entrainment and emergence of brain rhythms [9,11,46,52,53]. Here, we have observed that gal3 impairs the ring coordination relative to gamma oscillations of crucial neuronal populations such as inhibitory GABAergic FSN and excitatory glutamatergic PCs. We proceeded to investigate the effect of gal3 on the excitatory and inhibitory synaptic inputs onto PCs as well as excitatory drive onto FSN. For this purpose, we recorded excitatory and inhibitory postsynaptic currents (EPSCs and IPSCs, respectively) from PCs (see methods) 40 min after 2 µM gal3 application during ongoing gamma oscillations (Fig. 3).
Analysis of EPSC charge transfer showed that gal3 signi cantly decreased the overall excitatory input onto PCs Neuronal mechanisms underlying the activity-dependence of gal3-induced impairment of functional network dynamics Here we have observed signi cant reductions of gamma power and rhythmicity when slices are exposed to gal3 (1 µM) prior to gamma induction (see Fig. 1). However, when the network is under stable entrainment into the gamma rhythm (30 min after gamma induction by 100 nM KA), an increased concentration of gal3 is needed to observe comparable disruption (see Fig. 2). To investigate the effects of gal3 on FSN basal activity and its possible consequences for the network function during gamma oscillations, we set out to perform recordings of FSN activity concomitantly with LFP rst in the basal state and then during subsequent gamma oscillation induction in area CA3 of hippocampal WT mouse slices (Fig. 4A). Gal3 or gal3 plus gal3 inhibitor (TD139) were applied for 15 min prior to gamma oscillation induction and remained in the bath solution throughout the subsequent 30 min of rhythm establishment and stabilization. Basal state recordings also served as control to ensure that neither the whole-cell patch-clamp con guration nor the bath ACSF induced functional deviations during the total 45 min-long recordings (Fig. 4A).
Recordings of FSN membrane potential (Em) in quiescent state revealed that gal3 hyperpolarizes FSN Em 15 min after bath application (Fig. 4B, D). Co-application of gal3 and TD139 prevents this gradual hyperpolarization (Fig. 4B, E), which is also absent in the control condition (ACSF, Fig. 4B We also observed a drastic decrease of the ring rate at basal state following gal3 wash-in ( to gamma oscillation induction, which then overrides that hypoactivity to a limited extent. This is evident in the fact that concomitant LFP recordings revealed that the neuronal network fails to generate gamma oscillation power similar to that seen in control conditions (Fig. 4F, G). Such gal3-induced impairment of rhythmic network activity is counteracted by the presence of TD139 (Fig. 4F, G). Additionally, gamma oscillations induced in slices previously exposed to gal3, showed lower peak frequency compared to gamma oscillations recorded in control conditions or in the presence of the gal3 inhibitor (Fig. 4F, H).
Treating slices with gal3 led to a drastic decrease of FSN AP ring rate during ongoing gamma oscillations To investigate a possible effect of gal3 in some of the gal3 activation/responses-associated genes related to microglial AD-phenotype, we assessed the expression of Trem2, Tlr4, Clec7a, and Cx3cr1 transcripts in some of the recorded slices. Notably neither Trem2, Clec7a nor Tlr4 changed their expression in slices treated with gal3 prior to gamma oscillation induction (Fig. S6). Conversely, there was a decrease of the ∆Ct of the astrocyte marker Gfap in slices treated with gal3 in the presence of the inhibitor TD139. ∆Ct of the chemokine receptor Cx3cr1 revealed that KA induces a decrease of this parameter while neither gal3 nor gal3 upon inhibition increased the ∆Ct (Supplementary Figure 6, Additional le 2).
Interference with gal3-signaling prevents the disruption of gamma oscillations in two different AD models Acute application of physiological concentrations of Aβ42 (50 nM) to hippocampal slices causes a dramatic and rapid impairment of gamma oscillations [43-45, 47, 55, 56]. Also, cognition-relevant brain rhythms have been reported to be disturbed in several AD mouse models [24,[57][58][59][60][61]. We investigated whether inhibition of gal3 signaling in two AD-related mouse models is able to prevent the disruption of gamma oscillations, which is linked to the cognitive impairment in AD.
Firstly, we performed recordings of gamma oscillations in area CA3 of WT mice hippocampal slices previously incubated with Aβ42 or Aβ42 + TD139 ("acute Aβ mouse model"). Recordings were performed in the same con guration as for the experiments in Fig. 1 (interface-type recording chamber, gamma oscillations induced after 15 min of incubation with 50 nM Aβ42 or 50 nM Aβ42 + 10 µM TD139 and recorded 30 min after rhythm induction). In slices pre-incubated with Aβ42, gamma oscillation power and rhythmicity were signi cantly decreased, in accordance with previous reports [43,45,48,55,56]. Surprisingly, co-incubation with TD139 fully prevented this Aβ42-induced functional impairment (Fig. 5A, C, D).
Secondly, we tested whether gamma oscillations were impaired in a familial AD mouse model also ("chronic 5xFAD In the present study, we observed that the hippocampal CA3 network of 6 months-old 5xFAD mice exhibits a signi cant decrease of gamma oscillation power and a shift of the peak gamma frequency towards lower values. Notably, in age-matched 5xFAD-Gal3KO mice, gamma oscillation power and peak frequency were unaffected and at WT littermate control levels. Gal3 deletion did not cause further changes in both parameters (Fig. 5B, E, and F). Thus, gal3 parallels the toxic effects of Aβ42 on gamma oscillations as previously reported [43-45, 48, 55, 56]. This toxicity was prevented by gal3 inhibition (see Fig. 1) while gal3 deletion prevents functional impairment of the gamma network rhythm in the 5xFAD AD mouse model. effective not only against neurodegenerative processes, but also in rescuing normal neuronal signaling, cognition-relevant brain rhythms and so, presumably, having a positive effect on cognitive decline.

Discussion
Synchronization of fast neuronal activity between PCs and FSN within hippocampal CA3 circuits give rise to the emergence of periodic uctuations of the LFP in the gamma frequency-band (30- [32]. Particularly, binding to TREM2 and TLR4 has been described to be mediated by gal3 CRD [26,32]. Accordingly, the consistent prevention of gal3-mediated disruption of neuronal and network function observed in the presence of the gal3 inhibitor TD139 reinforces our ndings. TD139 is a 3,3'-Bis-(4-aryltriazol-1-yl) thio-digalactoside gal3 inhibitor with high a nity for the gal3-CRD. In this study we tested whether the drastic disruption of gamma oscillations observed following exposure of the hippocampal network to gal3 is accompanied by changes in the expression of genes related to microglial activity. We observed changes in the expression of very few genes, which could be ascribed to the relatively short exposure time (45 min total, see Fig. 5) that appears long enough to prevent an e cient induction of gamma oscillations but too short to trigger a signi cant change of gene expression. However, the overall quanti cation of the expression of the listed transcripts in the present study should just be taken as a descriptive clue and future experiments should be performed to further assess a wider range of microglia activation-related genes with longer gal2 exposure-times.
The signi cant electrophysiological changes to neurons and networks observed here appear to rely on a mechanism linked to the cell membrane (i.e., a particular microglial receptor), probably inducing an early microglial activation with further consequences if the application lasts longer as expected from the timedependence shown in Fig. 1, and as it may happen in vivo during sustained gal3 release. Again, the overall preventive effect of TD139 reinforces this notion. TD139 mainly acts extracellularly within the time of application employed here. The inhibitor does not reach the intracellular compartments unless incubation/wash-in lasts 24 hours or more [70].
Furthermore, we found that gal3 is less e cient in disrupting gamma oscillations if the network is already properly entrained into the gamma rhythm (see Fig. 2). However, if the network is treated with gal3 prior to FSN engagement into strong spike-phase coupling, gal3 prevents the proper establishment of a coordinated activity that leads to gamma oscillations of physiological relevance (see Fig. 4). In an AD scenario, it is tempting to hypothesize that at a certain point during very early pathology progression, cognition-relevant neuronal networks preserve their functionality due to homeostatic mechanisms (a large concentration of gal3 is needed to disrupt cellular and neuronal normal activities). As the pathology progresses, such homeostatic mechanisms become overwhelmed, perhaps even deleterious (i.e., microglia shifts to damage-associated microglia (DAM) phenotype (see Frere and Slutsky, 2018) [21]). Then, a progressively weakened neuronal network appears more susceptible to gal3 that could reach the neurons/synapses by diffusing throughout the brain parenchyma once secreted by the already activated microglia (a lesser gal3 concentration drives major dysfunction).
A plausible explanation for the drastic gal3-induced effects observed could be found in the complex purinergic signaling in the neuron-microglia crosstalk. It has been proposed that the initial increase of extracellular adenosine to levels far greater than reached in physiological conditions initially leads to a burst of adenosine receptor 1 (A 1 R)-mediated inhibition, and the continuous massive over ow of extracellular adenosine then overcomes the restricted activation of A 2A Rs, which results in a predominant role of A 2A Rs in the development of neurodegeneration [71]. In line with this proposal a paramount impact of ATP/adenosine signaling on hippocampal circuitry function has been observed. At mossy ber-CA3 synapses microglia-derived ATP differentially modulates synaptic transmission and short-term plasticity through activation of presynaptic P2X4 receptors and A 1 R, respectively, with the latter following its extracellular conversion to adenosine [72].
Additionally, blockade of A 2A Rs prevents lipopolysaccharides-induced impairment of long-term potentiation in rat in vivo, by counteracting the shift of microglia towards a pro-in ammatory phenotype [73]. Notably, A 2A R is upregulated in APP/PS1 mice model [74] as well as in cortical areas [75] and the hippocampal formation of AD In this study we found a drastic decrease of both excitatory and inhibitory drive onto PCs as well as a decrease of excitatory input onto FSN. In this regard, hippocampal A 1 Rs are known to hyperpolarize FSN, reduce the excitability of PCs and interneurons, and also reduce neurotransmitter release [79,80]. Changes in neurotransmitter release are mostly associated with changes of frequency of synaptic events, here observed as a decrease of EPSCs onto PCs (see Fig. 3B) and FSN (see Supplementary Figure 5, Additional le 2) in the presence of gal3. Notably, gal3 reduced the occurrence of larger IPSC components in PCs probably as a re ection of the observed FSN impairment since the major perisomatic inhibition of PCs is driven by FSNs [8,9].
The overall collapse of the network was also observed in the gal3-induced increased variability of the phase relation between both EPSCs and IPSCs with the corresponding LFP-gamma as well as the loss of the rhythmicity of the postsynaptic currents. This loss of rhythmicity of cellular electrical events likely accounts for the degradation of the network rhythm since the generation of gamma oscillations depends on balanced excitatory and inhibitory interplay [11]. However, due to the diverse evidence of purinergic signaling in the modulation of the operational capacity of the hippocampal circuitry, and commonalities found in our study regarding gal3-induced neuronal network collapse, a possible underlying mechanism involving ATP/adenosine receptor activation deserves further research without excluding an indirect involvement of astrocytes, nitric oxide and metabolic arrestment [30,31,76, 81] and a direct effect of gal3 on PCs and FSN.
Finally, there is mounting evidence suggesting that in AD synaptic and network failure starts long before the establishment of Aβ42 depositions into solid plaques and manifestation of cognitive decline expression [21,59,[82][83][84]. Interestingly, in the APP/PS1 mouse model associative long-term synaptic plasticity is impaired in CA3 PCs at the early onset of AD-like features due to the postsynaptic activation of upregulated A 2A R [74]. Moreover, focal glial activation has been reported to precede amyloid plaque deposition in APP transgenic mice associated with a vicious cycle of APP proteolytic cleavage that gives rise to soluble and amyloidogenic immunostimulatory mediators [27]. Using a proteomic approach, it has been found that immune alterations in microglia in 5xFAD mice are active before plaque deposition [85].
Recently, the notions on direct involvement of microglia-released gal3 and Aβ cross-seeding agents in plaque formation has been validated [26, 86, 87]. Particularly, the 5xFAD mouse model lacking gal3 (5xFAD-Gal3KO) fails to develop prominent Aβ plaques and cognitive impairment typical of the 5xFAD model at 6 months-of-age [26]. Here we found that the underlying functional reason explaining these previous reports could be the signi cant degradation of gamma oscillations (observed in the 5xFAD model at 6 months-of-age) that is absent in the age-matched 5xFAD mice lacking gal3 (see Fig. 5). 5xFAD mice also showed a signi cant slowing of gamma oscillation-frequency compared to WT control that was not evident in the age-matched 5xFAD-Gal3KO, which retained values similar to WT. Interestingly, slower gamma oscillations have been observed also in the CA3 area of organotypic hippocampal cultures under microglial priming with interferon gamma [88]. Inhibition of gal3 by co-application of TD139 prevents the decrease of both gamma oscillation power and frequency in the acute Aβ application model (see Fig. 5).

Conclusions
In summary, here we report for the rst time that gal3 -a proposed central microglial/neuroin ammatory regulator in AD -causes degradation of cognition-relevant neuronal network dynamics and their underlying neuronal synchronization mechanisms. These impairments are mediated by the gal3-CRD and prevented by the gal3 inhibitor TD139 in a dose-dependent manner. Additionally, gal3 prevents neuronal network entrainment into proper gamma rhythm. Such disruption is accompanied by the impairment of FSN-and PC-gamma spikephase locking and disturbances of excitatory and inhibitory synaptic transmission. Interestingly, we found a possible functional link for gal3 to AD since 1) TD139 prevents Aβ42-induced degradation of gamma oscillations ex-vivo and 2) gamma oscillations are impaired in the CA3 hippocampal area of the 5xFAD mouse model at 6 months-of-age while gamma oscillations recorded from 5xFAD mice lacking gal3 (5xFAD-Gal3KO) remain similar to age-matched WT counterpart. Moreover, our results bridge a gap between cellular and molecular notions on the central role of gal3 in AD progression and behavioral studies by providing missing functional evidence that is relevant for those behaviors. This reinforces the therapeutic potential of inhibiting/removing gal3 to counteract AD progression and putatively for disease-modifying interventions in other neurodegenerative disorders involving microglia activation and neuroin ammation.

Consent for publication
Not applicable

Availability of data and materials
All the data is provided in the gure legends, in the supplementary material and available upon request without restrictions.

Competing interests
The authors declare that they have no competing interests.  Gal3 impairs gamma oscillation power and rhythmicity through its carbohydrate recognition-domain (CRD).
Inset: Representative power spectra for slices co-incubated with 1 µM Gal3 + 10 µM TD139 (gray), co-incubated with 1 µM Gal3 + 3 µM TD139 (magenta) or co-incubated with 1 µM Gal3 + 1 µM TD139 (green). B) Representative example traces of recordings performed in the conditions shown in A. C) Summary bar graphs of gamma oscillation power for the conditions shown in A demonstrating that 10 µM TD139 confers the most effective prevention against gal3-induced decrease of gamma oscillation power (ordinary one-way ANOVA followed by Holm-Sidak's multiple comparisons test,  (Table S2)    Data is presented as a mean ± SE. Statistics performed: ordinary one-way ANOVA followed by Holm-Sidak's multiple comparisons test. Signi cance levels are shown as * p<0.05, ** p<0.01, **** p<0.0001.