Recombinant human galectin-3 (Gal3) was produced by the Lund Protein Production Platform (LP3). Gal3 production were prepared as previously described (Salomonsson et al., 2010). Briefly, Gal3 production was performed in strain E. 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 purified on a 20-ml lactocyl-sepharose column. Peak fractions containing Gal3 were pooled and dialyzed against phosphate buffer saline (PBS, Nzytech), pH 7.4 (Garcia-Revilla et al., 2023). Freeze-dried Gal3 was stored at −80 °C until use. Supplementary Figure 1a shows SDS-PAGE analysis of the final Gal3 product.

Murine microglial BV2 cell lines (RRID: CVCL_0182) were cultured in Dulbecco’s Modified Eagle Medium GlutaMax (DMEM GlutaMax, Gibco, Cat: 61965-059) supplemented with 10% heat inactivated Fetal Bovine Serum (FBS, Gibco, Cat: 17593595) and 1% penicillin–streptomycin (Cytiva, Cat: SV30010) and kept at 37 °C 5% CO₂. Galectin-3 knockout BV2 (Gal3KO) cells were generated with the Alt-R CRISPR-Cas9 System (Integrated DNA Technologies, IDT), following manufacturer’s instructions. Briefly, two Alt-R CRISPR-Cas9 CRISPR RNA (crRNA, IDT) guides were designed to excise the transcription starting site of the galectin-3 gene (guide 1: CTTCTAGTACTCTTTTCCCC, guide 2: TACCCGCAGGAGAGCCAAGG, IDT; protospacer-adjacent motif sequences not included). Each one was then mixed with trans-activating Alt-R® CRISPR-Cas9 CRISPR RNA (tracRNA, IDT) to form the single-guide RNA (sgRNA). Afterward, ribonucleoprotein (RNP) complex was formed by mixing Alt-R S.p. HiFi Cas9 Nuclease V3 (IDT) with both sgRNAs in Opti-MEM (Gibco, Cat: 31985070). Reverse transfection was performed by mixing the RNP complex with Lipofectamine RNAiMAX (Invitrogen) and added to a 96-well plate, where 40.000 BV2 cells were then seeded. The same procedure was carried out as negative control with the RNP complex without the crRNA (WT cells). After that, propidium iodide (PI)-negative single-cells (1:1000 dilution, BioLegend, Cat: 421301, 5 min incubation) were sorted using the BD FACSAria™ III Cell Sorter (BD Biosciences) in new 96-well plates containing one third of conditioned medium to promote cell growth to create separate colonies. When wells were >95% confluent, individual clones were seeded in 24-well plates to perform downstream experiments to assess the success of the knockout. Clone 3 was used for all subsequent experiments.

For live-cell imaging experiments, 7,000 cells were seeded on a 96-well PhenoPlate microplate (Revvity, Cat: 6055302). After cellular attachment to the plate bottom, 1 μM of Gal3 protein or PBS was added 24 h prior to initiating imaging (preGal3). The following intracellular compartments were labeled through the addition of cell-penetrating dyes: mitochondria (using MitoTracker Deep Red FM, Invitrogen, Cat: M22426, 1:2500 dilution, 30 min incubation), lysosomes (using LysoTracker Green DND-26, Invitrogen, 1:2500 dilution, 5 min incubation) and nuclei (using Hoechst 33342, Invitrogen, Cat: H3570, 1:1000 dilution, 10 min). Eight images per well were taken every 30 min for 20 h with Operetta CLS High Content Screening (Revvity) by confocal spinning disk technology with a 40 × water immersion objective, and cells were kept at 37 °C, 5% CO₂ for the duration of all experiments. Superoxide production analysis was performed by incubation of the cells with MitoSox 5 μM (MitoSox Red, Invitrogen, Cat: M36007) under the same live-cell imaging parameters, after, the images were acquired 30 min after incubation.

2,500 cells were seeded on 96-well PhenoPlate microplates (Revvity, Cat: 6055302). Pretreatment for 24 h with Gal3 was performed as described. Nuclei were stained with Hoechst 33342 for 5 min, and live-cell imaging was conducted by brightfield and Hoechst (at 350 nm excitation) channels at 15-min intervals over a 6-h period using Operetta CLS System (Revvity). Cells were kept at 37 °C, 5% CO₂ for the duration of the experiment. Displacement and total accumulated distance were analyzed using Operetta CLS Harmony Software (at least 250 cells per replicate).

For flow cytometry experiments, cells were detached by removing medium, washing once with sterile PBS and adding trypsin 2.5% (Gibco, Cat: 15090046) for 2 min. Cells were then spun down in fresh medium and resuspended in FACS buffer (PBS + 2% FBS). A total of 20,000 events per sample of the target population were analyzed in BD FACSAria III Cell Sorter. Data was analyzed with FlowJo software version 10.4.0.

For cell division assay, 100,000 cells were seeded on 24-well plates. They were stained with CFSE (CFSE Cell Division Assay Kit, Cayman Chemical, Cat: 154-10009853) according to the manufacturer’s instructions. 4 h after the labeling, cells were collected from 24-well plates and stained for 5 min with PI (BioLegend, Cat: 421301, 1:1000 dilution). The live (PI-negative population) cells were then gated for proliferation assessment with FACSAria III cytometer. The cells that divided were determined as a percentage of CFSE-positive cells below the intensity peak of cells with CFSE at 0 h.

For CD11b and CD45 receptors characterization in WT and Gal3KO cells with preGal3, 50,000 cells were seeded on 48-well plates and pretreated with preGal3 for 24 h, Fc-receptors were blocked using TruStain FcX™ (anti-mouse CD16/CD32) antibody (1:500, BioLegend, Cat: 101319) for 5 min on ice. Afterward, they were stained with rat anti-mouse CD45-AlexaFluor 488 (1:200 dilution, BioLegend, Cat: 110717) and rat anti-mouse CD11b-PE/CY7 (1:200 dilution, Biosite, Cat: ASB-XQYMH4-0.1). PI (BioLegend, Cat: 421301, 1:1000 dilution, 5 min) was used as viability dye for cells. Supplementary Figure 3a shows the gating strategy followed to discard doublets and dead cells to reach the target population in all flow cytometry experiments, utilizing CD45 and CD11b staining as example. Supplementary Figure 3b shows median fluorescence intensity (MFI) histograms of CD11b and CD45 stainings.

V-PLEX Proinflammatory Panel 1 (mouse) Kit (Cat: K15048D) or U-PLEX Metabolic Group 1 (mouse) Assay (Cat: K152ACM) from MesoScale Discovery (MSD) were used to evaluate cytokine levels in media from cell cultures. V-PLEX plate included IFN-γ, IL-1β, IL-2, IL-4, IL-5, IL-6, KC/GRO, IL-10, IL-12p70 and TNFα; U-PLEX plate included IFN-γ, IL-1β, IL-2, IL-4, IL-10, IL-12p70, IL-13, TNFα and MMP-9. Assays were performed following manufacturer’s instructions and read using the MSD 1300 Meso Quickplex SQ 120MM Imager with Methodical Mind software, and analyzed with Discovery Workbench software (MSD). TNFα was the only cytokine detected within the detection range consistently throughout all samples both in the V-PLEX and U-PLEX panels.

7,000 cells were seeded in a 96-well PhenoPlate microplates (Revvity, Cat: 6055302) to test their phagocytic and endocytic ability. To study phagocytosis, Zymosan A (Saccharomyces cerevisiae) Alexa Fluor 488 conjugated BioParticles (Invitrogen, Cat: Z23373) were then added to the cells for 6 h (1:5000 dilution). To study endocytosis, dextran Alexa Fluor 647 particles (Invitrogen, Cat: D22914) were added for 20 min (1:4000 dilution). The following intracellular compartments were labeled through the addition of cell-penetrating dyes: mitochondria (using MitoTracker Deep Red FM or MitoTracker Green FM, Invitrogen, Cat: M22426 and Cat: M7514, at 1:2500 dilution), lysosomes (using LysoTracker Green DND-26, Invitrogen, Cat: L7526, at 1:2500 dilution) and nuclei (using Hoechst 33342, Invitrogen, Cat: H3570 at 1:1000 dilution). 1 μM of Gal3 was added either 24 h prior to the addition of zymosan and dextran (preGal3), or coincubated with zymosan for 6 h (coincubation Gal3 + zymosan), or 6 h before dextran addition (preGal3 6 h + dextran). Finally, cells treated with dextran were fixed with 4% paraformaldehyde (PFA, Histolab). For Supplementary Figure 2, imaging was performed in Zeiss 780 confocal laser-scanning microscope (Zeiss) and analysis in Fiji by ImageJ 1.53 software (NIH). Imaging and quantification was performed in Operetta CLS system (Revvity).

Total RNA from WT and Gal3KO BV2 cells was extracted with TRI Reagent (Sigma-Aldrich, Cat: AM9738) according to manufacturer’s instructions. RNA was quantified using the Thermo Fisher Nanodrop 2000c spectrophotometer using an extinction coefficient of 0.025 (μg/ml)-1 cm-1 at 260 nm. RNA was reverse-transcribed (500 ng) into cDNA in a total volume of 20 μL with the iScript cDNA synthesis kit (Bio-rad, Cat: 1708890) in the Bio-Rad T100 Thermal Cycler device. qPCRs were performed in 384-well plates in the Bio-Rad CFX384-C1000 Touch Thermal Cycler. 25 ng of cDNA were loaded in the reaction using the SsoAdvanced Universal SYBR Green Supermix (Bio-rad, Cat: 1725270) and the annealing temperature was 60 °C. Primer pairs for the target genes were developed with the Primer BLAST tool from the National Institutes of Health. As reference, gene Hprt or 18S was used. Primers were validated when efficiency was comprised within 80–110%. Relative expression ratio of each target gene was calculated by the ΔΔCt method.

100,000 cells from WT and Gal3KO BV2 were seeded on a 24-well plate (Sarstedt) and treated with 1 μM recombinant Gal1 produced by Lund Protein Production Platform (LP3, Lund University). DMEM FBS-free cell media was collected after 24 h of exposure with Gal1, Gal3 or combination for Immunoasssay plate cytokine determination (see above). Monolayer cells were later lysed for gene expression determination as described above.

10,000 cells were seeded on Lumox 96-well plate (Sarstedt Cat: 94.6120.096) and treated with Gal3 for either 360, 180, 60, 30 or 10 min and fixed with 4% paraformaldehyde (Histolab) for 20 min at room temperature. To evaluate cell morphology, the cells were stained with ActinRed 555 (Thermo Fisher, Cat: R37112) and Hoechst 33342 reagent. Images were recorded using a Zeiss 780 confocal microscope (×20 magnification).

BV2 cells were plated in 24-well plates and cultured to confluence. A cross-shaped scratch was introduced in the cell monolayer using a 200 μL pipette tip, after which the medium was replaced with fresh medium. For the Gal3 pre-treated condition, cells were incubated with Gal3 for 24 h prior to scratching. In the Gal3 post-treated condition, Gal3 was added to the culture medium immediately after the scratch, concomitant with medium replacement. Phase-contrast microscopy images were acquired at 0 h, 1 h, 6 h, and 24 h post-scratch. The percentage of wound closure was quantified using Fiji by ImageJ software, by measuring the open area devoid of migrated cells immediately after scratching and at 24 h (n = 3).

To study microglial markers, 50,000 cells were seeded in 24-well plate. After 24 h, the medium was changed to starvation medium (medium with 2% FBS) and 1.0 μM of Gal3 (preGal3) was added to the cells. Cells were then treated with RIPA buffer (Fisher Scientific, Cat: 89900) with PhosStop and cOmplete phosphatase and protease inhibitor tablets (Roche, Cat: 4906837001 and 11836170001) on wet ice, extracts were sonicated for 10 s, 80% amplitude, 0.8 cycles and stored at −80 °C. After centrifugation for 10 min at 15,000 g 4 °C, supernatant was stored at −80 °C. For Gal3KO cell characterization, cells were extracted as described above.

Primary antibodies used were: anti-galectin-3 (1:1000 dilution, goat, R&D Systems, Cat: AF1197), anti-TREM-2 (1:750 dilution, sheep, R&D Systems, Cat: AF1729), anti-TLR4 (1:1000 dilution, rabbit, Santa Cruz Biotechnology, Cat: sc-10741), anti-Dectin-1 (1:500 dilution, rabbit, Abcam, Cat: ab140039), anti-β-Actin Peroxidase (1:10000 dilution, mouse, Sigma-Aldrich, Cat: A3854). Secondary antibodies used were: horse anti-goat IgG peroxidase (Vector Labs, Cat: PI-9500-1), rabbit anti-sheep IgG peroxidase (Invitrogen, Cat: 61-8620), goat anti-rabbit IgG peroxidase (Vector Labs, Cat: PI-1000). All secondary antibodies were used at 1:3000. Blot images were taken using ChemiDoc platform (Bio-rad). Stripping was performed utilizing Restore PLUS Western Blot Stripping Buffer (Thermo Fisher). Western Blot band quantification was carried out using Image Lab Software (Bio-Rad).

Cell lysates were normalized at 0.15 mg/mL utilizing Pierce BCA Protein Assay Kit (Thermo Fisher) and analyzed using Mouse Dectin-1/Clec7a ELISA Kit (AssayGenie, Cat: SBRS1334) following manufacturer’s instructions. The detection limit of the kit is determined to be 0.12 ng/mL.

Cell cycle phase distribution in BV2 WT and Gal3KO cells was assessed based on nuclear Hoechst intensity using the Operetta CLS high-content imaging system. Cells were classified into three populations (G1, S, and G2/M phases) according to their sum nuclear fluorescence intensity, and the proportion of cells in each phase was quantified. Supplementary Figure 2 presents representative images, the corresponding intensity histogram, and the resulting phase quantifications.

5xFAD and 5xFAD/Gal3KO mice used for Clec7a immunostaining and quantification were generated and genotyped according to Boza-Serrano et al. (2019). To generate 5xFAD mice, wild-type dam and hemizygous sire were crossed. Experimental mice were homozygous for Gal3KO and hemizygous for 5xFAD mutation. Animal housing, handling and experiments were performed under international guidelines approved by the Malmö/Lund animal ethics committee (M30-16, Dnr5.8.18-01107/2018). Mice were housed in groups (maximum 5 animals/cage) and maintained in a 12-h light/dark cycle, with free access to food and water. Only female mice were studied.

For live-cell and other experiments using Operetta CLS, images were analyzed employing the High Content Analysis System Harmony (Revvity). The results shown are the mean of at least 6 imaged fields per well (with intraexperimental well duplicates). For the rest of image analysis experiments, Fiji by ImageJ software (NIH) was used. Immunofluorescence of mouse brains was analyzed utilizing a background subtraction of 50 pixel per each individual picture in the z-stack, followed by a maximum intensity projection of the channels. For Clec7a and Iba1 quantification, each plaque was manually selected in the MOAB2 channel, and the automatic threshold “Moments” available in Fiji was used quantify the % area covered by Clec7a per plaque. For CD45-CD11b quantification, an 8 μm enlarged area around plaque cores was used to quantify the % area covered by these markers utilizing automatic threshold algorithms (“RenyiEnthropy” was used in CD45 channel and “Triangle” in CD11b channel).

All statistical analyses were performed using GraphPad Prism 10. A significance threshold of p < 0.05 was applied. Exact p-values, including those approaching significance, are reported in the figure panels or legends. The specific statistical tests used for each dataset are indicated in the corresponding figure legends. For full ANOVA results and detailed test outputs, see Supplementary Table 1.

Galectin-3 (Gal3) is both endogenously expressed as well as released by microglia in the brain in pathological conditions (Garcia-Revilla et al., 2022), particularly in Alzheimer’s disease (AD) (Boza-Serrano et al., 2019). Its dual presence inside cells and in the extracellular space suggests that Gal3 may exert distinct, context-dependent effects on microglial function. Extracellular Gal3 can act in a paracrine or autocrine manner to influence neighboring cells as well as itself (Burguillos et al., 2015), while intracellular Gal3 contributes to cell-autonomous processes. Interestingly, homeostatic microglia barely express Gal3 at the transcriptomic level (Garcia-Revilla et al., 2022; Boza-Serrano et al., 2019). To investigate these functions, we generated Gal3-deficient BV2 microglia using the Alt-R CRISPR-Cas9 system (Integrated DNA Technologies). BV2 cell line is a murine microglial cell line with basal Gal3 expression (Figures 1a,b). We successfully generated several microglial BV2 cell line clones showing no levels of endogenous Gal3 at transcript and protein levels (Figures 1a,b). Clone 3 was selected for all further experiments (referred to as Gal3KO). In order to explore if other galectins were affected by this deletion, relative transcript expression was analysed (Figure 1c). Based on previously published transcriptomic data in microglia, galectins −1, −8 and −9 were studied (Krasemann et al., 2017). Besides Gal3, no other measured galectin showed any significant change (Figure 1c), suggesting no compensatory mechanism at least through another galactoside-binding protein. Curiously, we detected that Gal3bp, known to be a ligand for several galectins in the extracellular matrix, including Gal3, was also decreased in Gal3KO cells, indicating a transcriptional regulation driven by Gal3.

Microglia are highly mobile cells, and their proper mitochondrial and lysosomal function are key elements of neurodegeneration (Agrawal and Jha, 2020; Nixon, 2017; Lively and Schlichter, 2013). Therefore, to evaluate the differential involvement of endogenous and exogenous Gal3 in these microglial functions, we exposed WT and Gal3KO BV2 cells with 1 μM of exogenous Gal3 (preGal3, Supplementary Figure 1a). Cells were first tracked for 6 h using the Operetta CLS high-content imaging system (Revvity) (Figure 2a). Gal3KO cells displayed significantly greater accumulated traveled distance compared to WT counterparts, whereas preGal3 treatment decreased the motility to WT levels, effectively equalizing both groups (Figure 2b). To further investigate if exogenous Gal3 yielded an effect on cell motility, we performed a scratch in WT cells and determined that exogenous Gal3 at the time of scratch rather than before, was detrimental for microglial motility after 6 and 24 h (Supplementary Figures 1b–d), anticipating a fine tuning anti-migratory effect for Gal3.

Microglial proliferation is considered a signal for microglial activation and response to brain damage (Hammond et al., 2018). Therefore, to test whether Gal3 genetic deletion affected proliferation and cell division, we performed a CFSE assay, where CFSE fluorescent dye is incorporated to the cell and its content is diluted division after division. No significant differences were observed in the proportion of divided cells after 4 h, when the majority of cells had already divided in both genotypes (Figure 2c). Furthermore, we analyzed cell cycle stage by measuring nuclear DNA content based on the summed Hoechst fluorescence intensity within each nucleus (Olofsson et al., 2021). As observed, most of the cells present an intensity related to G1 phase which approximately doubles in phase G2/M, whereas intermediate intensities were assigned to phase S (Supplementary Figures 2a,b). Again, no difference was found between the genotypes in G2/M phase proportion of cells (Supplementary Figure 2c). Thus, supporting no difference in cell cycle and division rate.

Since microglia’s motility and tropism toward damaged brain areas are highly energy-consuming mechanisms, we evaluated the mitochondrial status after Gal3 deletion and exogenous exposure. Utilizing the Operetta CLS high-content imaging system, we quantified cell number, cell area, lysosomal content, and mitochondrial content after 24 h (Figure 2d). Notably, Gal3KO resulted in a highly significant genotype effect, with increased cell area and MitoTracker particle content, while preGal3 equalized WT and Gal3KO cells in terms of mitochondrial content (Figure 2e). Despite no change in cell area being detected after 24 h, we observed that exogenous Gal3 exposures on WT cells promoted a time-dependent altered microglial morphology over time, from a more rounded to a more elongated shape (Supplementary Figure 2d). In contrast, the total number of cells and the number of lysosome particles remained unaffected (Figure 2d). These findings suggest that Gal3KO cells may have an increased energy expenditure. To further investigate this hypothesis, we tested the production of superoxide with the fluorescent dye MitoSox (Figures 2f,g). Interestingly, Gal3KO cell had a significant overproduction of superoxide while exogenous Gal3 addition also increased the production of superoxide species (Figure 2g). Since superoxide production is mainly produced by complex I activity in the electron transport chain pathway (Kussmaul and Hirst, 2006), we next assessed the levels of mitochondrial external membrane TOM20 and complex I subunit NDUFB8 (Figures 2h,i). Complex I subunit levels correlated with superoxide production, thus supporting a change in this complex as the main source of superoxide production and explaining the observed MitoSox elevation (Figure 2i). However, the number of TOM20 mitochondria remained stable after preGal3 treatment and Gal3 deletion. This is intriguing as it does not correlate with the levels of MitoTracker, MitoSox or NDUFB8. MitoTracker accumulation initially depends on mitochondrial membrane potential, as well as MitoSox depends on complex I activity, while TOM20 is a receptor protein in the mitochondrial outer membrane not related to the oxidative phosphorylation pathway.

This led us to conclude that endogenous Gal3 is important on microglial metabolism and motility without affecting their division, while exogenous Gal3 mainly impacts migration. An increase mitochondrial oxidative phosphorylation pathway is suggested to occur by both Gal3 deletion and Gal3 exogenous stimulation, showing a dual role for Gal3, while total mitochondrial content may remain stable as suggested by levels of TOM20.

Another key feature of microglial cells is the surveillance of the extracellular matrix and the clearance of debris and pathogens. Thus, we next assessed phagocytosis and endocytosis function in WT and Gal3KO BV2 cells, as well as the impact of preGal3 treatment (Figure 3a). For this purpose, we measured the uptake of zymosan and dextran particles, which have been shown to be linked to phagocytosis (Venkatachalam et al., 2020) and endocytosis (Li et al., 2015) pathways, respectively. Here, exogenous Gal3 was added before (preGal3 24 h), or at the same time as zymosan (coincubation 6 h) or 6 h prior dextran (preGal3 6 h); while MitoTracker was used as a counterstain for cytosolic content. Phagocytosis, measured by zymosan internalization, was markedly enhanced in Gal3KO cells, however, a mild synergistic treatment effect of preGal3 appears, being more intense when Gal3 was added along with zymosan particles (Figures 3b,c). Again, a discrepancy emerges between the roles of endogenous and exogenous Gal3, suggesting a dual role. Endocytosis, measured as dextran uptake, was also increased under basal conditions in Gal3KO cells, while exogenous Gal3 had no effect on Gal3KO cells but further increased endocytosis in WT cells (Figures 3d,e). These results indicate that endogenous Gal3 constrains phagocytic and endocytic activity, as Gal3KO cells present increased levels of both mechanisms. The role of exogenous Gal3 is intriguing, as it limitedly increases phagocytic activity independently of endogenous Gal3 presence, resembling its role as opsonin (Karlsson et al., 2009). In contrast, exogenous Gal3 promotes endocytosis exclusively in WT cells but not in Gal3KO cells, arising the idea that extracellular/intracellular Gal3 balance is determinant in this pathway.

Next, we proposed to look into general immune features of microglial cells upon Gal3 deletion and exogenous presence. CD11b and CD45 are common surface markers used to define the microglial population in mouse models and human brain (Kettenmann et al., 2011). Thus, we performed flow cytometry analysis of CD11b and CD45 (Masliah et al., 1991; Czirr et al., 2017) in WT and Gal3KO cells exposed to preGal3 for 24 h (Figure 4a; Supplementary Figures 3a,b). Median fluorescence intensity (MFI) analysis revealed that CD45 surface levels were slightly reduced following Gal3 pretreatment in both genotypes (Figure 4b). CD11b levels also decreased after preGal3 treatment in WT cells, but, strikingly, Gal3 deletion produced a robust reduction in CD11b levels (Figure 4c), indicating a critical role of endogenous Gal3 in regulating this marker. Next, to assess the contribution of Gal3 to microglial inflammatory activity, we measured the NOX2 subunit p47phox. NOX2 is a hallmark of pro-inflammatory microglial activation and is capable of generating superoxide species (Qin et al., 2004). Interestingly, p47phox levels increased in WT cells after Gal3 exogenous exposure, however, Gal3 genetic deletion diminished its expression (Figures 4d,e). While p47phox increased levels are suggestive of a pro-inflammatory status of the cell (Yamagata et al., 2014), p47phox must be recruited by p22phox subunit and extensively phosphorylated during NOX2 complex activation (Paclet et al., 2022). Similarly, p22phox showed a detrimental effect following Gal3 genetic deletion, but no effect for exogenous Gal3 (Supplementary Figures 3c,d), however, colocalization of both markers showed no difference, meaning that no differential activation of NOX2 complex was produced (Supplementary Figure 3d). In line with p47phox, proinflammatory cytokine secretion was next evaluated. Similarly, preGal3 treatment in WT cells for 24 h increased TNFα release, as measured by TNFα concentration in the cell media, supporting both autocrine and paracrine inflammatory roles for Gal3 (Figure 4f). In contrast, Gal3KO cells exhibited reduced TNFα secretion under both basal and preGal3 conditions (Figure 4f). Other measured cytokines were not consistently within detection range. Curiously, we also detected that WT BV2 can secrete TNFα in response of both Gal3 or its galectin relative Gal1 (Supplementary Figure 4a). Thus, we proposed to verify whether Gal1 TNFα release was dependent on endogenous Gal3. To verify the unique autocrine/paracrine effects of Gal3, we exposed BV2 cells to increasing Gal1 dose. Both WT and Gal3KO cells responded in similar ways to Gal1 exposure in terms of Tnfa gene expression (Supplementary Figure 4b). Similarly, no genotype effect was observed for Gal1 induced Il1b gene expression (Supplementary Figure 4b). Indicating that the genotype effect for endogenous Gal3 in TNFα is only detected when exogenous Gal3 is the stimuli, but not other galectin, thus confirming a paracrine effect of Gal3 specifically.

Since microglial Gal3 has been implicated in multiple aspects of AD pathogenesis (Garcia-Revilla et al., 2022; Boza-Serrano et al., 2019; Tao et al., 2020; Boza-Serrano et al., 2022), we next investigated its relationship with microglial receptors that play central roles in microglial phenotype regulation and AD progression: TREM2, TLR4, and Clec7a (Hansen et al., 2018; Valiukas et al., 2025; Krasemann et al., 2017; Keren-Shaul et al., 2017; Huang et al., 2021; Miron et al., 2018; Roy et al., 2020). Western blot analysis demonstrated a clear genotype effect, with Gal3KO cells showing elevated TREM2 expression and, most strikingly, a dramatic increase in Clec7a levels compared to WT (Figures 4g,h). Interestingly, TLR4 remained unchanged, suggesting that Gal3 selectively modulates a subset of AD-associated receptors. To further confirm the robust Clec7a upregulation, we performed an ELISA assay, which validated the strong increase in Clec7a content in Gal3KO cells (Figure 4i). Both Clec7a and TREM2 converge into the phosphorylation of Syk (pSyk) in their downstream activation pathway (Wang et al., 2022; Nascimento et al., 2025), thus we stained the BV2 cells to assess the levels of pSyk (Figure 4j). WT and Gal3KO cells present similar basal levels of pSyk, increasing after preGal3 treatment only in WT but not in Gal3KO cells (Figure 4k). This means that the increased levels of TREM2 and Clec7a provoked by Gal3 deletion do not correlate with downstream Syk phosphorylation.

Overall, these findings suggest that endogenous and exogenous Gal3 differentially shape microglial receptor regulation and inflammatory output. Exogenous Gal3 appears to drive CD45 internalization and promote autocrine and paracrine proinflammatory responses, whereas endogenous Gal3 is required to maintain CD11b surface expression and support robust cytokine release. The loss of Gal3, therefore, shifts microglia toward a state with reduced CD11b availability but increased TREM2 and Clec7a receptors, but only WT cells are responsive to exogenous Gal3 addition in terms of pathway activation as shown by Syk phosphorylation and TNFα secretion. These data highlights distinct mechanisms by which endogenous versus exogenous Gal3 fine-tunes microglial activation.

Our group has previously characterized a reduction of plaque burden in the 5xFAD model following Gal3 genetic deletion at 6- and 18-month-old mice (Boza-Serrano et al., 2019). In this study, a Gal3-TREM2 interaction was described, in line with our in vitro results. Thus, we then tested if the observed changes in CD45, CD11b and Clec7a after Gal3 genetic deletion were observed in vivo. In this line, we used 3- and 6-month-old female 5xFAD and Gal3KO-5xFAD mice brains (Figures 5a,c). Firstly, microglia recruitment, measured as Iba1-positive area around the plaque was similar between genotypes at both age timepoints (Figures 5a,b). Next, we assessed Clec7a peri-plaque levels. In this case, Clec7a coverage per amyloid plaque was significantly greater in Gal3KO mice at both 3 and 6 months of age (Figures 5a,b), indicating that the Clec7a upregulation observed in vitro in the absence of Gal3 is maintained during AD pathology. Similarly to Iba1, CD11b and CD45 showed no differences between genotypes at any timepoint (Figures 5c,d), thus supporting no differences in microgliosis and no effect of Gal3 toward these receptors in the context of AD.

Taken together, these results identify Gal3 as a negative regulator of Clec7a in vivo. This connection links Gal3 to one of the most relevant AD-associated microglial receptors and suggests that Gal3 influences microglial phenotype both under basal conditions and in the context of neurodegeneration.