Gene expression profiles in MAGL KO mice

Previous studies provided evidence that pharmacological or genetic inactivation of MAGL alleviates neuropathology and improves synaptic and cognitive functions in several animal models of neurodegenerative diseases [21, 23,24,25,26,27,28]. To determine the cellular and molecular mechanisms underlying these beneficial effects of MAGL inactivation, we profile expression of genes in microglia and astrocytes from tKO, nKO and aKO mice to compare with WT mice using 10 × genomics chromium single-cell RNA sequencing analysis. A total of 67,452 cells passed the quality control. To visualize the genes relationship between cells, an unsupervised learning method, t-distributed stochastic neighbor embedding (t-SNE) was used to arrange barcodes in two dimensions. The resulting tSNE plots revealed 43 distinct clusters across all cells (Fig. 1A). Figure 1B shows the numbers of differentially expressed genes (DEGs) in tKO, nKO, and aKO versus WT. We identified that there are more upregulated DEGs in tKO and aKO mice, while there are more downregulated DEGs in nKO mice. Heatmap and Venn plots that display the distribution and average Log2 (fold change) of DEGs also show the differences between cell type-specific MAGL KO mice (Fig. 1C and Additional file 1: Fig. S1). Figure 1D lists top 5 terms of influenced biological processes (BP) based on comparison of the DEGs by Gene Ontology (GO) analysis in different genotypes. It appears that BP, including brain development, RNA splicing, mRNA processing and gliogenesis (Additional file 7: Tables S16–18), are differentially affected by cell type-specific inactivation of MAGL, suggesting that regulation of biological processes by inhibition of 2-AG is cell type-specific.

Fig. 1

Transcriptome-based cell classification in mouse cortex and hippocampus. A T-distributed stochastic neighbor embedding (t-SNE) plots showing the cell clusters in wild type (WT), total MAGL knockout (tKO), neuronal MAGL KO (nKO), and astrocytic MAGL KO (aKO) mice. B Volcano plots display up- and down-regulated differentially expresses genes (DEGs) in the brain from different genotypes. C Heatmap shows DEGs in the brain of tKO, nKO, and aKO mice. D Gene Ontology (GO) analysis of biological process (BP) terms of DEGs in different MAGL KO mice

Transcriptomic changes in microglia and astrocytes are different in cell type-specific MAGL KO mice

Microglia and astrocytes in the brain are the main players in neuroinflammatory responses and in endocannabinoid signaling-mediated solution of neuroinflammation [37, 38]. To determine whether expression of genes in microglia and astrocytes is altered by inactivation of MAGL, we analyzed DEGs in microglia and astrocytes, which are identified based on known cell type-specific markers: aif1, itgam and tmem119 for microglia; and aqp4, gja1 and slc1a2 for astrocytes (Additional file 2: Fig. S2 A–F). A total of 7536 microglia (2141 cells from WT, 2185 cells from tKO mice, 1857 cells from nKO mice, and 1353 cells from aKO mice) and 6769 astrocytes (1399, 1970, 1636, and 1764 astrocytes from WT, tKO, nKO, and aKO mice, respectively) passed the quality control and were selected for identification of DEGs (Fig. 2A). We identified that there are 1102 upregulated DEGs and 179 downregulated DEGs in microglia from tKO mice, 830 upregulated DEGs and 169 downregulated DEGs from nKO mice, 862 upregulated DEGs and 81 downregulated DEGs from aKO mice when compared to WT mice (Fig. 2B and Additional file 7: Tables S1–S3). In astrocytes, there are 949 upregulated DEGs and 63 downregulated DEGs in tKO mice, 1247 upregulated DEGs and 68 downregulated DEGs in nKO mice, and 1027 upregulated DEGs and 25 downregulated DEGs in aKO mice compared to WT mice (Fig. 2C and Additional file 7: Tables S4–S6).

Fig. 2

Cell type-specific inactivation of MAGL induces distinct gene expression profiles in microglia and astrocytes. A t-SNE plots of microglia and astrocytes clusters. Each colored dot represents a cell originated from WT, tKO, nKO or aKO mice. B Up- and down-regulated DEGs in microglia from different MAGL KO mice. C Up- and down-regulated DEGs in astrocytes from different MAGL KO mice. D, E The number of up- and down-regulated DEGs and their relationships in microglia and astrocytes from different genotypes. F, G Representative DEGs in microglia and astrocytes ***P < 0.0001 compared with WT; ###P < 0.001 compared with tKO; §P < 0.05, §§§P < 0.001 compared with nKO

To determine relationships between up- and down-regulated DEGs induced by cell type-specific inhibition of 2-AG metabolism, we generated Venn distribution plots. As shown in Fig. 2D, in microglia, there are 402 tKO-specific upregulated genes and 91 tKO-specific downregulated genes, 140 nKO-specific upregulated genes and 73 nKO-specific downregulated genes, and 166 aKO-specific upregulated genes and 11 aKO-specific downregulated genes (Fig. 2D). There are 454 upregulated and 48 downregulated genes overlapped among three groups. The numbers of overlapped DEGs between any two groups are as shown in Fig. 2D. In astrocytes, there are 280 tKO-specific upregulated genes and 33 tKO-specific downregulated genes, 429 nKO-specific upregulated genes and 38 downregulated genes, and 202 aKO-specific upregulated genes and 6 downregulated genes (Fig. 2E). There are 462 upregulated DEGs and 7 downregulated DEGs overlapped among three groups, which are likely expressed in astrocytes in MAGL KO mice (Fig. 2E). Figure 2F and G shows a few representative up- and down-regulated DEGs and their expression levels in microglia and astrocytes from different genotypes. Among top genes dysregulated in microglia, we observed a few important genes involved in learning and memory, including ttr and b2m, and genes that are crucial for cell apoptosis, including zfp36, rpl41, hk2, and malat1 [44,45,46,47,48,49]. In astrocytes, a few dysregulated genes are associated with generation of precursor metabolites and energy (e.g., cox17, sdhd, and bloc1s1) and with cell migration (e.g., pfn1, vegfa, and dclk1) [50,51,52,53,54,55]. Interestingly, we noticed that up- and down-regulated DEGs in both microglia and astrocytes are more pronounced in aKO mice than in tKO or nKO mice, suggesting that inhibition of 2-AG degradation in astrocytes plays an unique role in regulation of cellular function.

To identify the biological processes (BP) potentially affected by the up- and down-regulated DEGs in different genotypes, we performed GO analysis. As shown in Additional file 3: Fig. S3A and B, altered DEGs in microglia from tKO mice are associated with the actin polymerization or depolymerization, leukocyte migration, cytoplasmic translation and ribosome biogenesis (Additional file 7: Table S13). In microglia of nKO mice, altered DEGs are likely involved in ATP metabolic process, response to endoplasmic reticulum stress, cytoplasmic translation and ribonucleoprotein and complex assembly (Additional file 7: Table S14). Altered DEGs in aKO mice participate in gliogenesis, response to oxidative stress, cytoplasmic translation, ribonucleoprotein complex biogenesis and response to interleukin-4 (Additional file 7: Table S15). Additional file 3: Fig. S3C and D displays altered DEGs in astrocytes. It appears that there are only upregulated DEGs in astrocytes from tKO mice, which may affect several biological processes, including glial cell differentiation, RNA splicing, cellular respiration and regulation of nervous system development (Additional file 7: Table S16). Altered DEGs from nKO mice may linked to regulation of neurogenesis, nervous system development, and neuron differentiation (Additional file 1: Table S17). DEGs in astrocytes from aKO mice may affect the cellular respiration, regulation of RNA splicing, synapse organization, central nervous system neuron development and axon genesis (Additional file 7: Table S18). These results suggest that cell type-specific inhibition of 2-AG metabolism may result in different transcriptome changes in microglia and astrocytes, which, in turn, regulate or control a variety of biological processes.

Expression of cytokines or chemokines in microglia are promoted by inhibition of 2-AG metabolism in astrocytes

Inactivation of MAGL produces significant anti-inflammatory and neuroprotective effects in both in vitro and in vivo [11,12,13, 21, 23,24,25,26,27,28, 56]. To determine whether inhibition of 2-AG metabolism alters expression of immune/inflammation-related genes, we analyzed immune/inflammation-related genes (IGs). As shown in Fig. 3A, many IGs are up- or down-regulated in MAGL KO mice, but the patterns of up- or down-regulated IGs are different between tKO, nKO, and aKO mice. For instance, there are 44 up- or down-regulated IGs in tKO mice, 27 in nKO mice, and 43 in aKO mice. These differentially expressed IGs display high confidence in the gene network supported by STRING analysis using Cytoscape software (Fig. 3B).

Fig. 3

Differentially expressed immune- and inflammation-related genes (IGs) in all cells from MAGL tKO, nKO, and aKO mice. A Heatmap displays IGs in different genotypes. B Gene networks of IGs in different genotypes

Next, we analyzed IGs in microglia and astrocytes from MAGL KO mice. As shown in Fig. 4A and B, the expression levels of many IGs are altered in microglia and astrocytes from MAGL KO mice compared to WT mice. In microglia, there are 27 altered IGs in tKO mice, 18 in nKO mice, and 30 in aKO mice. In astrocytes, there are 13 altered IGs in tKO, 12 in nKO mice, and 8 in aKO mice. Apparently, expression of IGs in glial cells is differentially up- or down-regulated, in particular, in microglia, in cell type-specific MAGL KO mice. Remarkably, expression levels of C-C motif chemokine ligand ccl2, ccl3, ccl4, ccl12, and il1a, which are important in innate immunity [57], are robustly upregulated in microglia from aKO mice (Fig. 4C), but expression of ccl6 and ccl9 is upregulated in tKO microglia (Fig. 4A). Interestingly, only very few cytokines are changed in astrocytes from MAGL KO mice (Fig. 4B). The significantly increased expression of genes that are associated with inflammation in astrocytes are gja1, f3, cd9 and il15ra in aKO mice (Fig. 4D). We also observed that there are differences in changes in expression of IGs in microglia and astrocytes between different MAGL KO mice (Fig. 5A and B). In particular, expression of ccl2, ccl3, ccl4, ccl12, and ii1a in microglia are significantly upregulated in aKO mice when compared with tKO and nKO (Fig. 5C). These results indicate that cell type-specific inactivation of MAGL, especially inhibition of 2-AG metabolism in astrocytes, induces distinct gene expression profiles of chemokines in microglia, suggesting that inhibition of 2-AG metabolism in astrocytes promotes immune/inflammatory vigilance in microglia by signaling interacting between astrocytes and microglial cells.

Fig. 4

Differentially expressed immune- and inflammation-related genes (IGs) in microglia (A) and astrocytes (B) from MAGL tKO, nKO, and aKO mice. C, D Representative IGs differentially expressed in microglia and astrocytes from aKO mice. **P < 0.01, ***P < 0.001 compared with WT mice

Fig. 5

Comparison of differentially expressed immune/inflammation-related genes between tKO, nKO, and aKO mice. A, B Differentially expressed immune/inflammation-related genes in microglia and astrocytes between different genotypes. C Representative immune/inflammatory genes in microglia from different MAGL KO mice. ###P < 0.001 compared with tKO; §P < 0.05, §§§P < 0.001 compared with nKO

We then used STRING analysis to identify predicted gene networks relevant to transcriptomes of microglia and astrocytes in different genotypes. In microglial gene networks, cx3cr1ccl6, itgam, ccl9, and c1qc are in specific confident centers in tKO mice (Additional file 4: Fig. S4A); p2ry12, trem2, tyrobp, and il6 are in specific confident centers in nKO mice (Additional file 4: Fig. S4B); ccl4, csf1r, cx3cr1, ccl2 and itgam are in specific confident centers in aKO mice (Additional file 4: Fig. S4C). Our results suggest that itgam and cxcl12 mice are the important genes in the confident centers primarily in microglial gene networks when MAGL is inactivated.

MAGL inactivation regulates expression of genes involved in oxidative stress and neural functions

Previous studies reveal that 2-AG protects neurons against oxidative stress and regulates functions of the nervous system [58]. In analyzing immune- and inflammation-related DEGs in MAGL KO mice, we also noticed significant changes in expression of genes regulating oxidative stress and neuronal functions. As shown in Fig. 6 and Additional file 5: Fig. S5, genes (e.g., fos, cst3, txnip, prdx1, ndufs8, cfl1, atp2a2, apoe, ppia, rhob, selenok, and sirpa) that are important for responses to oxidative stress are significantly changed in microglia in MAGL KO mice. In addition, we observed changes of genes (e.g., ntrk2, gpm6a, ntm, slc6a1, epha5, pax6, cadm1, ndrg2, dclk1, chchd10, syne1, and tnik) that regulate neural functions in astrocytes. Changes in expression levels of these DEGs in microglia are similar in all genotypes. For example, expression of apoe, ctnnb1, sirpa, and ppia in microglia is upregulated in tKO, nKO, and aKO mice. However, changes in expression levels of the DEGs in astrocytes are heterogeneous. For instance, expression of epha5, pax6, syne1, and tnik in astrocytes is altered only in nKO mice. In contrast, changes in expression of ntrk2 and cadm1 in astrocytes are only seen aKO mice. These results suggest that selective inactivation of MAGL, which changes expression of genes in microglia and astrocytes, may differentially regulate oxidative stress and neuronal functions.

Fig. 6

A Differentially expressed genes involved in “Oxidative response” in microglia from different genotypes. B Differentially expressed genes involved in “Neural function” in astrocytes. **P < 0.01; ***P < 0.001 compared with WT; ###P < 0.001 compared with tKO; §§§P < 0.001 compared with nKO

Regulation of intercellular communications by inactivation of MAGL

Expression of genes and functions in individual cells are not only affected by intracellular molecules, but also regulated or modulated by intercellular communications [59]. Based on the data shown above, inhibition of 2-AG metabolism in astrocytes significantly alters expression of genes in microglia, indicating signaling interactions between microglia and astrocytes in aKO mice. To this end, we used Cellchat package in R (4.0.4) to analyze signaling molecules that are changed by cell type-specific inactivation of MAGL in microglia and astrocytes. Since intercellular communications are primarily regulated by ligand–receptor pairs, we explored the number of interactions (ligand–receptor pair) and interaction strengths in different genotypes. We found that there are many significant changes in ligand–receptor pairs MAGL KO mice. These ligand–receptor pairs are categorized into several different signaling pathways, including PTN, PSAP, TGFb, MK, GRN, CALCTIN, GAS, CCL, ENHO, TNF, PDGF, EGF, FGF, WNT, PROS, and CHEMERIN (Fig. 7A). It appears that the interaction strength of the CCL pathway is attenuated in microglia from tKO mice, but those of EGF and PROS pathways are boosted in microglia of tKO, nKO and aKO mice (Fig. 7A). In astrocytes, relative strengths of FGF, WNT and CHEMERIN pathways are enhanced in conditional KO mice (Fig. 7A).

Fig. 7

Enhanced cell–cell communications in cell type-specific MAGL KO mice. A Signaling interactions are strengthened in microglia and astrocytes by inactivation of MAGL. B Representative ligands or receptors that are altered in astrocytes from different genotypes. *P < 0.05, **P < 0.01, ***P < 0.001 compared with WT; ###P < 0.001 compared with tKO; §§§P < 0.001 compared with nKO

We further analyzed expression levels of ligands or receptors in these signaling pathways. As shown in Fig. 7B, expression of some of ligands and receptors is upregulated in astrocytes. For example, expression levels of fzd1, fzd2, and fzd4, which are involved in Wnt signaling pathways, in astrocytes are significantly increased in tKO, nKO and aKO mice. In addition, expression levels of fgf1 and rarres2 in astrocytes are elevated in aKO mice. Similarly, expressions of pros1 and tgfa are elevated in microglia from all the genotypes, while expression of ncl is increased only in microglia of nKO mice (Additional file 6: Fig. S6A).

We hypothesized that these ligands or receptors likely play an important role in regulating expression of immune/inflammation-related genes. To this end, we used STRING analysis to reveal interactions between IGs and ligands/receptors. As shown in Fig. 8, these ligands or receptors may interact with multiple genes, including p2ry12, c1qa, csf1r, gpr34, mfge8, tgfb1, and ccl2. These ligands or receptors may also regulate other IGs indirectly via these hub genes (Fig. 8). Interestingly, fzd1 and fzd2 in astrocytes could modulate itgam, prkcd, and tgfbr1 in microglia via the hub gene rhoa in aKO and tKO mice (Fig. 8A and C). In aKO mice, fgf1 is crucial ligand in astrocytes, which may regulate IGs in microglia (Fig. 8C). Additional file 6: Fig. S6B and 6C display gene networks, which indicate ligands and receptors interacting with DEGs in microglial and astrocytes, which modulate “Response to oxidative stress” and neuronal functions in aKO mice. Particularly, receptors in Wnt signaling pathways interact with several hub genes in all gene networks.

Fig. 8

Gene networks between astrocytes and microglia from different genotypes. A Interactions of DEGs and IGs in astrocytes and microglia in tKO mice. B Interactions of DEGs and IGs in microglia and astrocytes in nKO mice. C Interactions of DEGs and IGs in astrocytes and microglia in aKO mice. Node color: green indicates DEGs in microglia; red indicates DEGs in astrocytes; yellow indicates DEGs in both microglia and astrocytes

Our results provide evidence that inactivation of MAGL induces changes in expression of ligands and their receptors, which may further regulate expression of immune- and inflammation-related genes in microglia and astrocytes. MAGL inactivation-induced changes in expression of genes involved in Wnt signaling pathways in astrocytes likely contributes to these intercellular communications.