NLRP3 KO mice have enhanced gut motility with increased excitatory ChAT + and VGLUT2 + neurons in myenteric plexus
Previous studies showed that gut motility could be regulated by CASPASE-11 or NLRP6, which directly affected the pyroptosis of neurons [5, 21, 27]. To investigate the effect(s) of another inflammasome NLRP3 on gut physiological motility, we performed gastrointestinal transit test using carmine red dye and glass beads. The time of eliminating carmine or glass beads was significantly shorter in NLRP3 KO mice than control wild-type (WT) mice (Fig. 1A), suggesting that there has stronger muscle force in the colons of NLRP3 KO mice. Indeed, enhanced gut force upon exposure to muscarinic agonist acetylcholine (ACh) was observed in NLRP3 KO mice (Fig. 1B), which was similar to control CASPASE-1/11 KO mice [27]. Gut motility is related to the numbers of enteric nervous system [5, 27]. There are two distinct networks, the submucosal and MPs in the enteric nervous system [1]. More myenteric neurons could be detected in NLRP3 KO mice (Fig. 1C). MPs are comprised of multiple heterogeneous neurons [1] such as excitatory ChAT+ and VGLUT2+ and inhibitory nNOS+ neurons [5, 7, 28]. Data showed that more excitatory ChAT+ and VGLUT2+ neurons were located in the muscularis externa (ME) of NLRP3 KO mice as compared to WT mice (Fig. 1D). Notably, only some excitatory neurons could co-express ChAT and VGLUT2 (Fig. 1E), which was distinguished from previous reports that VGLUT2+ neurons are ChAT positive [29]. We next used anti-cleaved CASPASE-1 (Cc1) antibody to detect pyroptotic neurons [30]. Pyroptosis could be found in ChAT+ and VGLUT2+ neurons of the MPs in WT but not NLRP3 KO mice (Fig. 1D), implying that there indeed had increased ChAT+ and VGLUT2+ neurons in the MPs of NLRP3 KO mice. Notably, inhibitory nNOS+ neurons, which play an important role in smooth muscle relaxation [31], revealed no significant changes in these mice (Fig. 1D). As a positive control, CASPASE-1/11 KO mice also had increased neurons in the colon tissues (Fig. 1D). Taken together, NLRP3 KO mice have enhanced gut motility with increased excitatory ChAT+ and VGLUT2+neurons in the MPs.
NLRP3 KO mice have increased resident macrophages in colon tissues
Since gut neurons are regulated by CASPASE-11 or NLRP6 through pyroptosis [5, 21], it is also possible for these neurons to be regulated by another inflammasome NLRP3. However, NLRP3 was not detected in the MP neurons of the mouse colon tissues (Supplementary Fig. S1). Intestinal resident macrophage populations play a role in the normal function of enteric neurons [5, 7, 8]. The MMs, located within and surrounding the MP, were shown to regulate the activity of enteric neurons and peristalsis [8, 32]. Furthermore, the number of neurons was related to the resident macrophages, which could protect neurons [5, 8]. Thus, we analyzed these resident macrophages. CD45+CD11b+ cells were first sorted from the colonic tissues, including submucosa, lamina propria (LP) and muscularis externa (ME) of specific pathogen free (SPF) WT, NLRP3 KO and also CASPASE-1/11 KO mice (6–8 weeks old, male mice). Then sorted CD45+CD11b+ cells from pooled sample (6 mice) were sequenced on a 10⋅ Genomics platform and clustered using DESC, a deep embedding algorithm for single-cell clustering [21]. The macrophages in the colon tissues includes 6 subpopulations (Fig. 2A), similar to other analyses [21]. Importantly, the cluster 4 subpopulation was markedly increased in the colon tissues of NLRP3 KO mice (Fig. 2B). This population of macrophages expressed high levels of F4/80+, Mrc1 (CD206+) and CX3CR1+ (Fig. 2C), which can be detected by resident macrophages, indicating that increased cluster 4 macrophages in NLRP3 KO mice are resident macrophages. Cluster 6 and 1 expressed high levels of Ly6c2 and F10 (Fig. 2C and supplementary Fig. S2), which were the markers of blood monocytes/macrophages [33], suggesting that these clusters belong to monocytes/macrophages. Previous studies indicated that Ly6Chi monocytes were able to enter into the colon, and then mature into F4/80hiCX3CR1hiMHCII+ CD64+ resident macrophages [34]. This developmental process involves a series of identifiable intermediaries in which CCR2, F10, Ly6c2, Hdc, Sell and Hp, markers of blood monocytes [33] are lost, while expressions of F4/80, CX3CR1, CD206, CD163 and CD64 are gained or upregulated [35, 36]. Our results supported this progression of monocyte differentiation through distinct phenotypic developmental stages (Supplementary Fig. S2). Increased resident macrophages in the colon tissues of NLRP3 KO mice were further confirmed using flow cytometry and immune staining (Fig. 2D and supplementary Fig. S3). These increased resident macrophages were related to reduced pyropotosis (Fig. 2E, F). Indeed, isolated NLRP3 KO resident macrophages from the colon tissues were stronger resistance to NLRP3 activator (LPS plus nigericin) mediated pyroptosis (Supplementary Fig. S4). In addition, F4/80+ macrophages could be detected in whole colon tissues; Whereas Ly6C macrophages did not be detected in the muscularis externa (ME) (Supplementary Fig. S5), suggesting that Ly6C+ macrophages, which belong to inflammatory macrophages [35, 36], do not contribute to the ME. Taken together, our data demonstrate that there are increased resident macrophages in the colon tissues of NLRP3 KO mice.
In addition, CD81 gene could be detected in the resident macrophage cluster 4 (Fig. 3A-C). CD81 is a member of the tetraspanin family, which encompasses membrane proteins characterized by four transmembrane domains [37]. CD81 marker in the resident macrophages could also be confirmed by flow cytometry and immunostaining (Fig. 3D-F), suggesting that CD81 is a surface marker of gut resident macrophages, consistent with the resident macrophages in kidney [38]. Notably, only partly F4/80+ or CX3CR1+ resident macrophages could express surface CD81 (Fig. 3G).
Increased resident macrophages in the colon tissues of NLRP3 KO mice are muscularis macrophages
There existed multiple resident macrophage populations such as mucosal macrophages, perivascular macrophages, crypt base macrophages and MMs in the colon tissues of mice [9]. Indeed, the resident macrophages in the cluster 4 of supplementary Fig. S1 could further be divided into 4 subsets by DESC, including cluster 4.0, 4.1, 4.2 and 4.3 (Fig. 4A and supplementary Fig. S6A). Interestingly, while the proportion of four subpopulations of resident macrophages in WT mice was basically equal (Fig. 4A), cluster 4.3 subpopulation in NLRP3 KO mice was markedly increased (Fig. 4A). This was also different from control CASPASE-1/11 KO mice, which had markedly increased resident macrophages in cluster 4.1 and 4.2, but much less resident macrophages in cluster 4.3 (Fig. 4A). Since activation of CASPASE-11 or NLRP3 inflammasomes upon exposure to their ligands can cause the loss of the macrophages [16], we speculate that increased cluster 4.1 and 4.2 in CASPASE-1/11 KO mice might be located in mucosa or submucosa since there have more LPS [39], which can cause pyroptosis of macrophages; Whereas increased cluster 4.3 macrophages in NLRP3 KO mice might be in the places far from gut cavity such as ME.
We further analyzed these subpopulations based on the marker genes identified in different clusters of cell type using a published RNA-seq database ImmGen. Of note, cluster 4.0 and 4.3 respectively cover 25% or 18% genes with microglia (Fig. 4B) [9, 40], implying that these subpopulations potentially are related to neurons. Bio-informatics analyses also confirmed that there indeed existed a signal pathway for the regulation of neuron death in cluster 4.3 (Supplementary Fig. S6B). The resident macrophages of cluster 4.3 but not others could express high levels of β2-AR (adrb2) (Fig. 4C, D), suggesting that this subpopulation belongs to MMs [5, 9]. Furthermore, arg-1, which could be expressed in MMs, was also detected in these MMs (Fig. 4D, E). The development of resident macrophages is dependent on colony stimulatory factor 1 receptor (CSF1R) [8]. CSF1R could be also detected in the macrophages surrounding the MP (Fig. 4D). However, BMP2, which could stimulate the neurons and influence gut motility [8, 41], was not highly expressed in this subpopulation (Fig. 4C). Consistent with other reports [5, 7, 9], resident MM cluster 4 not only expressed F4/80, CX3CR1, CD206, CD163 but also c1qa, c1qb and c1qc (Supplementary Fig. S7A-C) [38, 42]. Cluster 4.3 could also co-express F4/80, CX3XR1, CD206, CD81 and c1qa, c1qb and c1qc. Immunostaining showed that c1qa+F4/80+ macrophages not only existed in mucosa and submucosa but also ME of colon tissues (Supplementary Fig. S7D-F). Furthermore, c1qa+ macrophages in the ME could also express β2-AR (adrb2) and CSF1R (Fig. 7D-F). Taken together, increased resident macrophages of cluster 4.3 in NLRP3 KO mice belong to the MMs, which express not only F4/80, adrb2, arg-1 and CSF1R but also c1qa+.
Effects of NLRP3 on excitatory neurons depend on muscularis macrophages Depletion of self-maintaining macrophages results in loss of enteric neurons [7]. To further determine protective roles of the resident macrophages in the neurons, we deleted macrophages using liposome encapsulated dichloromethylene with demonstrated deletion of macrophages (Fig. 5A). Consistent with previous reporter [8] at 2 days after deleting macrophages, there are no significant changes in HuC/D+ enteric neurons. However, over 7 days after deletion of macrophages, there had markedly reduced HuC/D+ enteric neurons in MP (Fig. 5B), indicating that resident macrophages determine the loss of enteric neurons in MPs. Deletion of macrophages in mice caused also decreased elimination time of carmine and maximal contractile forces in response to ACh (Fig. 5C, D). Thus, resident macrophages were necessary for excitatory enteric neurons.
MMs can be efficiently repopulated by bone marrow monocytes (BMC) [7]. To further illustrate the role of NLRP3 KO resident macrophages in ChAT+ and VGLUT2+ neurons in the MPs, we employed BMC transplant model. After transplantation from CD45.2+NLRP3 KO to CD45.1+ mice for four weeks, resident macrophages in the colon tissues of CD45.1 mice were CD45.2+ cells, indicating that CD45.1 resident macrophages are replaced by NLRP3 KO CD45.2+ macrophages (Fig. 6A). NLRP3 KO macrophages might resist against NLRP3 ligands surrounding MPs [18]. Indeed, there had increased MMs in the mice transplanted by NLRC3 KO CD45.2+ BMCs (Fig. 6B). The markedly increased neurons, especially excitatory ChAT+ and VGLUT2+ neurons around the MPs in the mice transplanted by NLRP3 KO CD45.2 BMCs were also observed (Fig. 6C, D). However, the quantification of nNOS+ neurons did not reveal significantly changes in the mice transplanted by NLRP3 KO or WT CD45.2 BMCs (Fig. 6D). The increased NLRP3 KO CD45.2+ resident macrophages could also promote not only the elimination of carmine but also enhance force developed by isolated colon muscle strips in response to ACh (Fig. 6E, F). Taken together, the effects of NLRP3 on ChAT+ and VGLUT2+ neurons in the MPs depend on gut MMs.
More excitatory neurons in the colon tissues of mice treated by gut microbiota metabolite β-hydroxybutyrate
Age-related changes in gut microbiota alter the phenotype of MMs and disrupt gastrointestinal motility [43]. Thus, it is possible for gut microbiota to regulate these MMs. Gut microbiota metabolite β-hydroxybutyrate (BHB) can suppress activation of NLRP3 inflammasomes in response to urate crystals, ATP and lipotoxic fatty acids [44]. Since NLRP3 may exert a critical role in controlling the loss of MMs, we observed whether gut microbiota BHB could resist the loss of MMs to maintain the homeostasis of ChAT+ and VGLUT2+ neurons. Indeed, flow cytometry and immune-staining showed that F4/80+CX3CR1+and F4/80+CD81+resident macrophages, and MMs markedly increased in the colon tissues of WT but not NLRP3 KO mice after BHB (Fig. 7A, B). Excitatory ChAT+ and VGLUT2+ but not inhibitory nNOS+ neurons significantly increased in WT but not NLRP3 KO mice after administrating BHB (Fig. 7B, C). BHB also promoted elimination of carmine and glass beads, and enhanced force developed by isolated colon muscle strips in response to ACh in WT but not NLRP3 KO mice (Fig. 7D, E). We also examined the in vitro impact of BHB on the resident macrophages, which were isolated from the colon tissues of mice. Consistent with previous results [44], BHB could markedly decrease the pyroptosis of these resident macrophages through inhibiting nigericin-mediated NLRP3 activation (Supplementary Fig. S8). Taken together, BHB can increase ChAT+ and VGLUT2+ neurons in the MP of colon tissues through inhibiting pyropotosis of the MMs.