DLG2 expression was low in inflamed tissues and in colorectal tumors, whereas NLRP3 and NFKBIZ expressions were high in inflamed tissues.
We evaluated the expression of DLG2, NLRP3 and NFKBIZ genes using publicly available microarray data for the independent colon cohorts (GSE4183; Fig. 1a, e, i) (Galamb, Gyorffy et al. 2008), (GSE109142; Figure 1b, f, j) (Haberman, Karns et al. 2019), (GSE75214; Fig. 1c, g, k) (Vancamelbeke, Vanuytsel et al. 2017), (GSE10950; Fig. 1d, h, l) (Jiang, Tan et al. 2008), obtained from the R2 Genomics Analysis and Visualization Platform (http://r2.amc.nl). In the different datasets gene expression was investigated in samples from patients with inflammatory bowel disease (IBD), adenoma or colon cancer compared to healthy controls (Fig. 1a); ulcerative colitis (UC) patients compared to controls (Fig. 1b); or ulcerative colitis (UC) patients with active vs. inactive disease state (Fig. 1c) and case controlled colorectal tumor samples compared to paired healthy mucosa (Fig. 1d). DLG2 was downregulated in IBD (log2 FC = 0.81, p < 0.05), adenoma (log2 FC = 1.1, p < 0.1) and colon cancer (log2 FC = 1.3, p < 0.01; Fig. 1a). DLG2 also showed a decrease in expression in UC compared to the control (log2 FC = 0.14, p < 0.01; Fig. 1b) and further decrease in expression when UC was active (log2 FC = 0.22, p < 0.001; Fig. 1c). A large downregulation in DLG2 was seen in the paired healthy-tumor colon tissue from colorectal cancer patients (log2 FC = 12.6, p < 0.001; Fig. 1d). There was no difference in NLRP3 expression in samples from patients with IBD, adenoma or colon cancer compared to healthy controls (Fig. 1e). An increased NLRP3 expression was detected in UC compared to control samples (log2 FC = 0.59, p < 0.001; Fig. 1f), and a further increase in NLRP3 expression when the UC was active (log2 FC = 0.74, p < 0.001; Fig. 1g). A lower NLRP3 expression was seen in the colon tissue compared to paired healthy mucosa in colorectal cancer patients (log2 FC = 1.1, p < 0.01; Fig. 1h). The NFKBIZ expression was higher in IBD samples compared to controls (log2 FC = 0.59, p < 0.001; Fig. 1i), however no difference from controls was detected in adenoma or colon cancer samples (Fig. 1i). There was also an increased NFKBIZ expression in UC samples compared to the control (log2 FC = 1.8, p < 0.001; Fig. 1j) and a further increase in NFKBIZ expression when UC was active (log2 FC = 0.78, p < 0.001; Fig. 1k). However, a downregulation in NFKBIZ was seen in the paired tumor tissue compared to healthy mucosa from colorectal cancer patients (log2 FC = 0.52, p < 0.001; Fig. 1l).
DLG2 expression was initially upregulated followed by downregulation over time in response to inflammatory signals.
We evaluated the expression of DLG2, NLRP3 and NFKBIZ genes using publicly available microarray data in mouse colon from mice treated with Dextran Sulfate Sodium (DSS) to induce a colitis like phenotype (Fang, Zhang et al. 2012) (GSE22307) and T cell transfer (Fang, Bruce et al. 2011) (GSE27302) to model chronic colitis, obtained from the R2 Genomics Analysis and Visualization Platform (http://r2.amc.nl). DLG2 was upregulated in the colitis mouse model four days after DSS treatment with no difference between zero and six days of DSS treatment (log2 FC = 0.47, p < 0.001) (Fig. 2a). When given a T-cell transfer, DLG2 expression in mice was decreased after four and six weeks (log2 FC = 0.57, p < 0.05 and log2 FC = 0.52, p < 0.05, respectively; Fig. 2b). When THP-1 monocytes were treated with Lipopolysaccharides (LPS) to induce immune responses, there was an initial increase in DLG2 expression 12 hours after exposure (log2 FC = 0.879, p < 0.001; Fig. 2c) then a decrease in DLG2 was detected 24 hours post exposure (log2 FC = 1.63, p < 0.001; Fig. 2c). The expression of the Drosophila melanogaster DLG2 ortholog dmDLG increased in fly larvae gut cells in response to Bifidobacterium lactis Bl-04 and Lactobacillus acidophilus NCFM, 24 hours post treatment (log2 FC = 0.98, p < 0.01; Fig. 2d), with a progressive and gradual decrease in dmDLG over time until four days post treatment (log2 FC = -0.64, p < 0.05; Fig. 2d). NLRP3 expression increased after six days of DSS treatment (log2 FC = 1.9, p < 0.001; Fig. 2e) and after T cell transfer by increasing expression between four and six weeks after treatment (log2 FC = 1.0, p < 0.01 and log2 FC = 1.6, p < 0.01, respectively; Fig. 2f). When THP-1 cells were treated with LPS there was no alteration in NLRP3 expression over time (Fig. 2g). NFKBIZ expression responded to DSS treatment after six days (log2 FC = 0.52, p < 0.01; Fig. 2h) and to T cell transfer by increasing its expression across all time points against the control up to six weeks after treatment (log2 FC = 0.80, p < 0.05, log2 FC = 1.5, p < 0.001 and log2 FC = 1.4, p < 0.01, respectively; Fig. 2i). When THP-1 cells were treated with LPS there was an initial increase in NFKBIZ expression 12 hours after exposure (log2 FC = 0.83, p < 0.001; Fig. 2j) with the increase sustained 24 hours post exposure (log2 FC = 0.84, p < 0.001; Fig. 2j).
DLG2 altered expression of NFKB components
Using differentiated THP-1 monocytes we compared mock transfection to DLG2 overexpressed with subsequent activation with growth media, LPS or LPS with ATP. We confirmed that the transfection was successful by determining the DLG2 expression (data not shown), followed by gene expression analysis of NFKB1, NFKBIZ, RELA and RELB (Fig. 3a-d). We determined that there was no difference in RELA expression between the control and the DLG2 transfected cells (Fig. 3a). RELB showed a consistent upregulation in response to DLG2 overexpression, a stronger effect than both LPS and ATP treatments had (log2FC=3.55, p < 0.01, log2FC=4.46, p < 0.01 and log2FC= 4.17, p < 0.01; Fig. 3b). We investigated the expression of NFKB1 and showed that, like RELB, the expression was consistently upregulated in the DLG2 expressed cells, with no additional effect by addition of LPS or ATP (log2FC= 3.51, p < 0.01, log2FC= 3.27, p < 0.01 and log2FC= 3.65, p < 0.01; Fig. 3c). Finally, we investigated the expression of NFKBIZ which was upregulated across all of the activations compared to the control with DLG2 overexpressed cells showing higher expression (log2FC= 3.77, p < 0.001, log2FC= 4.18, p < 0.001 and log2FC= 2.667, p < 0.001; Fig. 3d). We subsequently confirmed that the effects of DLG2 overexpression seen on gene expression level, also affected the protein expression, visualized by immunoblot for RELA, RELB, NFκB1, and IκBζ, using GAPDH as loading control (Fig. 3e).
DLG2 stimulated inflammasome formation and increased apoptosis in macrophage like cells
Using differentiated THP-1 monocytes we compared mock transfection to DLG2 overexpressed cells and subsequently treated the cells with LPS and LPS with ATP. We determined the gene expression level of IL1B, IL6, BAX and BCL2 (Fig. 4a-d). IL1B showed higher expression in the DLG2 transfected cells regardless of activation when compared to the equivalent activation (log2FC= 5.45, p < 0.001, log2FC= 3.22, p < 0.01 and log2FC= 3.03, p < 0.01, for the control, LPS and LPS+ATP respectively; Fig. 4a). DLG2 attenuated IL6 expression after activation with LPS and LPS and ATP (log2FC= 2.78, p < 0.001 and log2FC= 3.29, p < 0.001), with no difference in non-activated cells (Fig. 4b). DLG2 overexpression also resulted in consistently higher BAX expression across all activations (log2FC= 1.45, p < 0.01, log2FC= 1.90, p < 0.001 and log2FC= 1.57, p < 0.01; Fig. 4c) and consistently lower BCL2 expression across all activations (log2FC= 0.99, p < 0.01, log2FC= 1.01, p < 0.01 and log2FC= 1.33, p < 0.01; Fig. 4d). We subsequently determined the protein expression by immunoblot for DLG2, BAX, BCL2, ser727 p-STAT3, total STAT3, ser235/236 p-S6, total S6 and GAPDH (Fig. 4e). DLG2 overexpression resulted in increased BAX expression in non-activated, LPS and LPS+ ATP stimulated cells, and protein expression of BCL2 was decreased across all activations, which agreed with the gene expression data. STAT3 phosphorylation increased stepwise with the LPS and LPS + ATP treatments. The overexpression of DLG2 however resulted in an increase of STAT3 phosphorylation during the LPS only treatment and a subsequent decrease during the LPS + ATP treatment (Fig. 4e). Furthermore, p-S6 was also decreased in all DLG2 transfections while total S6 expression remained unaffected (Fig. 4e). As we previously determined that overexpression of DLG2 resulted in an increase in IL1B gene expression across all conditions in THP1 cells, we investigated PYCARD/ASC speck formation. We detected that DLG2 overexpression resulted in an increase in PYCARD speck formation (8.3% more, p < 0.01; Fig. 4f) and DLG2 silencing inhibited PYCARD speck formation (6.99% less, p < 0.01; Fig. 4f) in THP1 cells with stably transfected GFP tagged PYCARD compared to the control.
DLG2 activated paracrine signaling and resulted in slower proliferation of colon cancer cells
To determine the effect of paracrine signaling on colon cancer cells we first measured the amount of IL-1β and IL-6 in the supernatant of transfected and activated THP-1 cells by immunoblot (Fig. 5a). We could show that silencing of DLG2 expression resulted in a slight decrease of IL-1β and a strong increase in IL-6, while overexpression of DLG2 had the opposite effect (Fig. 5a). We subsequently tested if the altered expression in IL-6 and IL-1β would affect the tumor microenvironment and modify signaling in colon cancer cells by treating COLO205 cells with the supernatant from THP1 transfected cells combined with regular growth media (1:1) followed by cell growth for 72 hours. We detected that the DLG2 knockdown THP1 cell media increased the proliferation of COLO205 (22.0% more cells/ml, p < 0.05; Fig. 5b), and increased the proportion of cells in G2/M phase (62.8% more G2/M cells, p < 0.05; Fig. 5c) in the COLO205 colon cells when compared to untreated cells. DLG2 overexpression resulted in the opposite of this, decreasing the cell proliferation (7.3% less cells, p < 0.01; Fig. 5b), and the number of cells in G2/M (34,9% less G2/M cells, p < 0.01; Fig. 5c). To show that NFκB and apoptosis signaling pathways were affected in the treated colon cancer cells we visualized protein expression of RELA, RELB, IκBζ, NFKB1, BCL2, BAX, p-STAT3 Ser727, total STAT3 and GAPDH by immunoblot. These results showed that media from THP-1 DLG2-silenced cells decreased protein expression of RELB and BAX, and increased ser727 phosphorylation of STAT3 in COLO205 cells (Fig. 5d and e). Media from THP 1 DLG2 overexpressed cells resulted in increased protein level of RELB, NFκB1 and decreased level of phosphorylation of STAT3 (Fig. 5d and e).
DLG2 expression was low in colon tumors and controlled signaling pathways
We could show that DLG2 gene expression was not significantly different between the ascending and descending colon in healthy controls (Log2 fold change = 0.04, p > 0.05; Fig. 6a). DLG2 expression in the tumor tissue was lower than the paired mucosa sample (Log2 fold change = 1.89, p <0.001), as well as the paired ascending and descending colon mucosa from the distal healthy colon tissue (Log2 fold change = 1.36, p< 0.05, Log2 fold change = 1.32, p < 0.05, respectively; Fig. 6a). Using publicly available microarray colon adenoma data (Sabates-Beliver, Van der Flier et al. 2007) (GSE8671) we determined the expression of DLG2 relative to adenoma size. We could also show that DLG2 expression decreased as colon adenoma size increased to 1.1-1.5 cm and larger than 1.5 cm when compared to tumors under 1 cm in diameter (Log2 fold change = 1.32, p < 0.01 and Log2 fold change = 1.32, p < 0.01; Fig. 6b). To confirm these results, we determined proliferation in the colon cancer cells SW480 after DLG2 silencing or overexpression, DLG2 silencing resulted in an increase in SW480 proliferation (29.4% more cells, p < 0.001; Fig. 6c) and overexpression resulted in a decrease in proliferation (19.6% less cells, p < 0.001; Fig. 6c) compared to the control, 48 hours after transfection. We detected lower level of NLRP3 (Log2 fold change = -1.1, p < 0.05; Fig. 6d), and higher levels of NFKBIZ (Log2 fold change = 1.05, p < 0.05; Fig. 6e) and FOXO3 (Log2 fold change = 0.53, p < 0.05; Fig. 6f) gene expressions after DLG2 overexpression and compared to the mock transfection. At the protein level we subsequently determined that the percentage of phosphorylation of AKT (15.0%, p < 0.05; Fig. 6g), FOXO3 (26.4%, p < 0.05; Fig. 6h), and S6 (51.2%, p < 0.05; Fig. 6i) was lower in Sw480 cells overexpressing DLG2, as visualized in a representative immunoblot (Fig. 6j).