The CRC tissue samples were obtained from 58 patients who were diagnosed and underwent surgery from January 2010 to December 2011 in Jiangmen Central Hospital, Affiliated Jiangmen Hospital of Sun Yat-sen University. A written informed consent to use the surgical samples has been signed by all the patients, and the study was approved by the Institutional Review Board of Nanfang Hospital affiliated to Southern Medical University. Collected samples were fixed by paraformaldehyde and embedded in paraffin.

Paraffin-embedded tissue was cut into 5 μm thick slices, dewaxed with xylene, and hydrated with ethanol. Then, the slides were heated at 95°C in a 0.01 M citrate buffer (pH = 6.0) and quenched for peroxidase activity with 3% hydrogen peroxide for 20 min to retrieve antigens. After being treated with 10% goat serum, the slides were incubated overnight with antibodies at 4°C. PBS washing was followed by incubation with goat anti-rabbit IgG for 1 h and then stained with 3,3-diaminobenzidine. When all sections were dehydrated and sealed, we selected images with an inverted microscope (ZEISS). Immunohistochemistry was performed with the following antibodies: Ki67 (27309-1-AP; Proteintech), MALT1 (66225-1-Ig; Proteintech), p-p65 (AB11014, phosphor-Ser536), fibronectin (66042-1-Ig; Proteintech), N-cadherin (22018-1-AP; Proteintech), E-cadherin (20874-1-AP; Proteintech), vimentin (10366-1-AP; Proteintech) Snail1 (26183-1-AP; Proteintech), and Snail2 (12129-1-AP; Proteintech).

The mRNA microarray datasets (GSE21510, GSE17536) and miRNA microarray datasets (GSE38389, GSE41655, GSE30454, GSE18392) were obtained from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/). In GSE21510, patients were divided into the normal group and tumor group according to the type of tissue resected in the surgeries. The expression of MALT1 is compared by the content of corresponding mRNA. Meanwhile, in GSE17536, patients were divided into high expression group (85%) and low expression group (15%) depending on the protein expression of MALT1, so that we were able to judge the relationship between MALT1 and prognosis by comparing the time of disease-free survival.

The human CRC cell lines HCT116 and SW480 and the human normal colorectal epithelial cell line NCM460 were purchased from American Type Culture Collection (ATCC, Manassas, VA, United States). All cell lines were cultured in Dulbecco’s modified eagle medium (DMEM; Life Technologies) supplemented with 10% fetal bovine serum (FBS; Beyotime Biotechnology, Shanghai, China), 100 μg/ml streptomycin, and 100 units/ml penicillin. They were all placed in a humidified incubator with 5% CO₂ at 37°C and a mild atmosphere.

Total RNA was extracted with TRIzol Reagent (ambin, United States). After gDNA was removed, 500 ng of total RNA was reversed with Evo M-MLV RT Kit with gDNA Clean for qPCR II AG11711 (Accurate Biology) according to the manufacturer’s instructions. Real-time PCR was performed using PerfectStart Green qPCR SuperMix (TransGen Biotech) on a LightCycler 96 Detection System (Roche). GAPDH was served as an internal reference gene, and the primers used for qPCR in this study were as follows: MALT1: F, TGG​AAG​CCC​TAT​TCC​TCA​CTA​CC; R, CAT​GAC​ACC​AG-TAG​GTT​CCT​TGG, GAPDH: F, GAT​ATT​GTT​GCC​ATC​AAT​GAC​C; R, AGC​C-TTC​TCC​ATG​GTG​GTG​AAG​A, miR-218-5p: F, TTG​TGC​TTG​ATC​TAA​CCA​TGT; miR-338-3p: F, TCC​AGC​ATC​AGT​GAT​TTT​GTT​G; miR-365-3p: F, TAA​TG-CCC​CTA​AAA​ATC​CTT​AT; miR-375-3p: F, TTT​GTT​CGT​TCG​GCT​CGC​GTG​A; and U6: F, CGT​TCA​CGA​ATT​TGC​GTG​TCA​T. The reverse primer used in the qPCR of miRNA was the mRQ 3′ primer supplied with the microRNA first-strand synthesis and miRNA quantitation kits (Takara).

Total protein was extracted using a cell lysis buffer (Beyotime, Shanghai, China) mixed with a protease inhibitor cocktail on ice, and protein concentration was determined by the Bradford method. SDS-polyacrylamide gel electrophoresis was use to separate the desired proteins. Subsequently, proteins were transferred to PVDE membranes and blocked with 10% milk powder solution. Membranes were placed in a primary antibody overnight at 4°C and then incubated with a secondary antibody at room temperature for 2 h. Immunoblotting was performed with the following antibodies: β-actin (47,778; Santa Cruz Biotechnology), GAPDH (47,724; Santa Cruz Biotechnology), p-p65 (AB11014, phosphor-Ser536), MALT1 (2494S; Cell Signaling Technology), A20 (WL00820; Wanleibio), RELB (WL02922; Wanleibio), Fibronectin (66042-1-Ig; Proteintech), N-cadherin (WL01047; Wanleibio), E-cadherin (20874-1-AP, Proteintech), Vimentin (10366-1-AP; Proteintech) Snail (WL01863; Wanleibio), secondary antibodies’ anti-mouse IgG-HRP (VC297958; Invitrogen), and anti-rabbit IgG-HRP (VC297287; Invitrogen). The data were acquired by Tanon 5200 system (Tanon Science) with Luminata Forte Western HRP substrate (Millipore).

HCT116 and SW480 were seeded into six-well plates at a number of 1 × 10⁶ cells per well, and the cells were transfected with plasmid (MALT1-pcDNA3.1-3xFlag-C; Hunan Fenghui Biotechnology Co.) and siRNA for 6 h. After transfection, cells were seeded into 96-well plates at 1,000 cells per well. CCK-8 was added to each well according to the instructions, and after waiting for 2 h of reaction, the OD value at 450 nm was measured with a microplate reader.

For the colony formation assay, transfected HCT116 were seeded into 6-well plates at 500 cells per well and transfected SW480 were seeded into 6-well plates at 500 cells per well. All cells were cultured in DMEM supplemented with 10% FBS for 14 days. After 14 days, the colonies were fixed with 3.7% formaldehyde permeabilized with pure methanol and stained with 0.1% crystal violet. The ability of colony formation was determined by counting the number of stained colonies.

The siRNA (RIBOBIO) used in this study were as follows: siMALT1_1: sense, CCGGAGAUAAUAAUGUGUGdTdT; antisense, CACACAUUAUUAUCUCCG-GdTdT, siMALT1_2: sense, CUACGAUGAUACCAUUCCAdTdT; and antisense, UG-GAAUGGUAUCAUCGUAGdTdT. The miRNA mimic and inhibitor (RIBOBIO) used in this study were as follows: miRNA-365-3p mimic: sense, AGG​GAC​UUU​UGG​GGG​CA-GAU​GUG; antisense, CAC​AUC​UGC​CCC​CAA​AAG​UCC​CU, miRNA-365-3p inhibitor: sense, CAC​AUC​UGC​CCC​CAA​AAG​UCC​CU, miRNA-375-3p mimic: sense, UUU​GUU​CGU​UCG​GCU​CGC​GUG​A; and antisense, UCA​CGC​GAG​CCG​AAC​G-AAC​AAA; miRNA-375-3p inhibitor: sense, UCA​CGC​GAG​CCG​AAC​GAA​CAA​A.

HCT116 and SW480 cells (2.5 × 10⁶) transfected with MALT1 or siMALT1 were seeded to the top migration chambers (Millipore) with 200 μl FBS-free DMEM, and 750 μl of DMEM with 10% FBS was added to the lower chamber. After 30 h, the upper chambers were washed with PBS, fixed with 3.7% formaldehyde, permeabilized with pure methanol, and stained with 0.1% crystal violet. Photographs were recorded under an inverted microscope (ZEISS), and the number of cells was then counted with ImageJ.

For the immunocytochemistry assay, transfected HCT116 were seeded on glass coverslips in 12-well plates at 2.4 × 10⁵ cells per well. About 24 h later, the cells on glass coverslips were washed by PBS twice, fixed with 4% paraformaldehyde for 30 min, and then permeabilized by 0.5% NP-40 for 20 min at room temperature. After being blocked with 3% BSA (bovine serum albumin; Biotopped), glass coverslips were placed in primary antibody dissolved in 3% BSA overnight at 4°C and then incubated for 1 h with specific fluorescence-conjugated secondary antibodies in the dark. 4′,6-diamidino-2-phenylindole was used to stain the cells on glass coverslips. Photographs were collected under an inverted microscope (ZEISS).

HCT116 and SW480 (8 × 10⁴) was seeded into 24-well plates, co-transfected with NF-κB-Luc reporter vector (Beyotime Biotechnology, Shanghai, China), MALT1 and siMALT1. About 48 h later, luciferase intensity was measured using a dual-luciferase reporter assay kit (Beyotime Biotechnology, Shanghai, China), and the expression of each component was analyzed according to the instructions.

For the xenograft experiments, BALB/c nude mice (6 weeks old, female) were adopted from Guangdong Medical Laboratory Animal Center. As mentioned earlier, HCT116 cells (5 × 10⁶ cells in 50 μl DMEM) transfected with siNC or siMALT1 were subcutaneously inoculated into the scapular region of nude mice. The body weight and tumor volume [largest diameter × (smallest diameter)²/2] were measured daily, and after 7 days, a mixture of DMEM (50 μl) with siNC or siMALT1 (12 μl) and LIPO 2000 (12 μl) was injected into the tumor location every 2 days. Tumors were removed immediately after the mice were sacrificed on Day 21 and frozen at −80°C to prepare for the following experiments.

For the in vivo suppression studies of MI-2, HCT116 cells (5 × 10⁶ cells in 50 µl DMEM) were subcutaneously inoculated into the scapular region of nude mice (6 weeks old, male). Three days later, MI-2 and dimethyl sulfoxide (DMSO) dissolved in PBS were injected intraperitoneally at a dose of 25 mg/kg body weight in a volume of 10 μl/g body weight every 2 days. Mice were killed and the tumors were removed after 14 days. All procedures were approved by the Southern Medical University Animal Care and Use Committee. All animal studies were conducted in accordance with institutional guidelines.

The experimental results were expressed as mean values ± SD. A two-tailed Student’s t-test, paired t-test, log-rank test, and one-way ANOVA were used for the analysis of statistical difference between two groups in IBM SPSS statistics 25. p < 0.05 was considered statistically significant.

To explore the relationship between MALT1 and the malignant progression of CRC, we first analyzed the difference of the MALT1 mRNA level between CRC and adjacent tissues based on the Gene Expression Omnibus (GEO) database (GSE21510). The data demonstrated that MALT1 expression was significantly increased in CRC tissues (n = 123) compared with adjacent tissues (n = 25) (Figure 1A). Another GEO series GSE17536 showed that patients with high MALT1 expression (n = 22) had a worse prognosis compared with those with low MALT expression (n = 123) (Figure 1B), which reminded us of the significance of MALT1 in CRC. To further confirm the expression of MALT1 in CRC, we examined the protein level of MALT1 by immunohistochemistry (IHC) staining in colorectal tissues from 58 CRC patients compared to adjacent tissues. It was observed by microscope that MALT1 was mainly expressed in tumor tissues, and the IHC scores of tumor tissues were significantly higher than those in adjacent tissues (p < 0.001) (Figure 1C). The protein and mRNA levels of MALT1 in CRC lines (SW480, HCT116) and a normal colorectal mucosa cell line (NCM) were detected by immunoblotting and qPCR. Consistent with the results from IHC, the protein and mRNA levels of MALT1 in the SW480 and HCT116 cells were both higher than those in NCM (Figures 1D,E). In light of these findings described, it is concluded that MALT1 was highly expressed in CRC cells and tissues, and we speculated that MALT1 may be associated with the malignancy of CRC.

In order to identify the role of MALT1 in CRC malignancy, we transiently transfected HCT116 or SW480 cells by siRNA and an expression plasmid, and the protein levels of MALT1 were determined by immunoblotting (Supplementary Figures S1A,B). The CCK8 assays were conducted, and the results revealed a significant decrease of absorbance in HCT116 or SW480 cells following MALT1 knockdown, compared with the control group (Figure 2A). As shown in Figures 2B,C, silencing the expression of MALT1 also decreased the number of colonies and the staining intensity of the proliferation marker Ki67. A small molecule inhibitor MI-2 was selected to inhibit MALT1 protease activity irreversibly. The cell proliferation of HCT116 and SW480 cells treated with MI-2 was significantly inhibited compared with control (Figure 2D). Then, a dose-dependent decrease of cleaved RelB or A20 was induced by MI-2, verifying its inhibitory effect (Figure 2E). When MALT1 was upregulated in CRC cells, the results were opposite with the silence of MALT1 (Supplementary Figures S1C–E). These data above suggested that MALT1 could promote the cell proliferation of CRC cells in vitro.

Metastasis is a vital hallmark of carcinomas progressing to higher pathological grades of malignancy. Therefore, to further investigate the effect of MALT1 on the malignancy of CRC cells, we performed transwell assays to test the migration ability of CRC cells after transfection. The results showed that the migration abilities in both HCT116 and SW480 cells were significantly suppressed after MALT1 was silenced by siRNA transfection (Figure 3A). In addition, the expression of EMT-related proteins, which were indicators of tumor metastasis, were detected by immunoblotting. As shown in Figure 3B, the knockdown of MALT1 resulted in decreased expressions of vimentin, fibronectin, Snail, and N-cadherin and the increased expression of E-cadherin. Meanwhile, we examined the expression of EMT-associated proteins by IF, and the results were consistent with those by immunoblotting (Figure 3C). Similar to the knockdown of MALT1, MI-2 treatment inhibited CRC cell migration in a dose-dependent manner (Figures 3D,E). Inversely, overexpressed MALT1 promoted cell migration in CRC cells (Supplementary Figures S2A–C). Generally, the data mentioned above demonstrated that MALT1 strongly enhanced the migration ability of CRC cells to promote the malignant phenotype.

To evaluate the pro-oncogenic function of MALT1 in vivo, we established a subcutaneous tumor model. HCT116 cells were transfected with two different siRNAs targeting MALT1, and 1 × 10⁷ cells with MALT1 knockdown or negative control were injected subcutaneously into the two flanks of nude mice, respectively. Compared with the negative control group, the tumor volumes in the MALT1 knockdown group were significantly decreased in both two siRNAs (Figure 4A and Supplementary Figure S3A). Tumors were removed 21 days after inoculation. Then, the results showed that the MALT1 expression level, detected by IHC, was significantly decreased in the MALT1 knockdown group (Figure 4C and Supplementary Figure S3B). In accordance with our in vitro observations, the staining intensity of the proliferation marker Ki67 was much lower in xenografts derived from the MALT1 knockdown group than those in the control group (Figure 4C and Supplementary Figure 3B). As for the migration indicator, we performed IHC assays to analyze the expression of EMT-associated proteins, and the results were similar to in vitro results (Figure 4C and Supplementary Figure S3C).

Meanwhile, HCT116 cells were injected subcutaneously to identify the effect of MI-2 in vivo. Three days after injection, mice were randomly divided into two groups, the DMSO and MI-2 groups. The results showed that the tumor volumes in the MI-2 group were significantly reduced compared to the DMSO group (Figure 4B). Consistently, the expression of Ki67 and EMT-associated proteins was lower in MI-2 treated mice than that in DMSO (Figure 4D). Taken together, these results showed that the inhibition of MALT1 or its protease activity could suppress tumor growth and migration in vivo.

To investigate the downstream molecular mechanism of MALT1 promoting CRC malignancy, we detected the activation of the NF-κB pathway, which has been reported as the downstream. The phosphorylation of p65 is a hallmark of NF-κB activation, which results in p65 nuclear translocation and further transcriptionally activates the target genes. Given that, we examined the expression of phosphorylated p65 by immunoblotting, and the results showed that the downregulation of MALT1 inhibited the phosphorylation of p65, whereas the upregulation of MALT1 increased it (Figure 5A). In vivo experiments also demonstrated that MALT1 knockdown or its protease activity inhibition led to a decrease of phosphorylated p65 in mice subcutaneous tumor tissues (Figure 5B and Supplementary Figure S3B). To verify the activation of the NF-κB pathway, luciferase reporter assays were also performed. As shown in Figures 5C,D, the luciferase activity of the NF-κB reporter was 4–6 times in the negative control group that of the inhibition of the MALT1 group, and overexpressed MALT1 increased the transcriptional activity compared with the control group. According to our data, MALT1 could promote CRC development by targeting the NF-κB signaling pathway (Figure 5E).

To gain a better understanding of the complete regulatory axis of MALT1-NF-κB, miRcode and GEO databases were carried out to identify a number of miRNAs targeting MALT1. Through miRcode database prediction, 97 miRNAs probably targeted MALT1 (Supplementary Table S1). Among these miRNAs, downregulation miRNAs in CRC tissues compared with normal tissues were sorted out based on GEO databases (GSE18392, GSE38389, GSE41655, and GSE30454, Supplementary Table S2). We chose miR-375, miR-365a-3p, miR-338-3p, miR-218, and miR-429 for further exploration, which was included in more than two GEO databases (Figures 6A,B). Then, real-time PCR experiments were exerted to detect the relative RNA expression of each miRNA in NCM and CRC cells, and the results showed that all those expressions were higher in NCM cells as expected (Figure 6C). Since MALT1 was targeting the NF-κB pathway to increase the malignancy of CRC, miR-365a-3p and miR-375 were selected as reported in previous studies (Sun et al., 2018; Yin et al., 2019). Through the luciferase reporter assays, the results confirmed that the overexpression of miR-365a-3p or miR-375 inhibited NF-κB activation, which is in accordance with the results of MALT1 inhibition, and the downregulation of miR-365a-3p or miR-375 promoted NF-κB activation (Figures 6D,E). Furthermore, the mRNA expression of MALT1 regulated by miR-365a-3p or miR-375 was detected by qPCR. As expected, MALT1 expression was significantly decreased after the miR-365a-3p or miR-375 mimic treatment and increased after the miR-365a-3p or miR-375 inhibitor treatment (Figure 6F). In conclusion, miR-365a-3p or miR-375 might target MALT1 to inhibit NF-κB activation in CRC cells (Figure 6G).