GABA induced by sleep deprivation promotes the proliferation and migration of colon tumors through miR-223-3p endogenous pathway and exosome pathway

The human colon cancer cells(SW480, LoVo), mouse colon cancer cells(MC38), human monocytic cell line(THP1) and 293 T cells were purchased from American Type Culture Collection (ATCC, USA). All cells were cultured in DMEM/high glucose(Gibco, USA).To induce differentiation into macrophages, THP-1 cells (1 × 10⁶) were treated with 100 ng/mL PMA(MCE,USA) for 24 h. All cells were cultured in a medium supplemented with 10% foetal bovine serum (Gibco, USA) at 37 °C under 5% CO2.

The 10 paired of CRC tissues and adjacent noncancerous tissues were collected from Union Hospital, Tongji Medical College, Huazhong University of Science and Technology (Wuhan, China). The consents were obtained from all patients and the use of clinical materials for research purposes were approved by the Ethics Committee of Huazhong University of Science and Technology (Wuhan, China). CRC tissues were taken from the general surgically resected CRC specimens, and the matched adjacent non-cancerous tissues were taken from the site around 5-10 cm away from the edge of the tumors. These specimens were immediately frozen in liquid nitrogen until use. All patients were not received preoperative chemoradiotherapy.

Male BALB/C nu/nu mice (4–6 weeks old) and male C57BL/6 mice (4–6 weeks old) were purchased from Vital River Laboratory Animal Technology Co. Ltd. (Beijing, China). The mice were housed in a specific pathogen-free facility under controlled environmental conditions with a room temperature maintained at 23–25℃. All mice have free access to food and water. The care and handling of the mice were processed following the National Research Council’s animal care guidelines and approved by the Institution Animal Care and Use Committee of Tongji Medical College, Huazhong University of Science and Technology.

To establish the sleep deprivation mouse model, male C57BL/6 mice (4–6 week old) were randomly divided into an experimental group (n = 10) and a control group (n = 10). The mice in the experimental group were placed in a sleep deprivation device and maintained under standard environmental conditions of temperature (21 ± 1° C) and relative humidity (50 ± 10%), with a regular 6-h light/2-h dark cycle. The bottom sliding rod rotated while illuminated to hinder mouse sleep. This process lasts for 7 days. The mice in the control group had the same diet as the experimental group, except that they were not subject to daily sleep deprivation.

One of the xenograft subcutaneous implantation model was that, after the sleep deprivation mouse model was established, a total of 1 × 10⁶ colon cancer cells were suspended in 200μL PBS and injected into mice subcutaneously. The tumors volume was measured every 5 days. All mice were sacrificed after 20 days.

The other one was that normal mice were randomly divided into different treatment groups(n = 5). A total of 1 × 10⁶ colon cancer cells, transfected with mimic, siRNA, overexpression plasmids or not, were suspended in 200μL PBS and inject into mice subcutaneously. The tumors volume was measured every week. All mice were sacrificed after 4 weeks.

One of the pulmonary metastasis model was that, after the sleep deprivation mouse model was established, approximately 5 × 10⁵ colon cancer cells were injected into the tail vein of mice.

The other one was that normal mice were randomly divided into different treatment groups (n = 5). Approximately 5 × 10⁵ colon cancer cells, transfected with mimic, siRNA, overexpression plasmids or not, were injected into the tail vein of mice.

All mice were sacrificed after 2 months in order to evaluate pulmonary metastasis. Metastases were fixed with 4% paraformaldehyde for HE staining and IHC.

We used a 0.4 μm pore Transwell chamber (Corning, USA) to build a co-culture system. One of the co-culture systems was that 4 × 10⁵ colon cancer cells (SW480, LoVo) were plated in the bottom chamber of 6-well plates, while 4 × 10⁵ macrophages (THP1) were added into the upper Transwell insert. The other way was that 4 × 10⁵ macrophages (THP1) were plated in the bottom chamber of 6-well plates, while 4 × 10⁵ colon cancer cells (SW480, LoVo) were added into the upper Transwell insert. The cells were co-cultured for 48 h and were used for subsequent experiments.

To collect the conditioned medium (CM) of THP1, SW480 and LoVo cells, 4 × 10⁵ cells were plated in 6-well plates and cultured in 1640 or DMEM/high glucose for 48 h. The supernatant was collected and centrifuged it at 2000 × g for 10 min to eliminate the cells and cell debris. All the CMs were used instantly or frozen at − 80 ℃.

Total RNA from cultured cells was extracted using Trizol(Japanese TaKaRa) reagent for 5 min at room temperature, then centrifuged at 3000 × g for 15 min at 4℃ to obtain the supernatant. The supernatant was then added to isopropanol and mixed well, then centrifuged at 3000 × g for 10 min at 4℃ to discard the supernatant. The pellet was washed with absolute ethanol and then added to diethylpyrocarbonate in water to measure RNA concentration (ng/μL). The cDNA was then reverse-transcribed by RT Master Mix (TaKaRa, Japan). RT-PCR was performed using SYBR master mix (TaKaRa, Japan) on the Applied Biosystems StepOne-Plus System (American ABI). The expression levels of cellular RNA and mRNA expression were normalized against the housekeeping gene GAPDH. U6 served as a control for miRNA. The primer sequences are listed in Supplementary Table 3.

Lysing cells and tissues with RIPA buffers containing PMSF(American Sigma) and phosphorylase inhibitors(American Sigma). First, protein specimens were separated by SDS-PAGE. Then, the target protein were transferred to the PVDF membrane (Millipore USA). Then, we incubated the PVDF membrane with the corresponding primary antibody (Table S4) overnight at 4℃. This was followed by incubation with secondary antibody(CST, USA) for 1-h the subsequent day. The protein bands were visualized using ECL (Pierce, USA) and collected by the ChemiDocTm XRS Molecular Imager System (Bio-Rad, USA). Finally, the band densities were analyzed by Image J software. (The details are listed in Supplementary Table 4).

HA-tagged CBLB(amino acids 1–982) and its truncations, including CBLB A(amino acids 1–343), CBLB B(amino acids 344–930) and CBLB C(amino acids 931–982),were subcloned into pcDNA 3.1-HA vector. Flag-tagged cMYC(amino acids 1–454) and its truncations, including cMYC A(amino acids 1–368), cMYC B(amino acids 369–421) and cMYC C(amino acids 422–454), were subcloned into pcDNA 3.1-Flag vector.(GeneChem, China). Lipofectamine 3000(Thermo Fisher Scientific, US) was used to transfect plasmids into serum-free Opti-MEM(Gibco).

All the overexpression plasmids targeting cMYC(pcDNA3.1-cMYC) and CBLB (pcDNA3.1-CBLB) were designed and synthesized by GeneChem (China), and the empty plasmid was used as a negative control. MiR-223-3p mimics, miR-223-3p inhibitors and matched control(mimic-NC or inhibitor-NC) were synthesized by RiboBio(Guangzhou, China). All the small interference RNAs were transfected at a final concentration of 50 nM, and the plasmid was transfected at a final concentration of 1.6 μg for 12 well plates. Lipofectamine 3000 transfection reagent (Thermo Fisher Scientific,US) was used for cell transfections according to the manufacturer’s instructions.

A lentiviral vector GV248 containing cMYC shRNA, CBLB shRNA and corresponding negative control were purchased from GeneChem (China). Colon cancer cells were transfected with a lentiviral vector containing cMYC shRNA and CBLB shRNA to establish stable cell lines with downregulated cMYC and CBLB expression. To select stable cell lines, lentiviral-transfected cells were cultured in medium with 1 μg/ml puromycin for 10 days.

Protein and total RNA were extracted after 48 h. (The details are listed in Supplementary Table 5).

Cells were cultured in medium with exosome-free serum to remove the interference of serum exosomes. Briefly, the serum was centrifuged at 100,000 × g for more than 16 h, and filtered with a 0.22 μm filter (Millipore, USA). The medium was collected after 24–72 h, and the exosomes were isolated according to the instructions on the kit (Invitrogen, California, USA). The medium was centrifuged at 2000 × g for 30 min to eliminate the cells and debris. The total exosome isolation reagent was added to it and incubated between 2 ℃ to 8 ℃ overnight. The exosomes were centrifuged at 10,000 × g for 1 h and resuspended in PBS. qNano and electron microscope were used to quantify the size and concentration of the exosomes and to visualize the morphology of the exosomes. Finally, the exosomes were labeled with PKH26 and were extracted again in the reagent. Exosomes were isolated from 4 × 10⁵ colon cancer cells. THP1 cells were plated on 12-well plates the day before treatment. 100 μg exosomes were added to the plates when THP1 cells were treated by PMA for 24 h. The cells were collected after 48 h for the subsequent experiments.

After colon cancer cell was treated by GABA, the supernatant of colon cancer cells were collected. Control group was treated by PBS. Then, the exosomes of supernatant were isolated according to the instructions on the kit (Invitrogen, California, USA). Then, we performed miRNA transcriptome profiling by RNA sequencing. Total RNA extraction, RNA sequencing and bioinformatics data analysis were performed by Qijing Biological Technology Co., Ltd(WuHan, China). The small RNA sequencing library adopts the PE150 sequencing scheme and evaluates the quality value of the sequencing library using fastqc. Use bowtie short sequence alignment tool to compare the Rfam library and remove ncRNAs such as rRNA and tRNA. Use miRDeep2 for quantitative analysis of small RNAs, and use DESeq2 for differential expression analysis.

For analysis of public datasets, RNA-seq-based gene expression data in colon cancer were obtained from the GEO database (GSE74602). For gene expression, P < 0.05 was used as the cutoff.

For prediction of miRNA target genes, the overlap of downstream genes of miR-223-3p were predicted by miRDB(http://​www.​mirdb.​org/​), miRWalk(http://​mirwalk.​umm.​uni-heidelberg.​de/​) and TargetScan(https://​www.​targetscan.​org/​vert_​80/​) database. The final prediction results were obtained by intersecting the prediction results of the three databases.

After transfection, the cells were lysed in RIPA buffer at 4 °C for 1 h. Lysates were centrifuged at 15,000 × g for 15 min at 4 °C, and a sample from the supernatant was collected for further analysis as total lysate. To remove unspecific binding, a prewashing with 0.7 μg/μl IgG-free BSA and 30–50 μl of Protein A/G PLUS agarose was done. Anti-cMYC (6 μg, proteintech) was added to cell lysates and incubated overnight after the addition of Protein A/G PLUS agarose and incubation for 8 h at 4℃. When necessary, a second round of immunoprecipitation was performed. The beads were washed 5 times with 1 ml of immunoprecipitation buffer and then subjected to Western blot analysis.

Te CCK-8 kit was used to assess cell viability in accordance with the manufacturer’s instructions. In brief, cells (3 × 10³ per well) were seeded into 96-well plates (200 μl/well) in culture medium supplemented with 10% FBS with six replicates for each sample. At the appointed time point, 100 μl of fresh medium and 10 μl of CCK-8 solution were added to each well. After incubation for 1 h at 37 ℃, the absorbance was recorded at 450 nm using a Quant ELISA Reader.

5 × 10⁴ cells in medium without FBS were plated in an 8 μm pore Transwell chamber (BD, USA) for the migration assays. And we placed culture medium supplemented with 10% FBS in the lower chambers. Then, the cells were gently wiped away on the top of the filters after incubation for 24 h for the migration assays. The cells were fixed on the membranes with 5% paraformaldehyde for 20 min followed by staining for 15 min with 0.1% crystal violet. Lastly, five fields of vision were chosen and the number of cells were calculated under the microscope.

SW480 and LoVo cells were cultured in 6-well plates with 500 cells per well and allowed to grow for 14 days in the recommended growth medium. The old medium was replaced with fresh medium every 3 days. Then, the clones were fixed with 4% paraformaldehyde for 20 min and stained with 0.1% crystal violet 15 min. Finally, every wells were chosen and the total number of colonies were counted to evaluate the results.

5 × 10⁵ cells were plated in 6-well plates overnight. 1 mL pipette tips were used to scratch a straight line when the cells achieved 60–80% confluence. Then cells were cultured with medium without FBS. Then, a picture of the cell wound width was taken under the microscope at 0 and 48 h.

Tissues were formalin-fixed, dehydrated and paraffin-embedded. Then, the tissue sections were incubated with primary antibodies overnight at 4 ℃. Next day, the tissue sections were incubated with HRP-conjugated secondary antibodies for 1 h at 37℃. Then sections were further washed with PBS and distilled water, freshly prepared DAB solution (diaminobenzidine) was subsequently used until the tissue sections were ready to observe. On the one hand, we evaluated the intensity of tissue staining. We calculated the percentage of positive tissue staining (graded as 0, < 5%; 1, 5–25%; 2, 26–50%; 3, 51–75%; and 4, > 75%). SI score was equal to the product of those two. Two experienced pathologists evaluated all the results from the IHC analysis of the tissue sections.(The details are listed in Supplementary Table 4).

Colon cancer cells and PMA-pretreated THP1 cells were placed on a glass slide and fixed with 5% paraformaldehyde when the cells reached 60–70% confluence. Blocked the cells with 5% donkey serum for 1 h and incubated with the primary antibody overnight at 4 ℃. The antibodies used for the IF assay were listed in Supplementary Table S 4. The next day, the cells were incubated with the corresponding CY3 secondary antibody (Jackson Immunology Research, Erie, UK) at 37℃for 1 h. After counterstaining the nuclei with DAPI (Sigma, St. Louis, Missouri, USA) for 15 min, the fluorescence images were captured by epifluorescence microscopy (Olympus, Tokyo, Japan).(The details are listed in Supplementary Table 4).

ChIP assay was performed using the SimpleChIP® Plus Enzymatic Chromatin IP Kit (CST, USA). Crosslinked the cells with formaldehyde and sonicate to an average size of 300–500 bp. The lysate was added to the EP tube and incubated with the cMYC antibody. Purified cross-linked DNA released from protein-DNA complexes and further evaluated eluted DNA by qRT-PCR. Used both input and IgG to confirm that the detected signal comes from a specific binding between chromatin and MYC. Used JASPAR to predict the binding site between cMYC and the miR-223-3p promoter. All ChIP assays were repeated 3 times independently.

To confirm whether CBLB was a target of miR-223-3p,1 × 10⁵ 293 T cells were seeded on 96-well plates and contransfected with dual luciferase reporter plasmid containing CBLB 3’UTR and mutated forms with or without 50 nM miR-223-3p mimic or mimic-NC.After incubation for 48 h,cells were lysed with diluted Passive Lysis Buffer.Luciferase Assay Buffer II was added and used to measure firefly luciferase activity.After stopping the reaction with 1XStop&Glo®Reagent,we measured the Renilla luciferase activity.Activity per well = firefly luciferase activity/Renilla luciferase activity.

After treatment, blood samples were prepared from C57BL/6 mice. Expressions of Gamma-Aminobutyric Acid (GABA), Norepinephrine (NE), Epinephrine (EPI) and Serotonin(5-HT) in blood were assessed using commercially available ELISA kits (Enzyme-linked Biotechnology Co., Ltd., Shanghai).

All data were analyzed by GraphPad Prism5.0 and all assays were repeated at least three times. We used t-test to analyze the frequency of Macrophages in colon cancer tissues and paired adjacent normal tissues. We used the chi-square test to identify the correlation between miR-223-3p and cMYC, CBLB, p-38 and p-ERK in the colon cancer specimens. χ² test was used to analyze the relationship between Macrophages frequency and the clinical features of colon cancer. P < 0.05 was considered to be statistically significant and all tests were two-sided.

We established a chronic sleep deprivation model mouse to explore the effect of sleep disturbance on tumors. The results indicated that in the sleep-deprivated group, volume and weight of subcutaneous colon cancer tumors were increased (Fig. 1A-C). Additionally, more pulmonary nodules and increased infiltration were observed in the sleep-deprivated group compared to the regular sleep group in the metastasis mouse model (Fig. 1D, E).

To further clarify the mechanisms linking sleep disorder and colon tumors, we assessed the levels of several sleep-related neurotransmitters in the serum of sleep deprivation mice, including GABA, NE, EPI, and 5-HT (Fig. 1F, S1A-C). Notably, the serum concentration of GABA significantly increased in the sleep-deprivated group. We examined brain proteins from the control and sleep-deprivated groups to determine the relationship between chronic sleep deprivation and GABA expression. The expressions of glutamate decarboxylase 1 (GAD1), GABA vesicular transporter (VGAT), and GABA transporter 1 (GAT1) in the cerebrum and cerebellum were examined by WB (Fig. 1G). The results showed that elevated expressions of GAD1 and VGAT in the sleep-deprivated group, indicating increased GABA synthesis and transport capacity in the brain tissue. Conversely, GAT1 expression was significantly decreased in the sleep-deprivated group. GAT1 can reuptake GABA in the synaptic space, which suggests that sleep deprivation maintains high GABA levels in the synaptic area, thus promoting GABA entry into the peripheral blood. We also observed the effect of GABA on the AOM-DSS mouse model (Fig. S1D). These results revealed that GABA increased the number of tumor lesions in the colon in mice (Fig. 1H-I, S1E). Furthermore, the immunohistochemical staining results suggested that GABA promoted proliferation and migration in primary colon tumors(Fig. 1J). In conclusion, our findings suggest that sleep deprivation promote colon cancer proliferation and migration by encouraging GABA synthesis and blood entry.

Previous studies have identified three kinds of GABA receptors: ionic GABAA, GABAC, and metabolic GABAB receptors. Type A and type B receptors have been widely studied and reported. We screened 30 pairs of colon cancer and paracancerous tissues through the GSE database. The heatmap showed it was the expression of the B receptor but not A receptor in colon cancer tissues that was significantly higher than in the paracancerous tissues (Fig. S2A). Then, we collected seven pairs of colon cancer tissues and detected the expression of GABARAP and GABBR2 in the tissues by qRT-PCR. The results showed that both GABARAP and GABBR2 were highly expressed in the tumor tissues, and the fold change of GABBR2 was more significant (Fig. S2B).

Therefore, we speculated that the B receptor were involved in regulating GABA in colon cancer. To further verify the mechanism of GABA-promoting colon cancer progression, we screened colon cancer cell lines with high expression of B receptors: SW480 and LoVo cells (Fig. S2C). We treated tumor cells with different concentrations of GABA in vitro and selected 100 μM for subsequent experiments (Fig. S2D). The results of CCK-8 assays (Fig. 2A), colony formation assay (Fig. 2B), wound-healing assays (Fig. 2C, D), and transwell assays (Fig. 2E) demonstrated that the proliferation ability and migration ability were enhanced when colon cancer cells were treated with GABA.

The results of Western Blot analysis also showed that GABA treatment upregulated the expression of N-cadherin, vimentin, cyclin D1, and cyclin E while suppressing E-cadherin expression (Fig. S2E). In addition, we found increased volume and weight of subcutaneous xenografts in GABA-treated group compared with the PBS group (Fig. 2F-H). At the same time, the effect of GABA on tumor metastasis was verified by tail vein injection of colon tumor cells (Fig. 2I, H). HE staining indicated that the number and size of pulmonary metastatic nodules in the GABA-treated group were higher than the control group (Fig. 2K). These results confirmed that GABA is able to promote the proliferation and migration of colon tumors.

Studies have suggested that cMYC is abnormally expressed in about 70% of human tumors, regulating cell proliferation, differentiation, metabolism, and apoptosis [33]. We found that GABA treatment can upregulate the expression of cMYC in colon cancer cells. Ubiquitination of cMYC protein is considered as an essential means of targeting cMYC to inhibit tumors and has been widely studied [34]. Therefore, we examined whether GABA inhibited the ubiquitination of cMYC protein. Cycloheximide (CHX), a protein synthesis inhibitor, was used to explore the degradation of cMYC protein. After treatment with CHX, the degradation rate of cMYC protein in GABA group was significantly lower than that in PBS group, indicating that GABA may increase the stability of cMYC protein (Fig. 3A). In addition, MG132, a proteasome inhibitor, take an important role influencing the effect of GABA on the deubiquitination of cMYC protein (Fig. 3B), which affirmed that the inhibition of cMYC degradation by GABA is dependent on protease. We further examined the ubiquitinated form of endogenous cMYC. The cMYC protein was immunoprecipitated with an anti-cMYC antibody, and the ubiquitinated condition of cMYC was detected with an anti-ubiquitinated antibody. In the presence of MG132, the ubiquitinated form of endogenous cMYC in the GABA treatment group was significantly lower than in the PBS group (Fig. 3C). It suggests that GABA treatment reduced the ubiquitination form of cMYC in colon cancer cells and enhanced the stability of cmyc protein. To further explore the effect of GABA treatment on promoting the proliferation and migration of colon cancer cells through regulating cMYC, we knocked down cMYC in GABA-stimulated colon cancer cells. The results revealed that cMYC knockdown weakened the regulation of GABA on the proliferation and migration of colon cancer cells (Fig. 3D-G, S2F).

In summary, our data suggest that GABA enhances the stability of cMYC protein, inhibits its ubiquitination degradation, and promotes colon cancer cell proliferation and migration.

It has been previously reported that B-cell-derived GABA could affect the classification of macrophages and inhibit anti-tumor immunity, and another study found that GABA secreted by tumor cells were to activate GSK3 β-Pathway to inhibit CD8⁺T cell-mediated tumor immunity [11, 12]. What is mentioned above suggests that GABA is probably involved in the mutual regulation of tumor immunity. We performed immunofluorescence to detect macrophage markers in mouse lung metastases. Interestingly, macrophage infiltration around lung metastases was increased in GABA-treated group mice. (Fig. 4A). Additionally, co-culturing with GABA-induced colon cancer cells increased the migration ability of macrophages (Fig. S4A). It suggests that GABA promotes the recruitment of colon cancer cells to peripheral macrophages, which is related to tumor progression.

TAMs are the primary members of immune cells infiltrating into tumors. M2 macrophages play an irreplaceable role in regulating tumor growth and metastasis. Here, we investigated the polarization of macrophages recruited in the tumor microenvironment. We induced THP1 cells with PMA to form M0 macrophages, and confirmed its morphological changes (Fig. S3B). The augment of M0 marker CD68 was verified by qRT-PCR (Fig. S3C). First, we found that the polarization of THP1 cells treated with GABA alone did not change significantly (Fig. 4B). However, M2 polarization was ensured after co-culturing with normal colon cancer cells through WB and qRT-PCR. Intriguingly, when co-cultured with GABA-induced colon cancer cells, it exhibited more pronounced M2 polarization (Fig. 4C, D, S3D). Combined with previous experimental results, we speculated that GABA could promote the interaction between colon cancer cells and macrophages. We further detected the effect of the conditioned medium of colon cancer cells on macrophage polarization, and the results showed that GABA-treated colon cancer cell medium induced M2 polarization of macrophages more significantly (Fig. 4E, F, S3E, F).

It has been reported that tumor cells can secrete numerous exosomes, which take a great part in cell interaction. We speculated that GABA may promote the secretion of exosomes by colon cancer cells, thereby causing M2 polarization of macrophages. We collected the culture medium of colon cancer cells in the control and GABA treatment groups respectively to extract exosomes, further identified the extracted exosomes (Fig. S 3G- I). Exosomes were labeled with the fluorescent dye PKH26, and our data showed that exosomes derived from colon cancer cells were fused to macrophages (Fig. 4G). The extracted exosomes of colon cancer cells were co-cultured with THP1 cells for 48 h and detected the polarization of macrophages via WB. These results revealed that the M2 polarization of macrophages induced by exosomes secreted by GABA-induced colon cancer cells was more prominent (Fig. 4H). Immunofluorescence showed the same results (Fig. 4I, S3I). In addition, GW4869, an exosome inhibitor, can reverse this M2 polarization effect of tumor cells on macrophages (Fig. 4J, S3J, K).

In conclusion, our data demonstrates that GABA-induced colon cancer cells recruit macrophages in the microenvironment and aggravate the M2 polarization-promoting effect of tumor cells on macrophages through the exosome pathway rather than direct impact of GABA on macrophages.

MiRNAs are the most abundant substances in exosomes and participates in regulating various cells. To explore the mechanism of GABA-induced colon cancer cell-derived exosome promotion of M2 polarization in macrophages, miRNA sequencing in exosomes was used to perform large-scale expression profiling(Supplementary Table 1). We screened nine up-regulated miRNAs and five down-regulated miRNAs (Fig. 5A) and verified the sequencing results via qRT-PCR (Fig. S4A). Reports have indicated that mir-150-5p and miR-223-3p are involved in the regulation of macrophage M2 polarization. MiR-150 is an immune-related miRNA, which is upregulated in various cancers [35]. MiR-223 is involved in the regulation of inflammation, tumor proliferation, and invasion after transcription. Zhuang et al. reported that miR-223-3p, a new regulator of macrophages, could inhibit standard M1 polarization and promote M2 polarization [36]. More importantly, miR-223-3p acts as an intermediate communication signal between tumor and immune cells in TME [18]. Therefore, we further detected the content of miR-223-3p and miR-150-5p in exosomes. QRT-PCR results demonstrated that GABA promoted the inclusion of miR-223-3p in exosomes, and the content of miR-223-3p in the medium decreased significantly, while the range of miR-150-5p did not change significantly (Fig. S4B). Additionally, treatment with RNAase A and Triton X-100 significantly reversed the increase of miR-223-3p expression in GABA-induced exosomes, however, RNAase alone had no significant effect (Fig. 5B).

After overexpression of miR-223-3p in macrophages, qRT-PCR, WB, and immunofluorescence results showed that the expression of the M2 polarization marker significantly increased (Fig. 5C-E). We found that miRNAs were correlated with the MAPK pathway via KEGG analysis (Fig. S4C). The activation of the MAPK pathway has been proven to be linkedwith the M2 polarization of macrophages. P38 inhibitor and ERK inhibitor were used in THP1 cells overexpressing miR-223-3p, with the results showing that inhibition of the ERK pathway could effectively reverse the M2 polarization of macrophages induced by miR-223-3p. In contrast, inhibition of the p38 pathway had no significant effect (Fig. S4D). We further co-cultured macrophages overexpressing miR-223-3p with colon cancer cells to observe the feedback effect of macrophages on colon cancer. The results proved that co-cultured with mimic transfected macrophages could significantly improve the proliferation and migration of tumor cells(Fig. 5F-I, S4E-H). Furthermore, the results of subcutaneous tumor implantation in mice indicated that co-culturing with macrophages transfected with mimic could accelerate the growth of colon cancer cells (Fig. S5A-C). The results of lung metastasis of tail vein tumor also showed that co-culturing with mimics transfected macrophages enhanced the migration ability of colon cancer cells (Fig. S5D-F).WB revealed that the expressions of increased N-cadherin, vimentin, cyclin E and cyclin D, decreased E-cadherin, in colon cancer cells co-cultured with THP1 cells transfected with mimic (Fig. S5G).

We also collected mimic transfected macrophage-conditioned medium for culturing colon cancer cells. Results showed that the medium of macrophages transfected with mimics significantly enhanced the proliferation and migration of colon cancer cells (Fig. 6A-C, S5H). WB also revealed the same results (Fig. S 5I). We further measured the level of cytokines in a macrophage culture medium with a cytokine Chip (Fig. 6D). At the same time, qRT-PCR was performed to verify the mRNA level of cytokines, and we found that IL-17 increased most significantly (Fig. 6E). The expression of IL-17 increased or decreased upon miR-223-3p overexpression or inhibition in macrophages (Fig. 6G). IL-17 is a typical pro-inflammatory cytokine, which plays a regulatory role in host defense, tissue repair, inflammatory immunity and tumor progression [37]. Moreover, IL-17 augmented the expression of vimentin, CyclinD1, and cyclinE, while attenuating the expression of E-cadherin in colon cancer cells (Fig. 6H). IL-17 was reported to activate JAK-STAT3 and NF- κ B, and other signaling pathways [38]. It can also promote the expression of PD-L1 and cause tumor immune escape [39]. Isibizumab was used to block the IL-17 receptor in colon cancer cells. The results of WB analysis suggested that the JAK-STAT3 pathway were inhibited, and the expression of the marker proteins of proliferation and migration were also reversed (Fig. 6F).

In conclusion, our data suggested that tumor-derived exosomes miR-223-3p activates the MAPK pathway of macrophages to cause M2 polarization and secrete IL-17 to aggravate the proliferation and migration of colon cancer.

Although miR-223-3p is a conservative anti-inflammatory miRNA, studies have indicated its function in tumor proliferation and invasion in prostate cancer, breast cancer, and other tumors [22, 40]. Therefore, we further explored whether the endogenous expression of miR-223-3p in colon cancer cells impacted on tumor proliferation and migration. Results showed that overexpression of miR-223-3p can promote the proliferation of colon cancer cells (Fig. S6A, B), while increasing the migration ability of colon cancer cells (Fig. S6C, D). Our previous data proved that GABA could increase the stability of cMYC protein in tumor cells and reduce ubiquitination degradation. Notably, after treating with CHX in miR-mimics transfected colon cancer cells, the degradation rate of cMYC was also significantly slower than that of the control group (Fig. 7A). Immunoprecipitation also showed that the overexpression of miR-223-3p could reduce the ubiquitinated form of cMYC in tumor cells (Fig. 7B). This suggests that overexpression of miR-223-3p inhibits the normal ubiquitination and degradation process of cMYC protein, resulting in the relative increase of cMYC protein expression in tumor cells.

In this study, we intended to investigate whether GABA promoted the proliferation and migration of colon cancer by regulating the ubiquitination of cMYC protein through miR-223-3p. Inhibition of miR-223-3p can significantly reverse the promotion of GABA on the increase and migration of SW480 and LoVo cells (Fig. 7C-F). GABA inhibited the ubiquitination and degradation of cMYC protein while inhibition of miR-223-3p reversed the inhibitory effect of GABA on cMYC ubiquitination (Fig. 7G).

In conclusion, our data suggest that GABA promotes the proliferation and migration of colon cancer cells by stimulating the endogenous expression of miR-223-3p and the stability of cMYC proteins.

Coincidently, we found that the expression of miR-223-3p increased or decreased upon cMYC overexpression or inhibition in colon cancer cells (Fig. S6E, F). As a transcription factor, cMYC is involved in the regulation of a variety of tumors. Therefore, cMYC may acts as a transcription factor to regulate the transcription of miR-223-3p.

Data from the Jaspar database indicated a potential binding site for cMYC in the miR-223-3p promoter (Fig. S6G). Chip assays showed that compared with IgG-bound samples, cMYC-bound complexes were significantly enriched in the promoter region of miR-223-3p (Fig. S6H).

In conclusion, these data suggest that cMYC regulates miR-223-3p reversely.

To further clarify the specific mechanism of miR-223-3p regulating cMYC ubiquitination, we screened the target genes of miRNA from miRDB, miRWalk and TargetScan databases respectively (Fig. 8A). Notably, there are 56 downstream target genes in the intersection, three of which related to ubiquitination regulation, namely FBXW7, CBLB and SIAH1. QRT-PCR and WB results demonstrated that CBLB differentiates under the intervention of mimic and inhibitor, while both FBXW7 and SIAH1 performed poorly (Fig. 8B, C). Together, we also found that miR-223-3p may bind onto the 3'UTR region of the mRNA of CBLB (Fig. 8D). To confirm that CBLB was a target of miR-223-3p, 293 T cells were transfected with dual luciferase reporter plasmid containing wild-type CBLB 3'UTR (WT) or mutated type (MUT), followed by transfection with miR-223-3p mimic (miR-mimic) or negative control (mimic-NC). The results showed that the luciferase activity in 293 T cells was reduced upon the overexpression of miR-223-3p in the WT group but not in the MUT group (Fig. 8E).

Following, we verified the function of CBLB in colon cancer. CBLB is a proto-oncogene considered as a new target for tumor immunotherapy [41]. However, its expression in colon cancer has not yet been reported. We collected colon tissue samples from 10 patients with colon cancer. WB analysis results showed CBLB was highly expressed in the non-cancerous tissue of colon but lower in the tumor tissues (Fig. S7A). Immunohistochemistry also showed the same results (Fig. S 7B). We also detected the expression of CBLB in subcutaneous colon cancer tumors of control group and SD group. WB analysis results showed that sleep-deprivation decreased the expression of CBLB in subcutaneous colon cancer tumors, while cMYC was highly expressed in subcutaneous colon cancer tumors of SD group (Fig. S 7C). We further transfected SW480 and LoVo cells with CBLB overexpression plasmid and knockdown lentivirus. We found that knockdown of CBLB enhanced the proliferation and migration of tumor cells, while overexpression had the opposite effect (Fig. S7D-F, S8A-C). WB analysis results revealed that knockdown of CBLB increased the expression of cMYC and promoted the expression of proliferation and migration marker proteins, while overexpression of CBLB led to the opposite effect (Fig. 8F). Immunofluorescence results also indicated that downregulating the expression of CBLB promoted the proliferation and migration of colon cancer cells (Fig. S9A). Whereas, the results of subcutaneous tumor formation in mice also revealed that the expression of CBLB was negatively correlated with tumor proliferation (Fig. S9B-D).

We further explored the specific mechanism of cMYC down-regulation induced by CBLB with CHX to detect the degradation of cMYC protein. It was found that the degradation rate of the cMYC protein was significantly faster when CBLB was overexpressed in colon cancer cells (Fig. 8G), compared to the cMYC protein, which was considerably slower when CBLB was downregulated (Fig. S9E). This indicated that CBLB reduced the stability of the cMYC protein. In addition, GABA-induced colon cancer cells were treated with MG132, resulting in increased cMYC expression. However, MG132 did not significantly increase the expression of cMYC in CBLB knockdown colon cancer cells. While MG132 treating CBLB overexpressed colon cancer cells, it showed consistent results (Fig. S 9F, G). It indicates that CBLB regulates the degradation of cMYC through a proteasome-dependent pathway. We further explored the polyubiquitinated form of endogenous cMYC in CBLB knockdown colon cancer cells. In the presence of MG132, the ubiquitination form of endogenous cMYC protein in CBLB knockdown group was significantly reduced (Fig. 8H). To investigate further mechanism CBLB regulates cMYC protein levels, we explored whether CBLB directly binds to cMYC. Endogenous immunoprecipitation demonstrated that CBLB protein binds to cMYC (Fig. 8I, S9H). Immunofluorescence analysis of SW480 and LoVo cells revealed colocalization of CBLB and cMYC in cells (Fig. 8J). These results revealed that CBLB, as an E3 ligase, can recognize and bind the substrate cMYC. It promotes the ubiquitination of cMYC protein, which destroys the stability of cMYC protein and allows it to be degraded by proteasome.To determine the key domains responsible for this interaction, three HA-labeled CBLB truncations were generated based on previous studies on the structure of CBLB, and the interaction of each truncation with cMYC was examined (Fig. S10A). The amino acids 344–930 and 931–982 were responsible for the interaction with cMYC (Fig. S10B). Similarly, three cMYC truncations with Flag tags were designed and tested for their interaction with CBLB (Fig. S10C). The results confirmed that the amino acid 369–421 was vital in maintaining this interaction (Fig. S10D).

Our data suggests that miR-223-3p can target E3 ligase CBLB, and CBLB protein can interact with cMYC to promote its ubiquitination degradation. MiR-223-3p negatively regulates the expression of CBLB and reduces the ubiquitination degradation of cMYC, thus enabling the proliferation and migration of colon cancer cells.

Finally, we conducted a rescue experiment to verify the miR-223-3p/CBLB/cMYC axis involvement in mediating the effects of GABA on colon cancer. GABA increased colon cancer cell proliferation and migration by regulating the ubiquitination of cMYC and promoting the expression of miR-223-3p (Fig. 7C-G). Here, we further investigated whether CBLB mediated the regulation of GABA on cMYC. CBLB was more expressed in GABA-induced colon cancer cells. Overexpression of CBLB significantly decreased the expression of cMYC and inhibited the proliferation and migration of colon cancer cells, while GABA was able to rescue the inhibitory effect of overexpression of CBLB on the proliferation and migration of tumor cells (Fig. 9, S11A-D).

We further explored the effect of the miR-223-3p/CBLB/cMYC axis on colon cancer by another rescue experiment. We transfected mimics into SW480 and LoVo cells and then transfected the CBLB overexpression plasmid. The results demonstrated that overexpression of CBLB could reverse the promotion of mimics on the proliferation and migration of tumor cells (Figs. 10 and 11, S11E). Immunoprecipitation results revealed that overexpression of miR-223-3p enhanced the stability of cMYC and reduced the ubiquitinated degradation of cMYC, while overexpression of CBLB reversed this effect (Fig. 9E). In addition, we also transfected inhibitor into CBLB knockdown tumor cells, and the results of WB analysis and immunoprecipitation also suggested the same results (Fig. 9F). The development of the rescue function experiment is also consistent with our conclusion (Fig. S11, F).