NE/β2-AR Promotes Directional Migration of Colorectal Cancer Cells to the Liver by Inducing M2 Polarization of Kupffer Cells and Upregulating CXCL12 Expression

Background and Aim The exact mechanism of colorectal cancer (CRC) liver metastasis remains unclear. This study aimed to explore the role and mechanism of the norepinephrine (NE) in the directional migration of CRC cells to the liver. Methods In mouse models of CRC liver metastasis, the effects of NE on the number of liver metastases and the density of intrahepatic Kupffer cells (KCs) were observed. In vitro experiments were performed to detect KC polarization markers by ow cytometry, cytokines by enzyme-linked immunosorbent assay (ELISA), migration of colon cancer cells by Transwell migration assay, AR expression and PI3K/Akt signaling pathway-related protein expressions by Western blotting, and chemokine mRNA expressions by reverse transcription-polymerase chain reaction (RT-PCR). Results Two weeks after intraperitoneal injection of different doses of NE in CRC liver metastasis mouse models, the number of liver metastases were (111.00±51.43) and (102.40±54.85) in the low-dose (0.28 nmol) NE group and high-dose groups (2.8 nmol), respectively, which were signicantly higher than those in the control group (both P(cid:0)0.05); in addition, the density of intrahepatic KCs in the NE group was signicantly reduced compared with the control group (P<0.05). In vitro experiments showed that low-concentration NE induced M2 polarization of KCs; NE upregulated the expression level of NE receptor β 2 AR on KCs and activated the PI3K/Akt pathway, while blocking β 2 -AR or using a PI3K inhibitor inhibited this process (P< 0.05). M2 KCs promoted the migration of colon cancer cell line CT26. Eight macrophage-associated chemokines were screened from the TISIDB website. CXCL12 expression in KCs was signicantly higher in low-concentration (10 -9 M) NE group than in controls, and a β2-AR blocker down-regulated CXCL12 expression in KCs (both P < 0.05). Conclusions NE/β 2 -AR may induce intrahepatic KC M2 polarization through the PI3K/Akt pathway and promote its secretion of CXCL12 to induce the directional migration of CRC to the liver. Our ndings provide important evidence in the search for new strategies to prevent liver metastasis from CRC.


Introduction
Colorectal cancer (CRC) is the third most common cancer and the second most common cause of cancerrelated death globally. [1] Due to a variety of reasons, most CRC patients are already in the advanced stages when they are diagnosed, and 15%-25% of them have liver metastases. [2] The rate of liver metastases can reach 50% throughout the course of CRC, and the median survival time for patients with untreated liver metastases is only 6.9 months. [3] The main treatment options for liver metastases from CRC currently include surgery, radiotherapy, chemotherapy, and molecularly targeted therapy; however, more than half of patients will experience recurrence of colorectal liver metastases within 2 years. [4] Therefore, compared with the treatment of liver metastases, preventing or reducing the occurrence of liver metastases will help improve the prognosis of CRC patients.
Clinically, it is common that the incidence of liver metastases found during the follow-up period varies considerably among CRC patients in the same stage and receiving the same treatment, which highlights the potential role of the microenvironment of the target organ in the body. [5] The liver is not only the largest digestive and metabolic organ but also a very important part of the immune system. [6] A variety of cells including liver sinusoidal endothelial cells, Kupffer cells (KCs), hepatocytes, myeloid-derived suppressor cells (MDSC), and tumor-associated neutrophils (TAN) interact to form a complex immune microenvironment and jointly participate in the immune regulation of the liver. [7] KCs were liver-resident macrophages located in hepatic sinusoids, accounting for 80-90% of the monocyte-macrophage cell system. The functions of KC can be in uenced by a variety of factors including endotoxin, lipopolysaccharide, and stress. [8] When activated, KCs can produce a variety of cytokines such as vascular endothelial growth factor (VEGF), TGF-β, IL-6, and TNF-α, thus exerting biological effects (e.g. killing effect). [9] It has been found that NOTCH signaling promotes intrahepatic KC polarization towards M2 macrophages, which ultimately promotes CRC liver metastasis. [10] The interaction between the liver and the intestine, known as the "liver-gut axis", are always present under both pathological and physiological states. For example, intestinal ora disorders can induce in ammatory responses and brosis in the liver, leading to cirrhosis and hepatic encephalopathy. [11] In addition, abnormal gut ora can affect the occurrence of liver metastases from CRC. [12] There is growing evidence that neurons and their neurotransmitters play an important role in CRC liver metastasis. [13] The levels of the stress-related hormone NE and its β-adrenergic receptor (β-AR) dramatically increase in CRC tissue and are strongly associated with poor prognosis in CRC patients. [14,15] Further studies have revealed that β-AR is also widely expressed on immune cells and can in uence the immune microenvironment of tumours by regulating the function of immune cells and thus affect tumor occurrence and progression . [16] Activation of β2-AR reduces the in ltration of CD8+ CTL cells in the tumor microenvironment and inhibits the activation of CD8+ T cells; NE also inhibits the cytotoxicity of NK cells and reduces the secretion of IFN-γ through the β2-AR/cAMP/PKA/p-CAEB pathway; activation of the β 2 -AR pathway promotes STAT3 phosphorylation, which in turn signi cantly increases the survival of MDSCs in the tumor microenvironment and enhances their immunosuppressive function, ultimately promoting tumor progression. [17][18][19] To date, it is still unclear whether and how NE can affect the development of CRC liver metastases. This study aims to explore the potential role and mechanism of the enteroneurotransmitter NE during CRC liver metastasis, in order to provide evidence for establishing new prevention and treatment strategies of CRC liver metastasis.

Animal experiments
The mouse models of CRC liver metastasis were injected intraperitoneally with different doses of NE (TCI, China). Eighteen SPF-grade male BALB/c mice aged 8 weeks were randomly divided into three groups 3 days after successful modeling: a) control group, injected with normal saline; b) low-dose NE group, injected with 0.28 nmol NE; and c) high-dose NE group: injected with 2.8 nmol NE. NE was prepared with 0.9% saline and stored in a refrigerator at -20℃. NE was prepared every 3 days and injected intraperitoneally every 2 days (0.1 ml) for a total of 8 sessions.

Immunohistochemistry
Mouse liver tissue was collected, para n-embedded, and sliced into 4-µm-thick sections. After dewaxing, hydration, antigen repair, and incubation with 3% H 2 O 2 solution in the dark at room temperature, rabbit anti-mouse primary anti-F4/80 monoclonal antibody (1:500, Servicebio, China) was added, and color development was performed using a DAB staining immunohistochemistry kit (Zsbio, China). Under microscope, three high-magni cation elds were randomly selected for image acquisition, and the positive areas were analyzed using the Image J software.

Cell culture
Both the murine colon carcinoma cell line CT-26 (Cell Bank of Chinese Academy of Sciences, China) and the KCs (Jennio Biotech, Guangdong, China) were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (100 U/mland 100 µg/ml, respectively) at 37°C with 5% CO 2 in a thermostat incubator.

Enzyme-linked immunosorbent assay (ELISA)
All the ELISA kits were purchased from Arigo, USA. After blocking, antibodies TGF-β, IL-6, and VEGF were added for inoculation. After the enzyme reaction was terminated, the optical density (OD) of each well was measured at 450 nm.

Flow cytometry
After KCs were inoculated at a density of 1⋅10 5 cells/well in 6-well plates for 24 h, 100 µl PBS as well as Subsequently, the culture medium was replaced with 500 µl RPMI-1640 medium containing 20% FBS.
CT26 cell suspension free of FBS was inoculated at a concentration of 4⋅10 4 cells/100 µl in the upper chamber and co-inoculated with the pre-treated KCs for 24 h. The cells were xed in 4% paraformaldehyde solution for 30 min and stained with 0.1% crystalline violet for 20 min. Three elds viewed at 200X were randomly selected for each group. The number of cells crossing the microporous membrane in the upper chamber was counted using the Image J software.
TISIDB (http://cis.hku.hk/TISIDB/) enables the analysis of the associations of a selected gene with the clinical features of tumors, immunomodulators and chemokines. In order to understand whether chemokines are correlated with macrophages, we analyzed the chemokines including CXCL12, CXCL9, CCL2, CCL20, CXCL8, CXCL16, CX3CL1, and CCL17 in CRC through the TISIDB website.

Quantitative reverse transcription PCR (RT-qPCR)
KCs were treated with PBS and 10 − 9 M NE for 24 h, and then the total RNA was extracted using TRIzol reagent (Invitrogen, USA). cDNA reverse transcription was performed using PrimeScript™ RT reagent kit.
The obtained data were analyzed by the 2 −ΔΔCT method. The primer sequences of CXCL12, CXCL9, CCL2, CCL20, CXCL8, CXCL16, CX3CL1, and CCL17 are shown in Table 1.  1A-B). To observe the effect of NE on KCs, we performed immunohistochemical staining using F4/80 as a marker of KC cells. It was found that KCs were mainly distributed in the hepatic sinusoids and the central vein or con uent regions of the hepatic lobules, and the distribution of KCs in both the low-and high-dose NE groups was less than that in the control group (both P < 0.05), while there was no statistical difference between the two NE groups (Fig. 1C).

NE induced the M2 polarization of KCs
After 24 h incubation, KCs were treated with gradient concentrations of NE. The polarization status of KCs was measured at 24 h (A CD86 marker for M1 and a CD206 marker for M2). The results showed that all concentrations of NE inhibited CD86 expression on the surface of KCs (P < 0.01). In addition, low to medium concentrations of NE (10 −9 M, 10 −8 M, and 10 −7 M) promoted CD206 expression on the surface of KCs (P < 0.01). However, when the concentrations increased to 10 −6 M and 10 −5 M, NE had no effect on CD206 expression on the surface of KCs ( Fig. 2A-B).
Furthermore, we examined the effect of NE on KC-secreted cytokines by ELISA. The results showed that each concentration of NE inhibited the secretion of IL-6 by KC in a dose-dependent manner (P < 0.01).

NE induced KC polarization via β 2 -AR on the surface of KCs
To identify the speci c receptor via which NE exerts an effect on KC polarization, we selected the lowest concentration (10 −9 M) of NE in the above experiment to treat KC rstly and then detected the expressions of α 2 -AR and β2-A 2 on the surface of KCs after 24 h. The results showed that low-concentration NE induced up-regulation of β2-AR expression in KCs but had no effect on α 2 -AR expression (Fig. 3A), suggesting that NE may exert its effect via the β2-AR on the surface of KCs.
KCs were further pretreated with gradient concentrations ((10 −9 M, 10 −8 M, 10 −7 M, 10 −6 M, and 10 −5 M) of a β 2 -AR blocker ICI118551, and an equal concentration of NE (10 −9 M) was added 2 h later before incubation for 24 h. It was found that the expression level of CD206 on KC decreased after the addition of a β 2 -AR blocker compared to the control group, and there were statistical differences (P < 0.05) when the concentrations reached 10 −6 M and 10 −5 M (Fig. 3C).
ELISA revealed that the secretion of IL-6 showed an increasing trend after treatment of KC with gradient concentrations of the β 2 -AR blocker ICI118551, and the difference was statistically signi cant (P < 0.01) at the concentrations of 10 −5 M. Although no statistical difference was observed, KCs treated with the β2-AR blocker showed a decreasing trend in the secretion of cytokines VEGF and TGF-β (Fig. 3B). Therefore, NE might promote the M2 polarization of KCs via β 2 -AR on the surface of KCs.
NE affected KC polarization and function by activating PI3K/Akt pathway Differential genes were screened by comparing genes from CRC non-liver metastasis samples and liver metastasis samples in datasets GSE81980 and GSE18105. The obtained differential genes underwent further functional and pathway enrichment analyses using the DAVID website. It was found that GO categories were enriched in cellular component, biological process, molecular function, serine/threonine protein kinases, serine phosphorylation, threonine phosphorylation, and DNA damage checkpoint signaling. The KEGG analysis revealed that the most signi cant enrichment pathways were PI3K-Akt signaling pathway, sphingolipid signalling pathway, β-Adrenergic signaling in cardiomyocytes, and neurotrophin signaling pathway (Fig. 4. A).
To verify that NE promotes KC M2 polarization through PI3K/Akt signaling, we treated KCs with PBS in the control group and gradient concentrations of NE in the NE groups and detected the expressions of PI3K, PAN-Akt, and p-Akt proteins in KCs by Western blotting after 24 h. The results showed that low concentrations (10 −9 M and 10 −8 M) of NE induced an increase in PI3K protein expression compared to the control group (P < 0.01). In addition, the expressions of PAN-Akt and p-Akt proteins in KCs also increased in the low-concentration (10 −9 M) NE group (P < 0.05) (Fig. 4B).
All the gradient concentrations of the β 2 -AR receptor blocker inhibited the expressions of PI3K, Akt and p-Akt proteins on KCs (Fig. 4C). Thus, NE might promote high expressions of PI3K, PAN-Akt, and p-Akt proteins in KCs by acting on β 2 -AR. We further applied the PI3K inhibitor LY294002 to block the PI3K/Akt signaling pathway and then gave NE treatment; it was found that the expression of CD206 on the surface of KCs decreased, and the amounts of VEGF and TGF-β secreted by LY294002-treated KCs showed a decreasing trend, and the difference was statistically signi cant when the concentration was 40 µmol/L (P < 0.01); IL-6 showed an increasing trend but did not yet reach statistical difference. (Fig. 4) It could be concluded that NE/β2-AR promoted KC M2 polarization through the PI3K/Akt pathway.

M2 KC promoted the directional migration of CT26 cells via CXCL12
After Eight macrophage-assocaited chemokines were screened from the TISIDB website. As veri ed by RT-PCR, the expression of CXCL12 in KCs induced by low concentration (10 −9 M) of NE was signi cantly higher than that in the control group (P=0.006), while the expressions of CXCL9, CCL2, CCL20, CXCL8, CXCL16, CX3CL1, and CCL17 were signi cantly lower than those in the control group (all P<0.05) (Fig. 5C-D).
CXCL12 expression in KCs was detected using the RT-PCR, which showed that high concentrations (10 −6 M and 10 −5 M) of the β2-AR blocker signi cantly down-regulated CXCL12 expression in KCs (P < 0.05) (Fig. 5E). Therefore, M2 KC might promote the directional migration of CT26 cells by upregulating CXCL12 expression.

Discussion
In the present study, we found that low-concentration NE/β2-AR could promote intrahepatic colorectal metastases by inducing the M2 polarization of Kupffer cells (KC) and upregulating the CXCL12 expression on KC. Our ndings are of great signi cance for elucidating the mechanism of colorectal cancer liver metastasis and developing strategies to prevent and treat liver metastasis.
We found for the rst time that in animal models, NE promoted colorectal liver metastases, which was consistent with the literature. Palm et al. reported that tumor progression was signi cantly accelerated after intraperitoneal injection of NE in prostate cancer-bearing mice. [20] Similarly, NE also promoted the progression of colorectal cancer. Han et al. incubated colon cancer cells (HCT-116) with NE at a concentration of 10 −7 M for 24 h and found that NE promoted the proliferation, invasion, and migration of colon cancer cells; and NE promoted colorectal cancer progression by activating the CREB1-miR-373 axis. [21] Several studies have demonstrated that stress-induced activation of the adrenergic system not only promotes the proliferation of cancer cells themselves but also contributes to the metastasis of malignant tumours. [21,22] It has been shown that NE promotes the migration of colon cancer cells by binding to the β2-AR on tumor surface, while propranolol, a β-blocker, prevents the liver metastasis of colon cancer. [23,24] In fact, enteric neurotransmitters are also involved in the remodeling of liver immunity. [25] Recent studies have shown that the microenvironment of colorectal cancer intrahepatic metastasis is rich in immunosuppressive cells (e.g. MRC1+ CCL18+ M2-like macrophages) derived from intrahepatic KC. [26] The present study con rmed that NE not only reduced intrahepatic KC content but also promoted KC M2 polarization, leading to an immunosuppressive microenvironment in the liver, which facilitated the metastatic colonization of cancer cells into the liver. As liver resident macrophages, KCs form the rst-line defense against pathogens passing through the portal circulation, playing a key role in regulating liver immunity. [27] It has been found that tumor cells can form pre-metastatic niches by releasing exosomes to induce the M2 polarization of KC. [28] A meta-analysis showed that the rate of liver metastasis from colorectal cancer was signi cantly lower in rats with cirrhosis. [29] Liver KC in rats with cirrhosis secretes cytokines that induce upregulation of Fas receptor protein expression on colorectal cancer cells andpromote cancer cell binding to FasL on intrahepatic cytotoxic T lymphocytes, thereby killing cancer cells. [30] In addition, KC secretes cytokines (e.g. VEGF) and matrix metalloproteinases, which accelerate tumor cell invasion into the liver and promote tumor proliferation and angiogenesis, enhancing colorectal cancer liver metastasis. [31] However, some studies have also shown that colorectal cancer liver metastasis is associated with NE-induced activation of NK cells in the tumor microenvironment. [32] Therefore, the effect of NE on intrahepatic immune status may be multifaceted.
We further found that NE was induced by β2-AR on the surface of KC to polarize KC towards the M2 type. Interestingly, we noted that the effect of NE on KC polarization was closely related with dose, showing bidirectional regulation. The known M1-type and M2-type KCs are not completely opposing, but rather they can be interconverted. Huan et al. found that the sympathetic nervous system could promote KC activation and maintain the in ammatory microenvironment in the liver by activating α-AR on KCs, ultimately accelerating the progression of hepatocellular carcinoma. [33] In contrast, Loegering et al. found that catecholamines could inhibit the clearance function of KCs via β-AR on KCs. [34] Our results showed that low-concentration NE promoted KC polarization towards the M2 phenotype, whereas highconcentration NE promoted KC conversion to the M1 phenotype, which may be related to the different receptors on which they act. Other studies have shown that low-concentration NE (10 −9 to 10 −7 M) bind mainly to α 2 -AR to exert the pro-in ammatory effects, whereas high-concentration NE (10 −7 to 10 −5 M) bind to β 2 -AR to exert the anti-in ammatory effects. [35] NE exerted its anti-in ammatory effects by inhibiting KC polarization towards pro-in ammatory M1 phenotype via β 2 -AR and by down-regulating the expressions of perforin, granzyme B and IFN-γ [36] , and the NE-induced liver injury was associated with its up-regulation of α 2 -AR in intrahepatic KCs, which caused TNF-α release [37] . In our experiment, we further determined the expressions of β 2 -AR and α 2 -AR on the surface of NE-treated KCs and found that NE-induced expression of β 2 -AR on KCs was signi cantly higher than in the control group, while there was no signi cant change in the expression of α 2 -AR. When a β 2 -AR receptor blocker was added, the markers of M2-type KC were down-regulated, which in turn affected the migration of CT26. Our results showed that NE binded to the β 2 -AR on the surface of KCs and promoted its M2 polarization, ultimately affecting the migration of colon cancer cells. The mechanism may be related to the reduced secretion of TNF-α and NF-κB due to impaired cholesterol metabolism in adipose tissue. [38] The activation of the phosphatidylinositol 3-kinase (PI3K) pathway may be another possible mechanism. It has been reported that the tumor-secreted metastasis-associated secretory protein histone K can stimulate the M2 polarization of TAM via the PI3K/Akt pathway. [39,40] By analyzing the differential genes in colorectal cancer non-liver metastasis and colorectal cancer liver metastasis samples, we found that PI3K/Akt pathway played an important role in colorectal cancer liver metastasis. Our results also revealed that the expressions of PI3K, PAN-Akt and p-Akt proteins on KCs were signi cantly higher in NE induction group than in the control group, while the expressions of these proteins decreased signi cantly after β 2 -AR inhibition. Further inhibition of the PI3K/Akt pathway downregulated the expression of CD206 on KCs, suggesting that NE binds to β 2 -AR on KCs and induces KC M2 polarization by activating the PI3K/Akt pathway. Kim et al. found that fucoxanthin (FCX), a xanthophyll carotenoid, could inhibit LPS-induced macrophage polarization and reduce the secretion of IL-6, TNF-α and IL-1β by activating the PI3K/Akt pathway, thus promoting the in ltration of tumor cells. [41] Wang et al. showed that green tea polyphenols could inhibit the expressions of CXCL12 and VEGF in tumor cells through PI3K/Akt/mTOR/hypoxia inducible factor-1αpathway, thereby resisting tumor angiogenesis. [42] However, the mechanism by which NE affects the directional migration of colorectal cancer cells is still unclear. Our results suggested that KCs treated with low-concentration (10 −9 M) NE signi cantly induced the migration of CT26 cells, suggesting that M2 KC has a role in promoting CT26 migration.
Cancer cell homing has been reported to be in uenced by the binding of chemokine receptors in cancer cells to speci c ligands in target organs. [43] During tumor metastasis, CXCL12 can activate its receptor CXCR4 in a paracrine or autocrine manner to initiate the early-stage -metastasis; when tumor cells enter the circulation, CXCR4-positive tumor cells tend to migrate to target organs with high CXCL12 expression, such as the liver, lungs, and lymph nodes; after these cells settle in the target organs, the CXCL12/CXCR4 axis may provide further support for their growth. [44,45] M2 macrophages or TAMs can secrete chemokines such as CXCL8, CXCL12, and CCL2 to promote the proliferation of lung cancer cells. [46] In addition, M2 macrophages promote CXC ligand expression, which is associated with high in ltration of CRC. CXCL12 can promote distant metastasis of colorectal cancer by upregulating Homeobox B5 (HOXB5), a member of the HOX transcriptional factor family, to activate CXCR4. [47,48] The activation of the CXCL12-CXCR4 axis initiates not only CXCR4-positive tumor cells but also CXCR4-positive bone marrow-derived dendritic cells, which promote angiogenesis in a VEGF-dependent manner and are involved in tumor metastasis. [49,50] In addition, CXCL12 in the tumor microenvironment can promote colon cancer migration by activating the PI3K/Akt pathway via binding to CXCR4 on colon cancer cells; the CXCL12-CXCR4 axis can activate the PI3K/Akt pathway to promote craniopharyngeal carcinoma migration; and CXCL12 signaling can also induce epithelial-mesenchymal transition of glioblasts through the PI3K/Akt pathway to promote glioblastoma migration. [51][52][53] Our results suggested that M2 KCs might induce the directional migration of CT26 cells via CXCL12 and might further activate the PI3K/Akt pathway to induce the M2 polarization of KCs.
There are some limitations in the present study: a) only one cell line (CT26) was used; and b) the ndings were not validated in in vivo experiments.
In conclusion, the effect of NE as an enteric neurotransmitter on colorectal cancer liver metastasis may be related to the polarization of intrahepatic KCs. Low-concentration NE/β 2 -AR may induce intrahepatic KC M2 polarization through the PI3K/Akt pathway and promote its secretion of CXCL12 to induce colorectal cancer liver metastasis.

Declarations
Ethics approval and consent to participate

Consent for publication
Not applicable Availability of data and materials The datasets generated and/or analyzed during the current study are available from the corresponding author.

Competing interests
The authors declare that they have no competing interests.