Introduction

Colorectal cancer (CRC) is the third most common cancer and the second leading cause of cancer-related mortality worldwide [1]. Although colonoscopy screening has become popularised, the morbidity and mortality of CRC remain high among men [2]. Early-stage CRC is eligible for curative treatment [3]; however, 25–50% of patients with early-stage disease progress to metastatic disease [4]. The liver is the most frequent site of metastasis in patients with CRC [5]. Blood draining from the gastrointestinal tract enters the liver through the portal vein, which promotes the dissemination of CRC into the liver [6, 7]. Approximately 15–25% of patients with CRC have synchronous liver metastasis (LM) [8, 9], and 18–25% of patients with CRC may eventually develop metachronous LM within 5 years of the initial diagnosis [10]. The 5-year survival rate dramatically declines when the local disease develops into metastasis [11, 12]. Therefore, LM has been used as a prognostic marker for CRC. Despite the development of surgical techniques and targeted therapy, the prognosis of colorectal liver metastasis (CRLM) remains poor [13].

The tumour microenvironment (TME) is composed of cancerous and noncancerous cells, including fibroblasts, endothelial cells and immune cells, as well as noncellular components such as the extracellular matrix (ECM), cytokines, growth factors and extracellular vesicles (EVs) [14, 47, 52, 53]. (1) Microvascular phase: Liver-infiltrating CRC cells that are trapped in sinusoidal vessels are killed via phagocytosis by Kupffer cells (KCs) and natural killer (NK) cell-mediated antitumour cytotoxicity [54, 55]; they may also remain alive by esca** cytotoxic effects and adhering to LSECs [53], which facilitates CRC migration into the space of Disse to avoid immune killing. (2) Extravasation and preangiogenic phase: CRC cells relocate to the space of Disse, thus recruiting stromal cells, including hepatic stellate cells (HSCs) that are responsible for the secretion of fibronectin and collagen to form a framework for neovascularization [56, 57] and portal tract fibroblasts, which generate IL-8 to promote invasion and angiogenesis [58]. (3) Angiogenic phase: After LSECs are activated and co-opted to the tumour–liver interface, activated HSC-derived vascular endothelial growth factor (VEGF) induces the formation of intrametastatic vessels, which appear to be continuous with sinusoidal vessels [59]. Various immunosuppressive cells, such as immunosuppressive regulatory T (Treg) cells, myeloid-derived suppressor cells (MDSCs) and macrophages, are activated to form an immunosuppressive microenvironment, which promotes the development of CRLM. (4) Growth phase: CRC cells acquire adequate blood supply and proliferate rapidly under the ‘protection’ of intrinsic hepatic immune tolerance and the immunosuppressive microenvironment, eventually forming a detectable metastatic tumour in clinical settings [47]. Therefore, the targeting of angiogenesis and the transformation of the immunosuppressive microenvironment into an immune-effective microenvironment are prospective therapeutic strategies for CRLM. Moreover, an understanding of the immune microenvironment of the liver may help to develop effective immunotherapeutic approaches.

Immune landscape of the TME in CRLM

The homeostasis maintained by organ innate resistance in the liver is attributed to various highly specialised resident cells and all types of immune cells [21, 60,61,62]. Each of these cells not only helps to balance protein, lipid and glucose metabolism but also orchestrates immune responses and oncogenesis [63,64,65,66,67,68]. The considerable inflow of antigens shapes the unique immune microenvironment of the liver to harmonise immune activation and immune tolerance [69, 70]. In the early stage of CRLM, the abovementioned cells act as defenders to destroy disseminated cancer cells. Specifically, LSECs arrest cancer cells, whereas KCs phagocytose and release tumour-killing cytokines. Additionally, APCs present antigens to T cells and transform them to effector T cells, which is strengthened by CD4+ T cells. Cytokines released from natural killer T (NKT) cells and M1 macrophages protect against cancer cells. However, when cancer cells escape the immune system, effector T cells are rendered dysfunctional by immune checkpoints, whereas cancer cells migrate into the space of Disse by adhering to LSECs. Treg cells impair the antigen-presenting activity of dendritic cells (DCs). Moreover, HSCs are activated to promote ECM remodelling, and M2 macrophages produce MMPs to regulate this process. Tumour-associated neutrophils (TANs) extrude chromatin fibres and form neutrophil extracellular traps (NETs), which trap CRC cells in the liver and eventually promote their invasive and metastatic capabilities. Furthermore, CRC cells adhere to hepatocytes and induce the release of serum amyloid A1 and A2 (SAA) and insulin-like growth factor-I (IGF-I) from hepatocytes, thereby making the liver a primary target for CRC metastasis (Fig. 2).

Fig. 2
figure 2

A schematic representation of the immune landscape of the TME in CRLM. GZM, granzyme; PRF, perforin; CTL, cytotoxic T lymphocyte; BA, bile acid; ex CD8+ T cell, exhausted CD8+ T cell; NET, neutrophil extracellular trap; CCRK, cell cycle-related kinase; sLewis-x, sialyl Lewis-x; CTLA-4, cytotoxic T lymphocyte antigen-4; SAA, serum amyloid A1 and A2; IGF-I, insulin-like growth factor-I. ① LSECs and NK cells produce IFNγ to upregulate functional Fas and induce apoptosis of cancer cells; PRF and GZM released from NK cells kill cancer cells. ② Disseminated CRC cells are phagocytosed by KCs along with the release of TNF-α, IL-1α and IL-1β. ③ APCs present neoantigens to CD8+ T cells, thus inducing the rapid proliferation of CD8+ T cells and their differentiation into CTLs. ④ CTLs secrete PRF and GZM, as assisted by IFNγ and TNF-α produced by Th1 cells to kill cancer cells. ⑤ LSECs are regulated by gut microbiota-modified bile acids to secret CXCL16, thus recruiting NKT cells to fight cancer cells. ⑥ M1 macrophages directly kill cancer cells by releasing cytotoxic ROS, NO and IL-12. ⑦ The function of cytotoxic CD8+ T cells is impeded due to the interplay between PD-L1 and PD-1. ⑧ The interaction between E-selectin and sialyl Lewis-x promotes the adhesion of CRC cells to LSECs. ⑨ Treg cells bind to APCs via the interaction between CTLA-4 and CD80/86 and produce TGF-β and IL-10 to suppress the activation of CTLs. ⑩ MDSCs, which are recruited by CXCL1 secreted from CRC cells, induce the activation of Treg cells to impair the antigen-presenting activity of DC cells. ⑪ M2 macrophages produce IL-10, TGF-β and MMP to regulate matrix remodelling. ⑫ As induced by TGF-β secreted from KCs, HSCs are transformed to aHSCs and release TGF-β to promote ECM remodelling. ⑬ Lactic acid causes NK cell apoptosis by downregulating their intracellular pH. ⑭ TANs release CCL2 and CCL17 to recruit CCR2+ M2 macrophages and CCR4+ Treg cells. ⑮ As induced by IL-8, NETs trap CRC cells in the liver. ⑯ Hepatocyte-derived CCRK increases CXCL1 production to recruit PMN-MDSCs, thereby impairing NKT cell-mediated immunosurveillance. ⑰ As mediated by integrins and desmosomes, CRC cells adhere to hepatocytes, thus inducing the release of SAA and IGF-I from hepatocytes

Full size image

Interaction of resident liver cells with cancer cells

Liver sinusoidal endothelial cells

LSECs perform important physiological and immunological functions, including filtration, endocytosis and antigen presentation [71,72,73]. As a selective barrier, LSECs allow for the entry of molecules such as plasma proteins, drugs, small chylomicron remnants, exosomes and smaller viruses (< 200 nm) into the space of Disse; however, they do not allow the entry of cells [74,75,76]. The mannose receptor, scavenger receptor and Fc-γ receptor IIb2 efficiently facilitate the clearance and degradation of blood-borne macromolecules by LSECs to perform endocytosis and scavenging functions [77, 78]. Additionally, LSECs exert antigen-presenting functions mediated by the mannose receptor and scavenger receptor, which mainly reshape the immunosuppressive microenvironment in the liver. However, LSECs can dampen effector immune responses. Specifically,, antigen presentation by LSECs mainly induces the differentiation of CD4+ T cells into Treg cells to promote the development of an immune-tolerant TME in the liver [44, 79]. In contrast, LSECs mainly drive a tolerogenic response mediated by an increase in the levels of coinhibitory PD-L1 that interacts with PD-1 to induce CD8+ T-cell dysfunction [80].

LSECs play a dual role in advancing tumorigenesis. When disseminated CRC cells enter the sinusoids, they are entrapped by LSECs and are either destroyed due to mechanical stress, phagocytosed by KCs or killed by perforin (PRF)/granzyme (GZM) from NK cells. LSECs and NK cells release interferon-gamma (IFNγ) and nitric oxide (NO) to upregulate Fas and induce apoptosis of cancer cells via the Fas–FasL pathway [81]. Recent studies have highlighted the fact that LSECs are influenced by gut microbiota-modified bile acids to secrete CXCL16, which recruits NKT cells to fight primary and metastatic liver tumours [82]. However, the anticancer proinflammatory response results in the high expression of vascular adhesion factors such as E-selectin, VCAM-1 and ICAM-1 on LSECs, thus leading to the susceptibility of LSECs to adhesion by cancer cells with the help of sialyl Lewis-x, PSGL-1 and ESL-1 [83,84,85]. Cancer cells can escape from the destruction of the initial assault through counterreceptor communication, after which they migrate into the space of Disse, where they are protected from the cytotoxic effects of KCs and NK cells [86]. A novel adhesion molecule known as LSECtin mediates the communication between activated T cells and LSECs [87, 88] to inhibit the tumour-killing effects of T cells; in addition, it facilitates adhesion and migration of CRC cells to the liver [89]. In multiple experimental LM models, melittin nanoparticles have been demonstrated to induce the activation of LSECs to reverse the hepatic immunological environment to the activated state, which recruits NK and CD8+ T cells and suppresses LM [90].

Due to the fact that LSECs induce a suppressive immune microenvironment in the liver and assist in the growth of disseminated cancer cells, the targeting of LSECs to modulate the hepatic immune microenvironment may be a novel approach to the management of LM in the future. For example, the abundance of beneficial gut organisms that optimise the metabolism and immunity of the liver can be enhanced by modulating the action of LSECs for the effective treatment of CRLM.

Kupffer cells

KCs, which are the resident macrophages in the liver, serve as a crucial part of the innate immune response, which is the first line of defence of the liver [91]. Localised in the hepatic sinusoid, KCs can recognise all types of antigens (such as immune complexes, senescent cells and cancer cells) from the portal or arterial circulation and exert anti-inflammatory effects to prevent the entry of gut-derived substances into the hepatic sinusoid [92, 93]. In the early stage of CRLM, the adherence of disseminated cancer cells to KCs prompts KCs to capture and phagocytose the cancer cells and release TNF-α, interleukin-1α (IL-1α) and IL-1β, thus reducing the metastasis of colon cancer cells to the liver [94, 95]. The innate receptor Dectin-2 on KCs promotes the phagocytosis and elimination of disseminated CRC cells to resist metastasis [96].

Although KCs mainly play a tumoricidal role in the early stages of metastasis, they also play a vital role in hepatic carcinogenesis [97]. KCs activate and expand FOXP3+CD4+ Treg cells through antigen presentation and induce tolerance by upregulating the inhibitory marker PD-L1, thereby resulting in the formation of an immune-tolerant environment to achieve homeostasis [23]. Moreover, HSCs are activated and produce fibronectin induced by KC-derived TGF-β, thus recruiting bone marrow-derived macrophages and neutrophils to form a favourable environment [53]. The corelease of TGF-β, fibronectin, EGF, VEGF and matrix metalloproteinases (MMP-2, MMP-9 and MMP-13) from KCs and HSCs leads to ECM remodelling, angiogenesis and cancer progression [98], which is augmented by the absorption of pancreatic ductal adenocarcinoma (PDAC)-derived exosomes by KCs in a PDAC model [99]. However, CRC-derived exosomal angiopoietin-like protein 1 (ANGPTL1) shuttles to KCs to decrease the expression of MMP-9, which subsequently reduces LM and inhibits vascular leakage mediated via the suppression of the JAK2-STAT3 signalling pathway [100]. In addition, KCs can phagocytose EV-packaged miR-135a-5p, thus mediating immunosuppression and facilitating the development of a premetastatic niche (PMN) in patients with CRLM [101].

It has been reported that a novel immunotherapy strategy by using bacterial genetic modification induces the reprogramming of KCs, which augments the phagocytic ability of cancer cells and strengthens the cytotoxic killing capacity of T cells to suppress LM [102]. An understanding of the functional role of KCs in CRLM may help to identify potential therapeutic targets and to develop novel therapeutic strategies, such as nanoparticle-mediated noncoding RNA-based therapy and bacterial treatment to reprogram the function of KCs. However, further research is required to identify the underlying mechanism and potential for application.

Hepatic stellate cells

As a resident nonparenchymal liver cell population, HSCs contribute to liver fibrosis and cancer development [67, 103]. HSCs maintain homeostasis in the liver by regulating the ECM, immune tolerance and inflammatory responses; additionally, they play a significant role in the colonisation and metastasis of cancer cells [104,105,106].

TGF-β is an important regulator of HSCs in the hepatic microenvironment. Its high expression blocks the initiation of CD4+ Th1 cells and weakens cytotoxic responses, thus facilitating LM and leading to a poor prognosis [107, 108]. TGF-β can induce the transformation of HSCs into a fibroblast-like (spindle-like and spread) phenotype (known as activated HSCs [aHSCs]) to promote ECM remodelling [98]. In addition, aHSCs play a vital role in secondary or primary hepatocellular carcinoma [109,110,111]. aHSCs can lead to hepatic fibrosis and portal hypertension, thus contributing to hepatocarcinogenesis and metastasis [109]. In a previous study, we demonstrated that CRC-derived exosomal miR-181a-5p facilitates CRLM by activating HSCs [112]. In addition, aHSCs engulf disease-associated lymphocytes, including CD8+ T, CD4+ T and NK cells, through cell adhesion [113]. As key cells involved in pro-tumour angiogenesis, HSCs have been demonstrated to upregulate fibroblast activation protein alpha (FAPα) and increase CXCL5 secretion, as regulated by cancer cell-secreted fibroblast growth factor-binding protein 1 (FGFBP1). This mechanism stimulates epithelial–mesenchymal transition (EMT) and induces vessel co-option that results in bevacizumab resistance in CRLM models [114]. Therefore, HSCs play an important role in sha** the immune microenvironment of the liver and in inducing resistance to antiangiogenic therapy. The targeting of HSCs expressing specific molecules (such as FAPα) to modulate the immune microenvironment of CRLM represents a beneficial strategy for strengthening the antitumour effects of immune cells and for effectively overcoming drug resistance.

Hepatocytes

Hepatocytes play an essential role in inducing an immune-tolerant TME, which is required for the implantation of disseminated cancer cells. Hepatocyte-mediated cross-presentation of soluble antigens can induce tolerance of antigen-specific CD8+ T cells [115]. After extravasation, disseminated CRC cells can deeply penetrate into the hepatocyte plate, where they proliferate and form metastatic foci. The adhesion of CRC cells to hepatocytes is considered an essential step in the formation of LM [116], which is mediated by integrins [116] or desmosomes [117]. The strongly expressed integrin subunit αvβ5 mediates cell migration and LM in CRC, and its effects are enhanced by hepatocyte-derived heregulin [118]. Hepatocyte-derived SAA can facilitate the development of LM and is highly expressed in patients with CRC. Mechanistically, hepatocytes promote LM by activating IL-6–STAT3 signalling and inducing SAA overexpression, thereby resha** the hepatic TME to facilitate the formation of a PMN in the liver [119]. Moreover, IGF-I can affect cancer growth and metastasis. The inhibition of IGF-1 released from hepatocytes reduces CRLM in mice [120]. A novel IGF-targeting protein (IGF-Trap) has been demonstrated to markedly block CRLM in experimental models to compensate for the function of the impaired insulin receptor system, thus inducing tumour cell apoptosis and reducing angiogenesis [121].

Altogether, the interplay between resident liver cells and cancer cells contributes to the progression and spread of CRC (Table 1). A better understanding of the communication between CRC cells and the hepatic TME may facilitate the development of new combination therapies for the efficient management of CRLM.

Table 1 Interaction between resident liver cells and cancer cells
Full size table

Immune cells contributing to the hepatic immune microenvironment of CRLM

CD4+ T cells

CD4+ T cells are essential in the defence against tumours because they regulate the activity of CD8+ T cells and influence the outcome of antitumour responses [122]. The classical effector CD4+ T helper 1 (Th1) and T helper 2 (Th2) subsets elicit important antitumour immune responses. Specifically, Th1 cells produce cytokines such as IFNγ and TNF-α, thus leading to cell-mediated killing, whereas Th2 cells secrete IL-4, which assists in the activation of humoral immunity [123]. In addition, CD4+ T cells can differentiate into new subsets, such as Th9 cells, Th17 cells and FOXP3+ Treg cells. Moreover, the role of Th17 cells in cancer is controversial [124, 125]. The low proportion of Th1 cells and high proportion of Th17 cells in liver metastatic tissue indicate a poor prognosis in patients with CRLM [126], which is consistent with the condition of patients with CRC [127]. Given the immunosuppressive activity of CD4+ T cells, we mainly focused on FOXP3+CD4+ Treg cells in this review.

One of the characteristics of LM progression is the high infiltration of FOXP3+ Treg cells [152]. Moreover, exosomal circCCAR1 expressed by cancer cells communicates with CD8+ T cells to impede the degradation of PD1, thus promoting the exhaustion of CD8+ T cells in the liver [153]. In addition, the high expression of MGP in cancer cells from both the primary CRC or LM sites increases intracellular Ca2+ to boost NF-κB phosphorylation, which mediates PD-L1 upregulation in CRC cells, thus promoting CD8+ T-cell dysfunction [154]. Furthermore, the induction of T cells to differentiate into regulatory cells is mediated by IL-10 release from LSECs, which are prone to activating the regulatory pathway of CD4+ T cells to FOXP3+CD4+ Treg cells [145].

Recent studies on the apoptosis of CD8+ T cells have demonstrated that tumour-associated macrophages (TAMs) induce apoptosis of CD8+ T cells and impair cytotoxic functions by reducing the expression of granzyme B and perforin in the liver [155]. The mechanism for this effect involves the fact that activated CD8+ T cells experience apoptotic cell death by the Fas–FasL pathway, as mediated by TAMs within the liver, which induces a decrease in activated T cells and transforms the hepatic immune microenvironment in CRLM [25] (Fig. 3C).

Given that cancer-reactive CTLs play a central role in cancer immunity, it is important to reactivate CD8+ T cells to suppress the progression and metastasis of CRC. It has been reported that hyper-IL-15, IL-15 and the sushi domain of the IL-15 receptor α chain augment the cytotoxic functions of CD8+ T and NK cells, which may be a prospective therapy to reactivate CD8+ T cells and recover their anticancer ability to manage CRLM [156].

Tumour-associated macrophages

As multifunctional APCs, macrophages are critical mediators of tumour immunity [157]. Macrophages present exogenous antigens to T cells through MHC-I and MHC-II aided by costimulatory signals, inhibitory signals or other cytokine signals to regulate T-cell activation [158]. Macrophages that infiltrate malignant tissues are known as TAMs.

With inherent plasticity and polarising characteristics, TAMs are conventionally categorised into two subtypes: M1 and M2 macrophages [159, 160]. M1 macrophages suppress cancer growth by releasing cytotoxic reactive oxygen species (ROS), NO and IL-12, which can directly kill cancer cells [161]. However, M2 macrophages induce the formation of an immunosuppressive TME by secreting cytokines, including IL-10, TGF-β, CCL17 and CCL22 [157, 162]. Due to their poor ability to present cancer antigens, M2 macrophages undermine Th1 adaptive immunity [163]. In addition, M2 macrophages produce MMPs to regulate matrix remodelling, thus facilitating the invasion and metastasis of cancer [164]. In CRC, the expanding liver metastatic tumour is rich in TAMs (primarily M2 macrophages) [157, 171]. Mechanistically, CRC-derived CTHRC1 interacts with the TGF-β receptor in macrophages to activate TGF-β signalling to promote CRLM.

With the development of single-cell profiling, TAMs have been classified as C1QC+ TAMs, SPP1+ TAMs and MRC1+CCL18+ TAMs [151, 165, 172]. In a previous study, single-cell analysis showed that MRC1+CCL18+ macrophages and SPP1+ macrophages are the predominant M2 cell subsets in liver metastatic tissue [151]. Consistently, the presence of SPP1+ macrophages in liver metastatic tissue was reported in a study by Liu et al. Therefore, SPP1+ macrophages may be a potential culprit in CRLM. Moreover, MRC1+CCL18+ macrophages infiltrating liver metastatic tissue exhibit high metabolic activities, thus suggesting that they may promote LM through metabolic pathways [151]. SPP1+ macrophages are found in mesenteric lymph nodes with metastasis but not in mesenteric lymph nodes without metastasis, thus indicating that SPP1+ macrophages play a role in facilitating the expansion of disseminated cancer cells [165]. Furthermore, it has been reported that in microsatellite-stable (MSS) CRC, SPP1+ macrophages and fibroblasts communicate very closely via the ligand‒receptor pathway, which may help to shape an immunosuppressive TME in the liver [173]. However, more studies are required to understand the mechanisms by which different subtypes of macrophages promote LM.

Overall, as the leading tumour-infiltrating immune cells in the TME [174,175,176], TAMs play a critical role in the progression and metastasis of CRC. Their high proportion is closely related to a worse prognosis [165, 177]. Specific subsets of macrophages, including MRC1+CCL18+ and SPP1+ macrophages, may serve as potential therapeutic targets for CRLM.

Myeloid-derived suppressor cells

MDSCs are one of the key contributors to the formation of an immunosuppressive TME in the liver [178]. They mediate immune evasion by inducing the production of Treg cells [179], thus inhibiting NK cell function [180] and impairing the antigen-presenting activity of DCs [181]. In addition, MDSCs facilitate cancer progression and metastasis in a nonimmune manner by producing MMP-9, which is a primary regulator of EMT [182], as well as VEGF, in order to promote TME remodelling and angiogenesis [20]. MDSCs are mainly classified as granulocytic or polymorphonuclear MDSCs (PMN-MDSCs) and monocytic MDSCs (M-MDSCs). The phenotypic and molecular features of these subtypes are difficult to identify [183]. Furthermore, the accumulation of MDSCs is one of the most dominant immunological features of CRC and is associated with disease progression and metastasis [184, 185].

MDSCs facilitate the formation of a PMN and the metastatic colonisation of CRC [186, 187]. Clinically, the high expression of CCL15 in patients with CRC results in the recruitment of more CCR1+ MDSCs, which is associated with the loss of SMAD4 (which is a TGFβ-relevant transcription factor) and promotes CRLM [188, 189]. In an orthotopic mouse model of CRC, CXCR2-expressing MDSCs are recruited from the circulatory system to the liver by CXCL1 secreted from CRC cells in the premetastatic liver, which facilitates the growth of disseminated CRC and its metastasis to the liver [187]. Mechanistically, sphingosine-1-phosphate receptor 1 (S1PR1)–STAT3 signalling in CRC cells results in the production of IL-6 to induce the activation of S1PR1 and p-STAT3 in MDSCs, thus leading to the formation of a PMN in the liver to promote CRLM [190]. Zeng et al. reported that the overexpression of hepatocyte-derived cell cycle-related kinase (CCRK) increases CXCL1 production to recruit PMN-MDSCs, thereby impairing NKT cell-mediated immunosurveillance, which dramatically promotes the metastasis of CRC cells to the liver [191]. M-MDSC-produced CCL7 binds to CCR2 on micrometastatic cells and stimulates the JAK/STAT3 pathway to activate dormant cells, thereby promoting the progression of CRLM [192]. Moreover, the inhibition of CCL7 may represent a potential strategy for preventing recurrent CRLM.

However, it is difficult to target MDSCs because they do not have a specific phenotype that differs from other mature granulocytes. Therefore, further research is required to identify therapeutic targets.

Natural killer cells

Under physiological conditions, NK cells are enriched in the liver and contribute to defending against infection and eliminating cancer cells [69]. After NK cells encounter cancer cells and are activated, they release PRF and GZM, thus leading to osmotic lysis and apoptosis of cancer cells [193]. Additionally, NK cells can directly kill target cells via the expression of TNF-related apoptosis-inducing ligand and FasL [194]. NK cells function in tumour immunosurveillance and elicit inflammatory responses by producing cytokines and chemokines [195].

NK cells can eliminate disseminated cancer cells to control metastasis [196]. A high proportion of NK cells indicates a good prognosis in patients with CRLM [197]. However, in the highly glycolytic environment of CRLMs, lactic acid causes the apoptosis of NK cells by downregulating their intracellular pH [198]. In addition, MDSCs attenuate the immunoreaction of NK cells by releasing NO, which interferes with FcR-mediated functions of NK cells, such as antibody-dependent cellular cytotoxicity (ADCC) and cytokine generation [199]. In a previous study on murine models of CRLM, compared with conventional NK (cNK) cells, liver-resident natural killer (LrNK) cells had a high expression of RORα, which is required to maintain LrNK cells but has no impact on cNK cells. The conditional knockout of Rorα aggravated CRLM, thus indicating that RORα is required for LrNK cell-mediated antitumour immunity. However, the RORα agonist SR1078 restrained CRLM [200]. Clinically, LrNK cells are significantly depleted in CRLM due to the accumulation of tumour-derived lactate, thus resulting in mitochondrial dysfunction and apoptosis of NK cells [198]. The targeting of lactate in the TME may restore the tumour-killing effects of NK cells and benefit patients with CRLM. Altogether, LrNK cells exert a great antitumour impact on CRLM and are closely related to prognosis; therefore, they may be qualified as specific therapeutic targets for CRLM.

Dendritic cells

Dendritic cells are the classical APCs that exert considerable influence in triggering antigen-specific immune responses and inducing immune tolerance [201,202,203,204]. The antigen-presenting function of conventional DCs (cDCs) is important for the antitumour response of effector T cells [205,206,207]. Efficient antigen presentation increases the polarisation of CD4+ Th1 cells and the activation of CD8+ T cells [208, 209].

DCs are heterogeneous and exhibit different characteristics. Compared with plasmacytoid DCs (pDCs), cDCs can more efficiently initiate an immune response against cancer cells [210]. In ICB-treated mouse models of orthotopic pMMR CRLM, the proportion of activated CD8+ T cells, CD4+ T cells and cDCs is lower in metastatic tumours than in subcutaneous tumours [211]. Liver-derived pDCs have a poor capability to stimulate the proliferation of T cells, thus resulting in anergy of effector T cells and immune suppression to maintain inherent liver tolerogenicity [212, 213]. Moreover, liver-resident regulatory DCs differentiated from bone marrow-derived progenitors secrete high levels of IL-10 but low levels of IL-12, thereby inhibiting effective T-cell function to maintain liver tolerance [214]. A subset of cDCs in CRLM identified as DC3s induces a proinflammatory phenotype and is correlated with a poor prognosis [165]. DC3s may be considered as a promising target for improving the therapeutic outcome of immunotherapy in CRLM. Further investigations are required to elucidate the mechanism by which DC3s promote CRLM.

Tumour-associated neutrophils

Similar to TAMs, TANs play a dual role in cancer progression by both promoting and inhibiting the growth and metastasis of cancer [215]. Specifically, TANs facilitate activated T-cell immune reactions by presenting antigens and releasing IL-18 to induce the activation of NK cells [216]. In contrast, TANs release CCL2 and CCL17 to recruit CCR2+ M2 macrophages and CCR4+ Treg cells, which shape a suppressive TME in the liver, thus promoting the progression and metastasis of cancer [217]. Additionally, TANs produce MMP-9 and neutrophil elastase to promote the extravasation of cancer cells and drive disseminated cancer cells to metastasise [218].

The accumulation of TANs has been demonstrated to be necessary for the formation of an omental PMN in orthotopic ovarian cancer models [219]. Ovarian cancer induces neutrophils to form NETs, which trap ovarian cancer cells and facilitate their implantation on the omentum [219]. Therefore, NETs can promote the metastasis of ovarian cancer. Several in vivo and in vitro studies on CRC have reported that the formation of NETs is enhanced by cancer-derived IL-8. These NETs can trap CRC cells in the liver and promote their invasive and metastatic capabilities [220,221,222]. In addition, anterior gradient-2 (AGR2) released from TANs can promote metastasis in murine models of CRLM. TAN–CRC cell crosstalk between TAN-derived AGR2 and CRC-derived TGF-β1 is considered the primary driver of CRLM [223]. Collectively, TAN is an effective potential target for the treatment of CRLM. However, further investigations are required to explore and develop TAN-based therapeutic strategies for CRLM.

In conclusion, immune cells involved in CRLM shape the susceptible suppressive immune microenvironment for tumour invasion and metastasis in CRC (Table 2). Immunotherapeutic strategies that can reverse the immunosuppressive microenvironment or strengthen effector immunity may be effective against CRLM.

Table 2 Immune cells involved in the liver immune microenvironment of CRLM
Full size table

Extracellular vesicles in the immune microenvironment of CRLM

EVs refer to various nanosized vesicles with membrane structures released by cells [224, 225]. According to their diameter and the mechanisms of biogenesis, they are classified into three subgroups (exosomes, microvesicles and apoptotic bodies). Exosomes have attracted substantial interest and have been widely investigated in recent years [226]. EVs carry bioactive molecules such as nucleic acids, proteins and lipids for intercellular delivery and facilitate intercellular communication [227, 228]. They are an important aspect of the immune microenvironment. CRC-derived EVs in the immune microenvironment facilitate the relocation and aggression of CRC, which contributes to LM [229,230,231,232]. In previous studies, we elucidated the molecular mechanisms underlying the involvement of EVs in the formation of a hepatic PMN and the metastasis of CRC, and we also identified promising functional biomarkers for CRC [35, 112, 233, 234], thus indicating that EVs may serve as a therapeutic target and a prognostic and diagnostic biomarker for CRLM (Fig. 4).

Fig. 4
figure 4

Schematic diagram depicting extracellular vesicles in the immune microenvironment of CRLM. Various cells in the hepatic immune microenvironment interact with CRC cells via extracellular vesicles to form a sophisticated immunosuppressive microenvironment that contributes to CRLM. The different pathways are indicated by different coloured arrows. ① CRC-derived hypoxia-induced exosomal miR-135a-5p is phagocytosed by KCs, thus blocking CD30-mediated CD4+ T-cell activation and promoting cell adhesion. ② Highly mCRC cells produce EV-packaged miR-181a-5p that activates HSCs. aHSCs release CCL20, which interacts with CCR6 expressed on CRC cells and activates CRC cells to promote the release of EV-packaged miR-181a-5p, thus contributing to resha** the hepatic TME and forming a PMN; CRC-derived exosomal HSPC111 promotes the activation of HSCs, thus leading to the upregulation of CXCL5, which targets CRC-expressed CXCR2, increases the secretion of exosomal HSPC111 from CRC cells and promotes CRLM. ③ M2 macrophages release exosomal miR-21-5p and miR-155-5p, after which they shuttle into CRC cells, which contributes to the migration and invasion of CRC. ④ Exosomal miR-934 secreted from CRC cells induces M2 macrophage polarisation to promote CRLM. M2 macrophages release CXCL13, which interacts with CXCR5 in CRC cells and promotes the transcription of miR-934. ⑤ CRC cells secrete exosomal miR-25-3p to stimulate endothelial cells, thus leading to vascular leakage and vasculogenesis

Full size image

Several studies have validated the pivotal role of EVs in the formation of a hepatic PMN. After being phagocytosed by KCs, CRC-derived EV-packaged miR-135a-5p can inhibit CD30-induced activation of CD4+ T cells and promote cell adhesion, which facilitates the development of a PMN for CRLM [101]. In a previous study, we reported that EV-packaged miR-181a-5p secreted by CRC cells activates HSCs. The interaction between CCL20 released from aHSCs and CCR6 expressed on CRC cells activates CRC cells to promote the release of exosomal miR-181a-5p, which generates a positive feedback loop to reshape the hepatic TME and form a PMN [112]. In mouse models of CRLM, endothelial cells stimulated by CRC-derived exosomal miR-25-3p can lead to vascular leakage and vasculogenesis, thus contributing to the formation of a PMN and enhancing CRLM [235].

EVs play an important role in intercellular communication, such as between CRC cells and other cells, in CRLM. M2 macrophage-derived exosomes can deliver miR-21-5p and miR-155-5p to CRC cells. Additionally, miR-21-5p and miR-155-5p are internalised by CRC cells and targeted to the BRG1 coding sequence, thus leading to a decrease in the expression of BRG1 in CRC cells and contributing to the migration and invasion of CRC [236]. In a previous study, we demonstrated that exosomal miR-934 in the immune microenvironment induces M2 macrophage polarisation to promote CRLM. Additionally, CXCL13 released by M2 macrophages interacts with CXCR5 on CRC cells to promote the transcription of miR-934 [35]. CRC-derived exosomal HSPC111 can promote the activation of HSCs, thus leading to the upregulation of CXCL5, which targets CXCR2 expressed on CRC cells, increases the release of exosomal HSPC111 from CRC cells and promotes CRLM [237].

Altogether, as coordinators of intercellular communication in the dynamic network of the TME, EVs are responsible for the progression and metastasis of cancer [238] and can serve as noninvasive markers for the screening and management of CRLM [239, 240]. A better understanding of the regulatory mechanisms of EVs can help to generate antitumour responses and design efficient EV-based diagnostic and therapeutic strategies for CRLM.

Therapies for CRLM

Surgical resection

At present, surgical resection is an effective therapeutic option for resectable CRLM [241]. The two commonly used surgical strategies include simultaneous resection and delayed resection. The results of existing studies on the selection of simultaneous or delayed resection are inconsistent [242,243,244]. The 5-year survival rate of patients with CRLM after resection can be improved to 50% [241]. However, only 10–30% of patients with localised LM are eligible for resection after diagnosis [245]. Moreover, 52% of patients develop postoperative recurrence of CRLM [246], thus resulting in a high mortality rate. Therefore, there is a need to explore novel therapeutic modalities for CRLM.

Systemic and conversion therapies

Systemic therapy is a more favourable treatment option for nonresectable CRLM. In addition to improving the quality of life and prolonging survival, effective systemic therapy can transform unresectable lesions into resectable lesions, which is known as conversion therapy [247]. According to the guidelines recommended by the National Comprehensive Cancer Network (NCCN) [248], first-line chemotherapy regimens for patients eligible for intensive therapy are FOLFOX (5-fluorouracil combined with leucovorin plus oxaliplatin), CAPEOX (combination of capecitabine and oxaliplatin), FOLFIRI (5-fluorouracil plus leucovorin and irinotecan) and injectable 5-fluorouracil/leucovorin or capecitabine. The use of FOLFOXIRI (5-fluorouracil combined with leucovorin, oxaliplatin and irinotecan) in conversion therapy may maximise tumour shrinkage and improve the eventual outcomes of surgery in patients with potentially resectable CRLM. Effective conversion therapy can allow for 12.5% of patients with unresectable CRLM to undergo liver resection, thus resulting in improved survival rates [249]. However, several adverse effects are associated with this regimen, which should be carefully considered.

Chemotherapy combined with targeted therapy can yield a better outcome for patients who are tolerant to aggressive therapy [250]. Drugs targeting epithelial growth factor receptor (EGFR) and VEGF are commonly used in combination with chemotherapeutic drugs. A phase III trial demonstrated that cetuximab and panitumumab (which are monoclonal antibodies against EGFR) can suppress the downstream signalling pathways of EGFR to effectively inhibit disease progression and provide clinical benefits to patients with mCRC [251]. However, cetuximab and panitumumab are only indicated in patients with wild-type RAS/BRAF [248]. Bevacizumab targets VEGF and plays a significant role in antiangiogenesis [5).

Table 5 Clinical trials targeting the immune landscape for the treatment of metastatic colorectal cancer
Full size table

Various vaccine strategies have been designed, including whole tumour cell-based, protein- or peptide-based and DC-based vaccines [297,298,299]. In a randomised clinical trial, patients with colon cancer who received an autologous cancer cell-based vaccine had a significantly longer recurrence-free interval (p = 0.011) and recurrence-free survival (p = 0.032); however, disease-specific survival and OS showed no improvement [300]. Based on this foundation, further clinical studies (Clinicaltrials.gov Identifier: NCT00016133, NCT02448173) have been conducted to examine the protective effects of the autologous cancer vaccine against tumour recurrence after colon cancer surgery [301]. DC-based vaccines are powerful contributors to antigen presentation and the initiation of antitumour immunity [302,303,304]; however, the development of DC-based vaccines for CRC is currently at an early stage. In a preclinical murine model, effective immunotherapy using tumour-associated antigen-loaded cDC-based vaccines increased the infiltration of activated effector T cells and inhibited tumour growth [305]. Two related clinical trials (Clinicaltrials.gov Identifier: NCT03730948, NCT02919644) on DC-based vaccines are ongoing to develop strategies for preventing the progression of surgically resected stage I and II hypermutated CRC or curatively resected stage IV CRC. Moreover, peptide-based vaccines have strong specificity and can easily elicit an effective immune response [306, 307]. In a phase I clinical trial, the adenovirus (Ad5)-GUCY2C-PADRE vaccine was efficient in patients with CRC and did not cause grade-3/4 toxic events during the 6-month follow-up period after vaccination [308]. Another phase I study demonstrated that the combination of a single-dose PolyPEPI1018 vaccine and maintenance therapy with fluoropyrimidine and bevacizumab was strongly effective, with 96% of vaccine peptides inducing T-cell responses without causing grade 3 or higher adverse events in patients with mCRC [309].

Based on these encouraging results, cancer vaccines may represent an immunotherapeutic strategy that is not limited by the DNA MSI or MMR status. Cancer vaccines can stimulate immune surveillance to combat initially undetected microscopic lesions and consequently enhance the survival of patients with CRLM.

There are multiple therapeutic options for CRLM; however, they fail to meet the requirements for a disease-free prognosis (Fig. 5). Immunotherapy is an emerging and effective weapon in the fight against CRLM that requires further research to explore its superior potential value.

Fig. 5
figure 5

Therapies for CRLM

Full size image

Discussion and perspectives

Liver metastasis is the most common site of metastasis in CRC and is the leading cause of death in patients with CRC. The liver is a characteristic immune-tolerant organ in which resident liver cells, recruited inflammatory and immune cells and active protein molecules interact with each other; additionally, EVs act as important mediators of intercellular communication. The intricate characteristics and mechanisms of the hepatic immune microenvironment that directly or indirectly induce immunosuppression and contribute to the regulation of cancer metastasis should be extensively investigated to explore potential therapeutic targets for CRLM.

The formation of a PMN in CRC is an important prerequisite for LM, which is a progressive process that triggers local changes such as vascular leakage, ECM remodelling and systemic effects on the immune system. Induced by the combined systemic effects of CRC-secreted factors and EVs, the PMN shapes a microenvironment that is favourable for LM, which makes the distant liver a favourable site for the colonisation of disseminated cancer cells. However, the underlying mechanisms responsible for the formation of the PMN remain uncertain and warrant further investigations.

ICB-based treatment is only effective in patients with MSI-H/dMMR CRC but not in patients with MSS/pMMR CRC. Moreover, another challenge involves the potentially deleterious side effect known as hyperprogression that occurs in some patients after ICB therapy and is independently associated with advanced age and higher metastatic load. With the increasing use of ICB therapy in clinical practice, more studies are required to elucidate the potential mechanisms and to identify the predictors of hyperprogression, which would allow for patients at high risk for life-threatening immune-related adverse events to be screened before ICB therapy. As emerging immunotherapeutic strategies, CAR-T-cell therapy and cancer vaccines may revolutionise the era of cancer immunotherapy. Despite the excellent efficacy of CAR-T-cell therapy in haematological malignancies, its use in solid tumours may be limited due to the trafficking barriers of CAR-T cells, their weak ability to infiltrate solid tumours and their off-target effects. Furthermore, tumour vaccines are a type of individualised immunotherapy with high specificity and few side effects. However, the effective translation of cancer vaccines in clinical practice remains challenging. Further preclinical and clinical trials should be conducted to demonstrate the efficacy of immunotherapy in mCRC and to identify novel therapeutic targets or to develop combination strategies to improve or activate antitumour immune responses for the effective treatment of CRLM.