The liver is the organ to which invasive malignant cells most frequently metastasize for most cancer types, and the mechanisms orchestrating the formation of the hepatic metastatic niche are a permanent field of investigation [1]. About thirty years ago, pioneering studies used different murine models to elucidate the reasons for this organ-specific metastatic process [2]. More recently, new strategies with simple biomaterial- or microfluidics-based co-cultures have tried to summarize this complex microenvironment to better understand the interactions between cancer cells and the stroma at the metastatic site [3].

Circulating tumor cells (CTCs) play a crucial role in this process, and considerable improvements have been made in identifying their molecular features and decreasing their detection levels in peripheral blood for diagnostic purposes [4]. However, this process is also related to the specificities of the liver vasculature system, this organ receives blood from two sources, the hepatic artery and the portal vein, with important immunological consequences [5]. As the main contributor is the portal vein [6], the hepatic sinusoids are directly exposed to CTCs emerging from primary malignant tumors developing in the gastrointestinal tract. The fenestrated vasculature that characterizes the liver [5] also explains why CTCs easily exit the vessels and preferentially colonize this organ [3]. The different steps of this colonization have been remarkably deciphered and illustrated, emphasizing the role of complex interactions between tumor cells and a great diversity of highly specialized resident cells, among which hepatocytes play an important role [6, 7]. However, once CTCs exit the circulation at their metastatic site, they also need to alter their metabolism to adapt to the new environment in the host organ [8], and only a small percentage of them can colonize and form metastatic lesions [9]. In this process, morphological and molecular adaptations play determinant roles, through changes in the mechanical properties of selected CTCs, an increased rat sarcoma virus/mitogen-activated protein kinase/extracellular-signal-regulated kinase (Ras/MAPK/ERK) signal output having been observed to be associated with this event [10].

In addition to its anatomical location and organization, the liver presents a unique metabolic and immunosuppressive environment [5] that, when affected, can promote the development of a premetastatic niche through intercellular communication between primary sites of malignant transformation and the heterogeneous cell populations present in the liver environment [6]. In this field, important breakthroughs identified 1) tumor necrosis factor-alpha (TNF-α) and transforming growth factor-beta (TGF-β) released by activated Kupffer cells (KCs), 2) trans-differentiation of hepatic stellate cells (HSCs) into myofibroblasts producing collagens and matrix metalloproteinases (MMPs), and 3) surface receptors expressed by liver sinusoidal endothelial cells (LSECs), such as intercellular adhesion molecule 1 (ICAM-1), C-type lectin domain family 4 member M (CLEC4M), and vascular cell adhesion molecule 1 (VCAM1), used by CTCs to adhere to the inner vessel layer, enabling tumor cell extravasation [11]. Extracellular vesicles (EVs) are also critical mediators of these complex bi-directional tumor-host cell interactions which are now being deciphered to understand this metastatic organotropism [12].

Among other factors produced by the primary tumor that precondition the liver microenvironment to form niches of activated resident cells promoting tumor expansion, bone marrow-derived immune cells recruited to the liver, including neutrophils, fibrocytes, and myeloid-derived suppressor cells (MDSCs) also appear to play an important role [13]. Studying the diverse signaling mechanisms involved in this “pre-metastatic niche”, released by the primary tumor to the liver, identified TGF-β, interleukin 6 (IL-6), integrin beta-like 1 (ITGBL1), glucose-regulated protein 78 (GRP78), the tissue inhibitor of metalloproteinases (TIMP-1), and a significant list of miRNA signatures transferred via exosomes [14]. TGF-β appears to play a central role in the formation of this pre-metastatic niche, inducing migration of tumor cells toward the liver, eliciting fibrosis, and stimulating epithelial-to-mesenchymal transition (EMT) in hepatic and stellate cells, while dampening immune responses via repression of CD8+ T cells and promoting regulatory T cells (Tregs) and macrophages [6]. The creation of this fibrotic and immune suppressive environment is favored by the transition from a quiescent to an active state in HSCs in the space of Disse, where there is an interplay between T cells and MDSCs [15]. Pro-inflammatory signaling can also contribute to a pro-metastatic environment via secretion of serum amyloid A proteins (SAA) by hepatocytes [16], supporting metastatic seeding and colonization, both at the microvascular and extracellular phases of liver metastasis [17]. SAA appeared to represent a downstream mediator of IL-6 signaling that drove myeloid cell accumulation and fibrosis in the liver [18]. Surgical resection of the primary tumor also frequently results in ischemic injury to hepatocytes, the associated hyperactivation of inflammatory pathways promoting the recurrence of liver metastasis [19]. Finally, hepatic progenitor cells (HPCs) were found to be activated at the periphery and in the interior of liver metastases of colorectal carcinoma, in the form of immature ductular structures, whose role remains largely unknown [20].

Besides cell changes, modifications in the composition of the liver extracellular matrix (ECM) promote cancer cell adhesion and growth. To test whether liver ECM could determine colon CTCs propensity to metastasize to this organ, pioneering studies by Zvibel et al. [21] revealed a specific stimulatory effect of hepatocyte-derived ECM on colon cancer cells, followed by an induction of erb-b2 receptor tyrosine kinase 2 (ERBB2) expression. Fifteen years later, a transcriptional profiling of genes altered by this hepatocyte-derived ECM showed it increased the expression of heparin binding EGF like growth factor (HBEGF) and the colon stem cell marker leucine rich repeat containing G protein-coupled receptor 5 (LGR5 gene) [22]. Moreover, the early phase metastasis of human colon cancer cells in the liver involved another kind of interaction, through desmosomal junctions with hepatocytes [23]. Additional interactions of interest with hepatocytes included integrins, osteopontin, and claudins [7]. Another component of ECM remodeling is represented by lysyl oxidase (LOX), which mediates the cross-linking of collagen and elastin. LOX was found to be involved in the generation of a fibrotic microenvironment supporting metastatic growth [24], and its increased expression was correlated with hepatic metastasis [25]. Increased production of fibronectin, some collagen isoforms, and cell-adhesion molecules also contribute to pre-metastatic niche formation [26]. Adhesion of cancer cells is also promoted by binding α2β1 integrin to type IV collagen and some cadherins containing arginyl-glycyl-aspartic acid (RGD) motifs. This appeared to be associated with aggressive forms of cancer [27]. Finally, in the last decade, much insight has been gained regarding the liver metastatic process. However, important questions remain, such as the parameters governing the dormancy and activation of quiescent disseminated tumor cells [28]. Moreover, there are additional questions regarding the signals that activate them to enable the growth of macro-metastases through the various patterns of the macro-metastasis/organ parenchyma interface (MMPI) [29].

Among the various initial cancer localizations, colorectal cancer initially attracted particular scientific interest [30], as it is the 3rd most common cancer in developed countries and the 2nd leading cause of cancer-related mortality. It is currently estimated that 50–60% of patients diagnosed with this cancer develop liver metastases during their disease course [31]. A decade ago, the need to improve the understanding of the molecular mechanisms involved in the metastasis of colorectal cancer to the liver was emphasized, identifying insulin-like growth factor 1 (IGF-1), hypoxia-inducible factor 1-alpha (HIF1α), vascular endothelial growth factor (VEGF), epidermal growth factor receptor (EGFR), and integrin subunit beta 1 (ITGB1) as interesting potential biomarkers in this process for improving diagnosis for patients [32]. Meanwhile, both cellular (bone marrow-derived cells, HSCs, KCs) and molecular events (ECM remodeling) were investigated [33]. Among the first biomarkers identified, the upregulation of VEGF appeared to promote liver metastasis through a positive feedback loop with H₂S anabolism and cystathionine-β-synthase [34]. HSC activation was found to involve 1) EVs secreted from highly metastatic colorectal cancer cells enriched in miR-181a-5p, 2) liver ECM remodeling [increased expression of alpha-smooth muscle actin (α-SMA) and fibronectin, decreased expression of vitronectin and tenascin C], and 3) secretion of C-C motif chemokine ligand 20 (CCL20), CCL24, and C-X-C motif chemokine ligand 10 (CXCL10) chemokines by activated HSCs [35]. In exosomes derived from the primary colorectal tumor tissue, the expression of angiopoietin-like protein 1 (ANGPTL1), a protein that regulates KC secretion patterns and especially decreases MMP-9 production, was also found to be downregulated [36]. Within proteins involved in ECM remodeling (see also the previous section for a general view of the liver metastatic process), the high level of TIMP-1 at the tumor front was found to be mediated by the release of EVs enriched in this protein by cancer cells, with a cytokine-like function in recipient fibroblasts, inducing ECM remodeling [37]. Besides collagens, which are the most abundant ECM components, the colorectal liver metastatic process involves changes in discoidin domain receptors (DDRs), which are part of the receptor tyrosine kinase family (RTK), cross-talking with several transmembrane receptors and thus playing the role of ECM sensors [38].

Given the very high incidence of colorectal cancer, and the fact that liver metastasis is the primary contributor to the death of patients with this cancer, therapeutic strategies have been directed towards therapeutic targeting of the tumor microenvironment through classification into four main molecular subtypes [39]. A recent review of the genomic profiling of liver metastases versus the primary colorectal tumors pointed to their considerable genetic heterogeneity with the co-existence of different intra-tumor sub-clones. The chromosome regions found to be preferentially altered in the liver metastases encoded genes encoding proteins degrading ECM (MMP-2), involved in cell adhesion [carcinoembryonic antigen-related cell adhesion molecule 7 (CEACAM7)], angiogenesis [angiopoietin 2 (ANGPT2)], cell growth (ERBB2) or cell dissemination [inhibitor of DNA binding 1 (ID1) and BCL2 like 1 (BCL2L1)], while an upregulation of genes from the TGF-β signaling pathway was also observed [40]. Another important discovery was the loss of the colon-specific gene transcription program found in liver metastatic colorectal cancer cells, substituted by a liver-specific one, driven by a reshaped epigenetic landscape where the transcription factors forkhead box A2 (FOXA2) and HNF1 homeobox A (HNF1A) appeared to play an important role [41]. Using a liver-specific transgenic model, Zeng et al. [42] also demonstrated the role of the overexpression of hepatocyte-intrinsic cell cycle-related kinase (CCRK) in liver immune microenvironment reprogramming to promote liver metastasis.

Reprogramming also concerns metabolic aspects as metastatic colon cancer cells are influenced by the specificities of the milieu they colonize. For example, the upregulation of aldolase B, involved in fructose metabolism, does not only represent a carbon source for stimulating tumor cell proliferation; but also a way to activate other pro-growth metabolic and signaling pathways [43]. As the liver is an organ rich in citrate, it is a perfect niche for attracting metastasizing cancer cells, helping them to survive [44]. Another driver of metastatic liver colonization associated with metabolic reprogramming is the increased expression of the red blood cell pyruvate kinase (PKLR), which negatively regulates the glycolytic activity of the major pyruvate kinase isoenzyme PKM2, increasing cellular glutathione levels [45]. Interestingly, expression of mitochondrial pyruvate carrier 1 (MPC1), the protein that transports pyruvate into the mitochondria and links glycolysis and the tricarboxylic (TCA) cycle, is also gradually decreased in liver metastasis relative to the primary tumor and normal tissue [46]. Additionally, several pathways of lipid metabolism are also concerned, such as fatty acids, the downregulation of one member of the protein tyrosine phosphatase family (PTPs), the receptor type O (PTPRO), promoting lipid accumulation and liver metastasis while decreasing overall survival [47]. The mevalonate pathway is also involved, its upregulation being documented in invasive cancers [48]. Finally, lipid metabolism reprogramming was also reported in cancer-associated fibroblasts, promoting liver metastasis through the secretion of CXCL5, TGF-β, MMP-2, and serpin family E member 1 (SERPINE1) [49].

All the factors described above contribute to producing a stromal-enriched microenvironment, initially characterized by leucocyte recruitment where L-selectins play an important role in association with the P-selectins present in platelets and endothelial cells, through the mediation of colorectal cancer cell extravasation [50]. A single-cell analysis of the spatiotemporal immune landscape of 97 samples of colorectal liver metastases revealed that among immunosuppressive immune cells, a population of highly metabolic M2 macrophages was present, characterized by their high expression of mannose receptor C-type 1 (MRC1) and CCL18, sharing the KC signature, suggesting their potential origin [51]. Besides this enrichment in immune suppressive cells, evidence of stem cell characteristics was reported one decade ago in 80% of clinical liver metastases, such as increased expression of the stem cell transcription factor gene Nanog homeobox (NANOG) [52]. Moreover, key processes in cancer stem cell metabolism during liver metastasis were recently reviewed, emphasizing the role played by the Warburg effect/glycolysis and fatty acid oxidation in increasing the production of ATP for energy production [53].

Many different types of cancer show a remarkable preference for the liver for establishing secondary tumors, including pancreatic cancer, which is characterized by a specific chemokine-chemokine receptor pattern where the early expression of C-X3-C motif chemokine receptor 1 (CX3CR1) appears to play a significant role [54]. Moreover, proteomic studies revealed the influence of exosomes released by pancreatic ductal adenocarcinoma cells taken by KCs, then reacting by increased TGF-β signaling, and releasing macrophage migratory inhibitory factor initiating the liver fibrosis pathway [55]. In ovarian cancer, the associated inflammation and subsequent destruction of the protective hyaluronan coat of the peritoneal mesothelium that follows co-optation of epithelial ovarian carcinoma cells with mesothelial cells, was found to involve the internalization of exosomes enriched in CD44, resulting in MMP-9 secretion, ultimately helping cancer cell invasion [56]. Neuroendocrine tumors [57] and prostate cancer, when found in the liver compared to other metastatic sites, present the worst prognosis, a specificity apparently related to phenotypic plasticity. E-cadherin downregulation (EMT) in prostate cancer cells allowed them to escape from the primary tumor mass, while hepatocytes were found to educate tumor cells to re-express E-cadherin to help tumor cells seed in and colonize the liver [58]. The same process, EMT, was mentioned in liver colonization by acute myeloid leukemia cells, a rarer event [59].

Investigations conducted on experimental malignant peritoneal mesothelioma (MPM) have provided insight into the molecular determinants of liver metastatic colonization. Proteomic analyses of the liver in an aggressive model of sarcomatoid MPM (M5-T1), established in the immunocompetent rat strain F344, resulted in identifying a set of proteins showing significant changes between liver tissues from three groups of rats. Cross-comparison of the proteomes from 1-normal rats (G1), 2-adjacent non-tumorous liver from untreated tumor-bearing rats (G2), and 3-curcumin-treated rats without hepatic metastases (G3), led to a list of 179 candidate biomarkers [60], of which 22 corresponded to proteins exclusively located in mitochondria, as shown in Table 1 below. This listing also includes 3 more proteins of interest, additionally located in the endoplasmic reticulum (encoded by Mgst1), cytosol (encoded by Decr1), and vesicles (encoded by Prdx3) (Table 1). For discussion on the 154 other non-mitochondrial proteins exhibiting changes in link with invasiveness, the reader is referred to previous work [60].

The 25 mitochondrial proteins listed above are involved in 10 main biological processes, the most important being lipid metabolism, heme biosynthesis, ion transport, the respiratory chain, protection against oxidative stress, mitochondrial dynamics, and hydrolytic processing of bioactive peptides in the extracellular environment (Figure 1). Five of these proteins were additionally reported as presenting some immune cell specificity for plasmacytoid dendritic cells (IDH3A), T cells (SARDH), and monocytes (NLN, PRDX3, MGST1). Interestingly, five proteins were also potential biomarkers in liver cancer, with their quantitative changes observed in [60] and Table 1 consistent with a favorable prognostic value established for three of them (encoded by Acads, Hmgcs2, and Aldh5a1), and a negative prognostic value was reported for the other two (encoded by Vdac2 and Phb).

Investigations into the molecular determinants of liver metastatic colonization are a growing field of research. This review first provides an update to the main features relevant to this exciting topic. Recent breakthroughs deciphered the role of key molecules involved in intercellular communication between CTCs and the heterogeneous population of liver cells, including HSCs, KCs, and LSECs. The role of exosomes derived from the primary tumor, ECM, and liver metabolic reprogramming was also emphasized. Genomic profiling of liver metastases vs. primary tumors, and single cell analyses of their immune landscape also completed these efforts. Secondly, based on previous cross-comparison of liver proteome changes between MPM-bearing rats and normal rats, the role of 25 mitochondrial proteins in tumor invasiveness, identified from a list of 179 candidate biomarkers, was discussed in link with the published literature. Their involvement in lipid metabolism, heme biosynthesis, electron transport chain, energy production, protection against oxidative stress, and mitochondrial dynamics and transport was particularly emphasized, all representing growing fields of interest. This tool is likely to be a good basis for the future design of innovative therapeutic strategies that could benefit patients facing the harmful consequences of liver invasion by aggressive cancers.