Regulation and functional roles of chemokines in liver diseases

Abstract

Inflammation is a major contributor to the pathogenesis of almost all liver diseases. Low-molecular-weight proteins called chemokines are the main drivers of liver infiltration by immune cells such as macrophages, neutrophils and others during an inflammatory response. During the past 25 years, tremendous progress has been made in understanding the regulation and functions of chemokines in the liver. This Review summarizes three main aspects of the latest advances in the study of chemokine function in liver diseases. First, we provide an overview of chemokine biology, with a particular focus on the genetic and epigenetic regulation of chemokine transcription as well as on the cell type-specific production of chemokines by liver cells and liver-associated immune cells. Second, we highlight the functional roles of chemokines in liver homeostasis and their involvement in progression to disease in both human and animal models. Third, we discuss the therapeutic opportunities targeting chemokine production and signalling in the treatment of liver diseases, such as alcohol-associated liver disease and nonalcoholic steatohepatitis, including the relevant preclinical studies and ongoing clinical trials.

Chemokines are small, basic, heparin-binding and chemotactic cytokines that were discovered in the late 1970s and early 1980s^1–3. In 1999, a systematic nomenclature was established and they were divided into four families: CXCLs, CCLs, CX₃CLs and XCLs⁴. Chemokines are a family of more than 50 small cytokine-like proteins that chemoattract and activate immune cells during inflammation and in diseases such as cancer. In human liver diseases and experimental animal liver injury models, chemokines can be produced from different cell types in response to various external stimuli. These small molecules have been well known to regulate immune cell infiltration and modulate the activation and proliferation of a variety of liver cells, including hepatocytes, hepatic stellate cells (HSCs) and endothelial cells⁵. Through interactions with 19 chemokine receptors expressed on both immune cells and liver-resident cells, chemokines have been shown to have crucial roles in a wide range of biological processes both in health and disease⁶. With the advances in functional genomics, the transcriptional regulation of chemokines has been enthusiastically studied. These studies might lead to new therapeutic targets for acute and chronic liver diseases such as alcoholic hepatitis, cirrhosis and hepatocellular carcinoma (HCC).

Chemokines interact through cell surface-specific receptors⁷, signalling to cells and changing cell behaviour such as cell migration. In addition, chemokines can lead to signalling pathway activation by binding to glycosaminoglycans (GAGs) on endothelial cells, epithelial cells and the extracellular matrix. GAG interactions can result in the generation of soluble chemotactic and tissue-bound hepatic gradients for neutrophils or monocytes^8,9.

In this Review, we discuss the function and regulation of chemokines as well as their associated signalling pathways. In particular, we address the genetic and epigenetic regulation of chemokine transcription and the cell-type specificity of chemokine production involving liver cells such as hepatocytes, liver sinusoidal endothelial cells (LSECs), HSCs and liver-associated immune cells (mainly macrophages (Kupffer cells), neutrophils, natural killer (NK) cells and natural killer T (NKT) cells) in healthy and disease conditions. The multiple functional roles of chemokines in homeostasis, in animal disease models (particularly mouse experimental models, including acute and chronic liver injury) and in human liver diseases are discussed in detail. Finally, several novel therapeutics targeting chemokine signalling have shown promising results in preclinical studies of animal liver disease models and in clinical trials. We highlight some of these agents, with an emphasis on anti-inflammatory agents and epigenetic-targeting drugs.

Chemokine biology and liver-specific chemokines

The chemokine superfamily is a group of 8–10 kDa, positively charged, secreted proteins with 20–50% sequence homology, which is reflected in their shared structural characteristics^10,11. All chemokines have a highly conserved, three-stranded β-sheet or α-helix tertiary structural fold^12,13. Their quaternary structures can vary. The sequential positions of the first two cysteines divide the chemokines into four families: CC-chemokines, CXC-chemokines, C-chemokines and CX₃C-chemokines. All chemokine families have genes encoding 50 chemokine ligands in humans and there are similar protein structures but different numbers of chemokines in mice. The connections between original chemokine names and modern classification are described in TABLE 1.

Table 1 |.

List of chemokines and their receptors in the liver

Standard name of chemokine ligand	Alternate name	Associated receptor	Key immunoregulatory functions
CXC
CXCL1	Human: GROa, MGSA Mouse: KC	CXCR2	Neutrophil trafficking
CXCL2	Human: GROb, MIP2α Mouse: MIP2	CXCR2	Neutrophil trafficking and transendothelial migration
CXCL5	Human: ENA78 Mouse: LIX	CXCR2	Neutrophil trafficking
CXCL6	GCP2	CXCR1 and CXCR2	Neutrophil–monocyte trafficking
CXCL8	IL-8	CXCR1 and CXCR2	Neutrophil–monocyte trafficking
CXCL9	MIG	CXCR3	TH1 immune response
CXCL10	IP-10	CXCR3	TH1 immune response
CXCL11	ITAC	CXCR3	TH1 immune response
CXCL12	SDF1	CXCR4 and CXCR7	Hepatic stellate cells, liver sinusoidal endothelial cells, hepatocellular carcinomas
CXCL13	BLC, BCA1	CXCR5	B cells
CXCL16	SRPSOX	CXCR6	Natural killer T cells
CC
CCL1	I-309	CCR8	TH2 immune response
CCL2	MCP1	CCR2	Monocytes–macrophages
CCL3	MIP1α	CCR1 and CCR5	T cells and natural killer cells
CCL4	MIP1β	CCR1 and CCR5	T cells and natural killer cells
CCL5	RANTES	CCR1 and CCR5	T cells and natural killer cells
CCL7	MCP3	CCR1, CCR2 and CCR3	TH2 immune response
CCL8	MCP2	CCR2 and CCR3	TH2 immune response
CCL17	TARC	CCR4	TH2 immune response
CCL19	ELC	CCR7	T cells
CCL20	MIP3α	CCR6	TH17 cell response
CCL21	SLC	CCR7	T cells and dendritic cells
CCL22	MDC	CCR4	T regulatory cells
CCL25	TECK	CCR9	T cells and monocytes

T_H, T helper.

Chemokine receptors are classic G-protein-coupled transmembrane proteins with seven domains and are embedded in the lipid bilayer of the cell surface¹⁴. These chemokine receptors are divided into four families that are named after their ligands (TABLE 1): CCR1–10 (R for receptor) bind to CC chemokines, CXCR1–7 bind to CXC chemokines, and so on. The selectivity of different chemokines to the receptors is largely determined by the ligands. Receptor expression also varies in different cell types (Supplementary Box 1). For instance, while pro-inflammatory CXCR2 is the most common receptor on neutrophils¹⁵, its ligands are primarily CXCL1 and CXCL2 in liver. The anti-fibrotic and anti-angiogenic CXCR3 is the key receptor on T cells, NK cells and possibly HSCs, with CXCL9 and CXCL10 as its ligands in the liver^16,17. Another important receptor, CCR2, is distributed mostly on inflammatory monocytes; CCL2, known to be pro-inflammatory and pro-fibrotic, is the ligand^18,19.

The advancement of genetic and epigenetic technologies in the past decade has opened up new possibilities for liver disease research. Here, we focus on the genetic-mediated and epigenetic-mediated regulation of chemokine ligands and receptors and on the cell-type specificity of chemokine production in liver homeostasis and inflammation.

Regulation of chemokine ligands and receptors

Genetic regulation.

The discovery of allelic variants of HIV co-receptors (such as CCR5 and CXCR4) that protect against virus entry into usually permissive cells led to increased interest in chemokine polymorphisms^20–24. Since then, investigations have searched for associations between allelic variants and phenotypes in both sterile and septic inflammatory diseases. One of the most studied is the CCL2-2518-A/G polymorphism, which has been shown to be a risk factor for Alzheimer disease^25,26. CCL2 has also been shown to be highly upregulated in patients with hepatitis C virus (HCV) infection and fibrosis with these single-nucleotide polymorphisms (SNPs). Other SNPs in CCR clusters have been studied in 465 patients with HCV infection and in 370 matched healthy individuals as controls²⁷. This study examined six SNPs and a 32 bp deletion in the genes encoding CCR3, CCR2 and CCR5 (all of which are located in a cluster on chromosome 3), including the G190A polymorphism (variant allele Ile64) in the first transmembrane domain of CCR2. However, the researchers found that host genetic factors in the CCR cluster only had a modest influence on the natural course of the infection, stage of fibrosis or response to therapy. Another example is CXCL9–11: a strong relationship was found between CXCL9–11 polymorphisms (specifically, CXCL9 rs10336, CXCL10 rs3921 and CXCL11 rs4619915) and liver fibrosis in patients with HCV infection²⁸. The three CXCL9–11 SNPs are located at the 3′ untranslated region (UTR) of their respective genes and are targeted by microRNAs. The role of microRNAs on chemokine production has been reviewed previously²⁹. It is thought that the combination of CXCL9–11 polymorphisms and other biomarkers (plasmatic, genetic and so on) could improve the prediction of the development of advanced liver fibrosis. Additionally, most of the SNPs for these chemokines are located in non-coding regulatory regions, such as promoters and enhancers, suggesting that transcriptional regulation is indeed critical for the functional role of chemokines.

Epigenetic regulation.

Given the importance of chemokines in biological homeostasis and disease, the regulation of their expression has been a focus of liver research in the past decade. Chemokines are closely regulated in a dose-dependent and time-dependent manner³⁰. Some of the upstream regulators are the inflammatory cytokines tumour necrosis factor (TNF), IL-1, interferon-γ (IFNγ), IL-17 and IL-6 (REF.³¹). TNF stimulation, for both short and long periods, can activate nuclear factor-κB (NF-κB)³². In addition to NF-κB, other transcription factors, such as activator protein 1 (AP-1), signal transducer and activator of transcription 1 (STAT1) and ELK1, are also involved in stimulating the transcription of chemokines induced by the aforementioned inflammatory cytokines^33,34.

Enhancers, a type of cis-regulatory element, are short DNA sequences that control gene expression by recruiting sequence-specific transcription factors and co-activators^35–37. Data show that enhancers have a crucial role in determining the cell type-specific transcriptome³⁸. In the past decade, the rapid application of next-generation sequencing has provided great insight into the regulatory roles of epigenetic modifications such as DNA methylation, histone modifications, non-coding RNA processes and chromosome conformation changes. Advances in epigenetics have fuelled the accelerated discovery of distal enhancers and revealed the importance of enhancers in the regulation of cell type-specific gene expression^39,40. For example, not only do the post-transcriptional modifications (such as methylation, acetylation and ubiquitination) of histone tails correlate with the activated or repressed chromatin state of associated genes, they can also identify the presence of enhancers. The acetylation of histone H3 at lysine 27 (H3K27ac) and trimethylation of histone H3 at lysine 27 (H3K27me3) are commonly associated with gene enhancer and promoter regions⁴¹. Meanwhile, many studies have demonstrated the importance of super-enhancers, or large enhancer clusters, for the cell-specific regulation of genes. The expression of genes associated with super-enhancers is sensitive to external stimuli, including inflammatory cytokines such as TNF. TNF induces NF-κB signalling through the activation of cis-regulatory elements, such as proximal promoters and distal super-enhancers, in a context-dependent manner to activate gene expression^42,43. NF-κB binding is crucial for the dynamic remodelling of the super-enhancer epigenetic landscape and chromatin structure, which enables rapid changes to clusters of gene expression. It has been well documented in human endothelial cells, such as human umbilical vein endothelial cells (HUVECs) and lung fibroblasts, that TNF-responsive super-enhancers are present in many chemokines, intronic or intergenic regions and support a very high occupancy of NF-κB, activation histone marks (such as H3K27ac and H3K4me1) and the epigenetic reader protein bromodomain-containing protein 4 (BRD4)³².

The development of chromosome conformation capture (3C)-based technologies, circularized chromosome conformation capture (4C) and other chromatin interaction techniques, such as chromatin interaction analysis by paired-end tag sequencing (ChIA-PET) and Hi-ChIP (which can map factor-directed chromatin conformation)⁴⁴, have enabled the discovery of the relevance of the 3D genome organization for the transcriptional control of inflammatory chemokines^32,41. Numerous studies using these new assays have suggested a role for chromatin loops (such as those formed by promoter–enhancer interactions) in chemokine expression programmes⁴⁴. A genome-wide 3C analysis method (Hi-C) has also determined that the genome is organized into various topologically associated domains⁴⁵ that have important roles in genome organization and control of gene expression, including genes encoding chemokines in HUVECs and lung fibroblasts^32,41. It is well established that alcoholic hepatitis is associated with hepatic neutrophil infiltration, partly through the activation of cytokine pathways such as TNF signalling, leading to the increased expression of chemokines^31,46. Data have demonstrated the role of super-enhancers in linking TNF expression with the amplification of chemokine expression in patients with alcoholic hepatitis (published in an abstract⁴⁷). The integration of RNA sequencing (RNA-seq) and histone modification chromatin immunoprecipitation followed by sequencing (ChIP–seq) of human liver explants showed the upregulation of multiple CXCL chemokines in alcoholic hepatitis livers compared with healthy livers along with dramatic histone mark changes. Published databases have identified LSECs as a dominant source of CXCL production in mice and humans⁴⁸. By using ChIP–seq and 4C-seq⁴⁹ and analysing published databases, a putative super-enhancer for multiple CXCLs located 20 kb upstream from the CXCL gene loci was identified⁴⁷ (Supplementary Fig. 1a). Owing to their broad activity against a large number of inflammatory genes and their specificity for their target genes, super-enhancers make attractive candidates for pharmacological intervention⁵⁰.

Additional studies^44,51 using ChIA-PET and Hi-ChIP technologies have found that super-enhancers that regulate inflammatory genes, such as those encoding CXCLs, IL-6 or macrophage colony-stimulating factor 1 (CSF1), can be modulated by long non-coding RNAs (lncRNAs) and histone demethylation (Supplementary Fig. 1b). For example, Fanucchi et al. found that a subset of lncRNAs called immune gene-priming lncRNAs (IP-lncRNAs or IPLs) can act in cis in human cells to direct the WD repeat-containing protein 5 (WDR5)–mixed lineage leukaemia protein 1 (MLL1) complex across the chemokine promoters (IL-8, CXCL1, CXCL2 and CXCL3), facilitating their H3K4me3 epigenetic priming⁴⁴. This mechanism was also shared among dozens of regulators of trained immunity (for example, IL-6 and CSF1), which refers to the ability of innate immune cells to acquire a non-specific enhanced response to pathogen reinfection. The researchers demonstrated that β-glucan, a classic inducer of trained immunity, can increase the transcription of several IPLs. As mentioned in Supplementary Boxes 2, 3, the mouse CXCL gene locus is different to that of humans (Supplementary Fig. 1a) — the mouse locus lacks IPLs in the chemokine super-enhancers, so trained immune responses are absent at the CXCL locus. However, when an IPL was inserted into the mouse chemokine super-enhancer, it resulted in the training of CXCL genes. This study provides strong evidence that IPL-mediated changes to the epigenetic status of immune gene promoters are important for trained immune responses. Meanwhile, a 2020 study⁴³ proposed a model in which lysine demethylases 7A (KDM7A) and 6A (also known as UTX) have vital roles in TNF signalling, where they are regulated by a novel TNF-responsive microRNA — miR-3679-5p. This work highlighted the role of histone methylation (such as H3K9me2 and H3K27me3) in regulating super-enhancer activity. The researchers showed that TNF rapidly induced the co-occupancy of KDM7A and UTX at NF-κB binding sites in HUVECs. KDM7A and UTX were then found to demethylate H3K9me2 and H3K27me3, respectively. H3K9me2 and H3K27me3 are histone-repressing marks required for the activation of NF-κB-dependent inflammatory genes, such as those encoding CXCL1 and CXCL8, and of other important genes such as those encoding vascular cell adhesion molecule 1 (VCAM1) and E-selectin. In addition, this study identified that increased interactions between TNF-induced super-enhancers and NF-κB-responsive target loci coincided with the recruitment of KDM7A and UTX by chromosome conformation capture-based methods. The pharmacological inhibition of KDM7A and UTX significantly reduced leukocyte adhesion in mice. The investigators concluded that the rapid reduction of repressive histone marks is essential for the NF-κB-dependent regulation of super-enhancers that control inflammatory gene expressions.

Chemokines and receptors in liver cells

The liver has a key role in the immune surveillance of enteric pathogens. However, the enteric circulation also carries large amounts of innocuous molecules, such as food antigens, to the liver, which requires the liver to have immunotolerant properties. Thus, a balance is needed between immunotolerance in health and mounting a rapid immunological response to enteric pathogens in disease; upsetting this balance can lead to allergic conditions, sterile inflammation, organ injury, chronic infections and carcinogenesis⁵². Chemokines have an important role in tissue homeostasis and pro-inflammatory processes in many organs, including the liver. Chemokines also contribute to immune surveillance — they direct immune cells via specific chemokine receptors into the liver⁶. Many factors can drive chemokines away from maintaining homeostasis and towards becoming more pro-inflammatory in different liver injuries. These injuries include viral infections, pathogen invasion, drug toxicity, metabolic dysfunctions, and acute or chronic alcohol intake. The functional roles of chemokines change during this process; it is surely context-dependent and continues to be the focus of much research. In general, many liver-resident cells as well as infiltrating cells can secrete chemokines in response to such injuries and can release cytokines, such as TNF, IL-6 and IL-1β, in addition to lipopolysaccharide⁶. These cytokines are common upstream regulators that can stimulate a huge production of chemokines that provoke immune cell infiltration into the liver and trigger inflammation. Meanwhile, the differential chemokine receptor expression of the targeted immune cells can shape the immune response (FIG. 1). Interestingly, the chemokine receptors primarily involved in modulating inflammatory responses are CCR1, CCR2, CCR5 and CXCR3, which ligate numerous chemokines, whereas ‘homeostatic’ receptors include CXCR4 and CXCR5, which are more specific and bind to only one or two ligands⁵³. A distinct chemokine receptor can be expressed by various leukocyte subsets with divergent functions to define the outcome of the immune response. This ligand–receptor diversity demonstrates the complex nature of the chemokine network in liver homeostasis and in the transition to inflammation. Although most chemokines are pro-inflammatory, such as CXCL1, CXCL8 and CCL2, some chemokines are thought to be more important for homeostasis such as CCL12 (REF.⁵⁴). According to in vivo studies, CXCL9, CXCL10 and CXCL11 can have anti-inflammatory and angiostatic effects as they target CXCR3 on endothelial cells, resulting in an inhibitory effect on cell proliferation^55–57 (FIG. 1).

Fig. 1 |. Chemokine pathways in liver homeostasis and inflammation.

In liver homeostasis, Kupffer cells (KCs), liver sinusoidal endothelial cells (ECs) and hepatic stellate cells (HSCs) are in control of surveillance of tissue stress, pathogens and other disturbances. Local immune cells are first responders to chemokines such as CXCL12, maintaining liver homeostasis. When injury escalates, it can cause the release of damage-associated molecular motifs, which activate KCs. This activation induces the production of chemokines and cytokines. Transforming growth factor-β (TGFβ) is produced by KCs and induces the activation of HSCs. CCL2 and CCL5 are important in the process of activation of KCs and HSCs. Tumour necrosis factor (TNF) and lipopolysaccharide (LPS)-induced CXCL1 and CXCL8 are important chemokines that chemoattract neutrophil adhesion and invasion into liver inflammatory sites. Other chemokines, such as CCL1, CCL2 and CCL25, promote the infiltration of monocytes, CXCL9, CXCL10 and CXCL11 promote the infiltration of T helper cells, and CXCL16 promotes the infiltration of natural killer T (NKT) cells. These immune cells also contribute to the local inflammatory milieu. The activation of HSCs by immune cells and chemokines leads to the deposition of collagen in the liver, causing fibrosis. TLR4, Toll-like receptor 4.

In the liver, homeostatic processes control hemodynamic changes, capillary permeability, leukocyte migration into tissues and balance the secretion of inflammatory mediators such as chemokines to avoid pathological and excessive inflammation. Liver sinusoids support a low-pressure blood flow with a fenestrated endothelium, which lacks a basement membrane and allows for increased interactions between resident immune cells and non-haematopoietic hepatic cells. The sub-endothelial compartment, the space of Disse, drains into lymphatic vessels that run parallel to the portal tract. Resident cells, including macrophages, reside within this space of Disse and liver sinusoids⁵⁸. Kupffer cells, LSECs, HSCs and local immune cells constituting the non-parenchymal liver cells, together with hepatocytes and cholangiocytes constituting the parenchymal liver cells, are all important sources of as well as responders to chemokines^59–61. In the inactivated state, liver is enriched for innate immune cells such as Kupffer cells, NKs and NKTs. Upon immune activation, mediated largely by chemokines and cytokines, large influxes of innate and adaptive immune cells infiltrate the liver, which include bone-marrow derived monocytes, neutrophils, T cells and B cells⁶²

The sources of different chemokines in health and disease are likely different^6,58,63 and these differences are poorly understood. For example, CXCL1, CXCL2 and CXCL8 are key chemokines that attract neutrophils, mainly via the chemokine receptors CXCR1 and CXCR2 (REF.⁶). Our screening of public databases such as the FANTOM5 (functional annotation of the mammalian genome 5; see Related links) consortium identified LSECs as a predominant cell type in normal liver tissue that generate CXCL1, CXCL3, CXCL5, CXCL6 and CXCL8 but appear to be a relatively poor source of CXCL2 compared with hepatocytes (S.C., M.L., T.S.S. and V.H.S., unpublished observations)⁶⁴. These points are important as LSECs represent a specialized and integral part of the local immune system owing to their roles in antigen capture and presentation and in directing immune cell extravasation⁶⁵. A study performed on HUVECs demonstrated that CXCL1 and CXCL2 have distinct roles in neutrophil diapedesis⁶⁶. Endothelial cell and pericyte-derived CXCL1 mediate neutrophil crawling, while CXCL2 is neutrophil-derived and retained at endothelial cell junctions by ACKR1, which is an atypical chemokine receptor. The ACKR1–CXCL2 axis is critical for unidirectional paracellular neutrophil transendothelial migration in mice in vivo. It is unclear whether the divergent functions of CXCL1 and CXCL2 are related to their varied production by liver cells. Further studies capitalizing on advances in single-cell sequencing technology have the potential to transform the field by identifying the cellular sources of chemokine production. A 2018 single-cell RNA-seq (scRNA-seq) analysis of normal human and mouse liver tissue was the first to show that a TNF receptor (TNFR1A; encoded by Tnfrsf1a) distributed to LSECs the most, and to hepatocytes to a lesser extent, suggesting that LSECs might be the main cellular sources of various chemokines in response to TNF^67,68. The next challenge is performing scRNA-seq studies using diseased liver tissues, which would lend valuable insight into the sources and, perhaps, functions of chemokines in liver diseases.

Related links.

FANTOM5 database: https://fantom.gsc.riken.jp/zenbu/gLyphs/#config=2Z4giXeUDRQulVO5sw_x8C;loc=hg19::chr4:74734631..74737497

In summary, chemokines are variably expressed in different types of cells in the liver in both physiological and pathological conditions. Their functions vary widely, from pro-inflammatory and angiogenic to anti-inflammatory and angiostatic; they can be regulated by multiple mechanisms, including by epigenetic regulation of enhancer and promoter interactions.

Chemokines in liver diseases

The role of chemokines in the coordination of immune responses in liver diseases has been well established. The balance of various chemokines regulates the constitution of the inflammatory milieu in many acute and chronic liver diseases. Liver immune activation might be conducive to combating diseases, as in the case of viral hepatitis and HCC, but can be detrimental in other contexts such as in alcohol-associated liver disease (ALD), nonalcoholic fatty liver disease (NAFLD) and ischaemia–reperfusion injury (IRI). The paradigm of progression from inflammation to fibrosis to carcinogenesis is well recognized in biology and has particular relevance in the liver. Over 90% of liver cirrhosis cases around the world are caused by ALD, NAFLD and viral hepatitis⁶⁹ and, according to an Italian cohort, 70–90% of cases of HCC occur in patients with cirrhosis^70,71. Given the importance of inflammation in liver disease pathogenesis, it is therefore important to address the role of chemokines in inducing and maintaining the liver inflammatory response.

Many chemokine ligand–receptor pairs have been shown to have important functions in perpetuating the inflammatory cascade and to be important in the pathogenesis of multiple liver conditions. For example, CCL2 and its receptor CCR2 are well characterized in many hepatic diseases, including ALD, NAFLD, viral hepatitis, cirrhosis and HCC. CCL2 is expressed by liver-resident cells, such as Kupffer cells, HSCs and LSECs⁶, and CCL2 acts on CCR2 expressed on bone marrow-derived monocytes (BMDMs) to promote liver chemotaxis as shown in mice studies⁷². Activation of the CCL2–CCR2 axis has been shown in animal studies to promote macrophage accumulation, inflammatory response, fibrosis and steatosis in the liver as well as in adipose tissue in NAFLD⁷³. CCR2-deficient mice have been shown to have a decreased accumulation of inflammatory cells in the liver and the inhibition of the CCL2–CCR2 axis reduces hepatic fibrosis in experimental models^74–76. RNA-seq of human cirrhotic livers found that CCL2 is one of the key chemokines upregulated by scar-associated macrophages after the differentiation from monocytes and is important in fibrotic niche formation⁷⁷. The fibrotic niche organized by these scar-associated macrophages enables cell–cell interactions between macrophages, fibroblasts, endothelial cells and parenchymal cells, leading to fibrogenesis.

Although many chemokine ligand–receptor pairs share similar cellular functions in different liver diseases, the same chemokine can have variable effects on disease progression according to specific pathogenic mechanisms. Certain chemokines or chemokine receptors have functions other than promoting immune cell chemotaxis in disease states. For example, CCL2 has been demonstrated in a mouse model to have a direct effect on lipogenesis gene expression in hepatocytes, thus impeding fatty acid oxidation and promoting steatosis, which might be independent of its actions on the CCR2 receptor⁷⁸. Given these disease-specific properties of chemokines, it is important to explore chemokine functions in the context of inflammatory responses and liver diseases. In this section, we discuss the role of chemokines in several common liver diseases and liver cancer. We also provide a general overview of chemokine function in these disease processes (TABLE 2) and include references to relevant reviews discussing these diseases and others here, which readers can refer to for further details^{53,73,79–87}.

Table 2 |.

Role of chemokines in human liver diseases

Chemokine	Receptor	Proposed role in disease	Function in animal models
ALD
CXCL1, CXCL4, CXCL5, CXCL6, CXCL8	CXCR1, CXCR2	Neutrophil chemoattractant; levels increased in ALD, elevated levels of CXCL1, CXCL5, CXCL6 and CXCL8 correlate with poorer clinical outcomes in alcoholic hepatitis46	Neutralizing CXCL1 by antibodies or genetic deletion or CXCR1-CXCR2 blockade ameliorated liver inflammation in alcohol-fed mice5,103
CCL2	CCR2	Monocyte chemoattractant; levels increased in ALD2; plasma levels and hepatic expression of CCL2 were associated with disease severity in alcoholic hepatitis99	CCL2-knockout mice showed attenuated liver inflammation and steatosis78; pharmacological inhibition of CCL2 ameliorated steatosis development207; CCL2 might have direct effects on steatosis via PPARα78
Nonalcoholic fatty liver disease
CCL2	CCR2	Monocyte chemoattractant; levels elevated in NASH and correlated with fibrosis in NASH217	CCL2 overexpression in adipose tissue led to insulin resistance and hepatic steatosis116; CCR2 inhibition decreased liver macrophage infiltration and fibrosis in NASH model218,219; CCL2 directly affects adipocytes favouring lipogenesis114,115
CCL5	CCR1, CCR5	Recruits T cells and NK cells; levels elevated in patients with NASH122	Levels elevated in animal models of NAFLD and CCL5 pharmacological inhibitors attenuated liver steatosis122,123
CXCL9, CXCL10, CXCL11	CXCR3	Regulate and activate T cells; elevated in NASH125; promote lipid accumulation and induce endoplasmic reticulum stress125; CXCL9 and CXCL10 levels correlate with fibrosis55,125; circulating CXCL10 levels correlated with the degree of lobular inflammation and was an independent risk factor for NASH in patients125	Mice with CXCR3-knockout or blockade of CXCR3 showed reduced steatohepatitis in NASH model125,220; mice with CXCL10-knockout or blockade of CXCL10 were protected against diet-induced NASH125,126; CXCL10-bearing hepatocyte extracellular vesicles were chemotactic for macrophages221
Cirrhosis
CCL2	CCR2	Monocyte chemoattractant; contributes to inflammatory responses leading to steatosis and fibrosis; CCL2 might promote hepatic stellate cell migration132	CCL2-knockout mice showed attenuated liver injury in CCl4 injection model133
CCL3, CCL4, CCL5	CCR1, CCR3, CCR5	Recruits T cells and NK cells; promotes liver fibrogenesis; elevated in patients with cirrhosis136,222; CCR5 may be important for hepatic stellate cell activation136	Deletion of CCL3, CCL5, CCR1, CCR5 or CCL5 inhibition are protective against fibrosis in CCl4 injection model122,137
CXCL9, CXCL10, CXCL11	CXCR3	Recruit T helper cells, particularly TH17 and Treg cells; elevated in patients with cirrhosis55,138; CXCL9 has antifibrogenic and CXCL10 has profibrogenic effects on hepatic stellate ceils55,140	CXCR3-knockout mice showed increased fibrosis in a CCl4 injection model55; CXCR3 inhibition is associated with reduced fibrosis in congenital hepatic fibrosis model223
CXCL16	CXCR6	Survival and maturation factor for NKT cells and promote NKT cell accumulation at injury site142; CXCR6 and CXCL16 expression was increased in the livers of patients with cirrhosis142	CXCR6-knockout mice showed attenuated inflammation and fibrosis in CCl4 injection model142
CX3CL1	CX3CR1	Promotes macrophage survival; levels elevated in cirrhosis; favours the development of anti-inflammatory macrophages143	CX3CR1-knockout mice showed increased fibrosis in CCl4 injection model144
HCC
CCL1	CCR8	Monocyte chemoattractant; recruit myeloid-derived suppressor cells, which inhibit anti-tumour immune surveillance147, but might also recruit anti-tumour immune cells, including CD4+ TH1 cells, CD8+ T cells and NK cells148; CCL1 level was found to be elevated in HCC and even higher in peritumoural liver tissue224; CCL2 is highly expressed in patients with HCC and is a prognostic factor for poor outcome225	CCR2-knockout or blockade inhibits tumour growth and host survival in murine HCC model; inhibiting CCR2+ tumour-associated macrophage infiltration using pharmacological CCL2 antagonist substantially reduced tumour growth226
CCL2	CCR2
CCL17,CCL20, CCL22	CCR6	Recruits Treg cells; elevated levels correlate with poorer survival in patients with HCC227; CCL20 might recruit TH17 cells as well, opposing the action of Treg cells139	CCL20 neutralization suppressed tumour growth and metastasis in murine HCC model228
CXCL12	CXCR4, CXCR7	Immune activation; CXCL12 was shown to have an angiogenic effect; CXCL12 can act on HCC cells and promote cell proliferation, survival and invasion160; CXCR4 and CXCR7 both implicated in tumour growth and metastasis156–161	CXCR7 inhibition decreases angiogenesis and tumour growth in murine HCC model156; the inhibition of CXCR4 decreases angiogenesis in animal models, HCC can be stimulated to produce vascular endothelial growth factor in the presence of CXCL12 (REFS156,157)
CXCL16	CXCR6	Elevated numbers of CXCR6+ cells were seen in patients with HCC164	Elimination of CXCR6+ NKT cells resulted in worsened HCC tumour growth in vivo162,163

This tabel is an abbreviated version of Supplementary Table 1. ALD, alcohol-associated liver disease; CCl₄, carbon tetrachloride; HCC, hepatocellular carcinoma; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NK cells, natural killer cells; NKT cells, natural killer T cells; PPARα, peroxisome proliferator-activated receptor-α; T_H, T helper; T_reg, regulatory T.

Alcohol-associated liver disease

Chronic alcohol use induces a spectrum of pathologies in the liver, ranging from simple steatosis (accumulation of fat) to steatohepatitis (steatosis concurrent with inflammation) to cirrhosis⁸⁸. Cytokines and chemokines have been well recognized to have important roles in the pathogenesis of ALD, largely by promoting inflammation. This effect is most evident in the development of alcoholic hepatitis, which is a highly morbid form of alcoholic steatohepatitis characterized by dense inflammatory cell infiltration, mostly neutrophilic, and hepatocellular injury⁸⁹. The pathogenesis of alcoholic hepatitis is complex and involves the crosstalk between multiple organ systems. Ethanol metabolism in the liver causes oxidative stress and alterations to lipogenesis and fatty acid metabolism, leading to hepatic steatosis⁹⁰. Alcohol not only causes direct hepatocyte toxicity but also increases gut permeability, resulting in the translocation of bacterial products and endotoxaemia^31,91. The activation of immune cells in the human liver, particularly Kupffer cells, by pattern recognition receptor pathways leads to the release of many pro-inflammatory factors, including TNF, IL-1β, IL-6, chemokines and others^92–94. TNF and other cytokines stimulate further chemokine release from immune and non-immune liver cells^95–97 (FiG. 2). This release of chemokines generates chemokine gradients that promote transendothelial migration of immune cells into the liver, which further exacerbates hepatocyte damage by immune-mediated cytotoxicity³¹. Indeed, elevated levels of CCL2, CXCL8 and CXCL5 have been shown to positively correlate with increased neutrophilic infiltration and higher mortality in patients with alcoholic hepatitis⁴⁶. CCL20, a CCR6 ligand that is chemotactic for lymphocytes, has also been shown to positively correlate with disease severity in patients with alcoholic hepatitis and it was suggested to act as a link between inflammation and fibrosis, not only through chemotactic functions but also by acting directly on HSCs⁹⁸. Genome-wide association studies have identified polymorphisms in the genes encoding CCL2, CCR2 and CXCL1 that are associated with the higher serum levels of these chemokines. These polymorphisms were found at higher frequencies in patients with ALD than in healthy individuals, providing correlative evidence of the importance of these molecules⁹⁹.

Fig. 2 |. Chemokines in the pathogenesis of alcohol-related injury.

Lipopolysaccharide (LPS) and bacterial products brought in via splanchnic circulation along with damage-associated molecular patterns (DAMPs) released by injured hepatocytes stimulate Kupffer cells (KCs) to release pro-inflammatory cytokines and chemokines. Immune cells and liver-resident cells respond to this paracrine stimulation by the production of chemokines. CXC chemokines are critical in the recruitment of neutrophils and CCL2 and CCL20 are important for the recruitment of circulating monocytes. Invading immune cells perpetuate injury to hepatocytes by a cell-mediated inflammatory response. CCRs, chemokine receptors; EC, endothelial cell; HSC, hepatic stellate cell; NF-kB, nuclear factor-kB; PAMPs, pathogen-associated molecular patterns; TLRs, Toll-Like receptors; TNF, tumour necrosis factor.

Many rodent models have been developed to study ALD¹⁰⁰. The Lieber–DeCarli chronic–binge alcohol-feeding model is one of the most widely used but, despite being a reliable model of alcoholic steatosis, it generates only a modest degree of liver inflammation¹⁰¹. Continuous intragastric alcohol-feeding models generate more liver inflammation but they are also more invasive and require technical expertise¹⁰². There is a continued need for better animal models of human alcoholic hepatitis to further advance the field. Despite these shortcomings, alcohol-feeding models have assisted in shedding light on the role of chemokines in ALD, particularly in alcohol-induced hepatic steatosis. In the chronic–binge alcohol-feeding model, compared with controls, alcohol-fed mice had higher liver and serum levels of inflammatory markers, including CXCL1, which is a murine homologue of human CXCL8 (REF.⁵). Blockade of CXCL1 by neutralizing antibodies or by G-protein-specific pepducins targeting CXCR1 and CXCR2 ameliorated liver inflammation in alcohol-fed mice^5,103. Alcohol-fed mice deficient in CCL2 (or the pharmacological inhibition of CCR2 and CCR5 signalling) attenuated liver inflammation and steatosis, likely by reducing the infiltration of pro-inflammatory macrophages⁷⁸. Interestingly, in mice, CCL2 might also exacerbate steatosis through the suppression of peroxisome proliferator-activated receptor-α (PPARα)-related pathways in hepatocytes, independent of cognate receptor CCR2 (REF.⁷⁸).

However, the infiltration of immune cells into the liver that occurs in ALD is not purely detrimental. One study found that the increased infiltration of neutrophils in liver biopsy samples was positively correlated with better clinical outcomes in patients with alcoholic hepatitis¹⁰⁴. Neutrophils have important functions in debris removal and release of growth factors, which might promote tissue regeneration¹⁰⁵. In an animal model of alcoholic liver injury, the activation of Kupffer cells through Toll-like receptor 3 (TLR3) was shown to favour an anti-inflammatory phenotype and to induce the release of IL-10, which is an important anti-inflammatory cytokine associated with improvement in inflammation and steatosis¹⁰⁶. Hence, it is unclear whether the elevated chemokine levels seen in patients with ALD are a biomarker reflective of the degree of underlying injury or rather a driver of disease through the promotion of increased immune cell-mediated cytotoxicity. Further research in this area is needed to better delineate the role of chemokines and immune cells in ALD.

Nonalcoholic fatty liver disease

NAFLD is the most common form of liver disease in the United States and its prevalence has continued to increase over the past several decades¹⁰⁷. NAFLD is characterized by increased fat accumulation in the liver and the histological characteristics range from asymptomatic steatosis to nonalcoholic steatohepatitis (NASH) to cirrhosis, sharing many features with alcoholic steatohepatitis¹⁰⁸. Animal models of NASH have been fairly successful in replicating human liver disease and the two most commonly used are the methionine choline-deficient diet and the high-fat and high-cholesterol Western diet, with the latter perhaps better simulating physiology¹⁰⁹. As in alcoholic steatohepatitis, chronic inflammation has an important role in the pathogenesis of NASH but advances in obesity research have highlighted other pathogenic factors. In particular, the crosstalk of the adipose–liver axis has been recognized to have important roles in the development of insulin resistance and NAFLD¹¹⁰. Human adipose tissue macrophages secrete pro-inflammatory cytokines, including TNF and IL-6, which contribute to chronic inflammation^111–113. Multiple chemokines, many of which are upregulated in the livers and adipose tissues of patients with NAFLD, are crucial in macrophage trafficking, both within adipose tissue and into the liver. For example, in mice, the overexpression of CCL2 in adipose tissue leads to the development of insulin resistance and hepatic steatosis and the pharmaceutical inhibition of CCR2 decreases macrophage infiltration and reduces liver inflammation and fibrosis^114,115. In animal models, CCL2 might also have direct effects on adipocytes: it has been shown to decrease insulin-stimulated glucose uptake and to affect the expression of adipogenic genes leading to adipocyte dedifferentiation¹¹⁶.

Within the liver, Kupffer cells are major sources of cytokine and chemokine production in NAFLD. In animal models of NAFLD, the depletion of Kupffer cells was protective against the development of NASH¹¹⁷. However, Kupffer cells and BMDMs in NASH are not universally pro-inflammatory. Single-cell transcriptomics data have shed light on the diversity of macrophage and monocyte populations in NASH and the interplay between these diverse macrophage populations modulates liver inflammation in NASH pathogenesis^118–120. In particular, the emergence of NASH-associated macrophages in the setting of metabolic derangement seems to be instrumental in regulating downstream pro-inflammatory signalling in NASH^118,119. These NASH-associated macrophages have a high expression of TREM2, are found in the liver and adipose tissue of mice and humans with NASH, are positively correlated with disease severity, and are responsive to dietary and pharmacological treatments for NASH in mice^118,119. These have been proposed to undertake an adaptive and protective response in NASH as the deletion of TREM2 (leading to NASH-associated macrophage depletion) in mice induced metabolic dysregulations and exacerbated immune-mediated liver injury^118,121. Similarly, the adoptive transfer of BMDMs from mice fed a normal chow diet to mice fed a Western diet resulted in an exaggerated response to acute sterile liver injury, suggesting the presence of a similar adaptive response to limit the excessive inflammatory response to liver injury in NASH¹²⁰.

In addition to macrophages, lymphocytes, NK cells and neutrophils have also been examined in NAFLD. CCL5, which regulates the recruitment of T cells by interacting with CCR5 and of NK cells by interacting with CCR1, was shown to be elevated in animal models of NAFLD and in livers of patients with NASH¹²². Furthermore, the use of pharmacological inhibitors of CCL5 in mouse models of NAFLD was shown to attenuate liver steatosis^122,123, likely mediated by the inhibition of lymphocyte chemotaxis¹²⁴. Cenicriviroc, a dual CCR2–CCR5 antagonist, garnered much attention owing to its simultaneous effect on two important chemokine pathways — it was shown to be effective in reducing inflammation, steatosis and fibrosis in vivo⁷⁴. Other T cell chemoattractants include CXCL9, CXCL10 and CXCL11, which interact with a common receptor, CXCR3, and have been shown to promote cholesterol-induced steatohepatitis, facilitate lipid accumulation and induce endoplasmic reticulum stress¹²⁵. The pharmacological blockade of CXCR3 in mice can prevent the development of as well as reverse established steatohepatitis. Serum levels of CXCL9 are associated with liver fibrosis in patients⁵⁵ and CXCL10 has been identified as an independent risk factor for NASH¹²⁵. CXCL10 not only induces inflammation but also directly promotes steatosis by stimulating lipogenesis and promoting macrophage-associated liver injury in mouse models of NASH¹²⁶. Neutrophil-attracting chemokines might be upregulated in NAFLD as well. For example, CXCL2 expression was higher in the adipose tissue of patients with obesity¹²⁷ and serum CXCL8 levels were elevated in patients with NASH¹²⁸.

Cirrhosis

Chronic liver inflammation from a variety of injuries leads to the development of liver cirrhosis. Advanced cirrhosis can lead to portal hypertension, liver dysfunction, and multi-organ failure and it predisposes the development of liver cancer. Many cell types can contribute to the progression of liver fibrosis but activated HSCs are considered the most important source of fibrogenic myofibroblasts that deposit collagen and extracellular matrix as part of the pathogenesis of cirrhosis¹²⁹. Chemokines are important factors in this chronic process as they regulate the degree and composition of immune cell infiltration that drives HSC activation¹²⁹. Macrophages, in particular, are the major inflammatory cell type responsible for HSC activation by the production of cytokines, including transforming growth factor-β (TGFβ)¹²⁹. In addition, several chemokines have been shown to act directly on liver parenchymal cells, such as HSCs, to exert profibrogenic or antifibrogenic effects¹³⁰.

The CCL2–CCR2 axis is one of the best-described chemokine ligand–receptor systems in liver fibrogenesis. The binding of CCL2 to CCR2 recruits and activates BMDMs to join the CD14⁺CD16⁺ inflammatory macrophage pool in fibrotic mouse livers, leading to further inflammatory responses and resulting in steatosis and fibrosis¹³¹. Additionally, CCL2 might promote the migration of HSCs via a CCR2-independent mechanism in vitro¹³². The importance of CCL2–CCR2 signalling has been confirmed by various models of liver injury using knockout mice. CCL2-knockout mice have attenuated liver injury after the administration of chronic carbon tetrachloride (CCl₄) injections, a well-established model of liver fibrosis¹³³. Other chemokines in the same subfamily, including CCL25 and its cognate receptor CCR9, have also been implicated as important pathways of BMDM recruitment in acute liver injury as well as in chronic liver fibrosis^130,134. These recruited monocytes are important producers of TNF and were shown to activate HSCs in vitro¹³⁰. CCR9-knockout mice showed attenuated liver fibrosis after CCl₄-induced liver injury¹³⁰.

CCL5 is another chemokine that is important in liver fibrogenesis. CCL5 binds to CCR1, CCR3, and CCR5 and is produced by HSCs and immune cells¹³⁵. CCL5, along with the related chemokines CCL3 and CCL4, is elevated in patients with cirrhosis in the liver¹³⁶. CCR1 is present on bone marrow-derived macrophages and CCR5 can be found on HSCs and seems to be necessary for their activation in animal models¹³⁶. In animal models, deletion of the genes encoding CCL3, CCL5, CCR1, or CCR5 or the pharmacological inhibition of CCL5 are protective against fibrosis^122,137. Another group of chemokines, CXCL9, CXCL10, CXCL11 and their receptor CXCR3, have also attracted attention. These chemokines are elevated in the livers of patients with cirrhosis and are positively correlated with fibrosis^55,138. Functionally, these chemokines recruit T helper (T_H) cells, particularly IL-17-producing T_H17 and IL-10-producing regulatory T (T_reg) cells in the liver¹³⁹. In animal models, CXCR3 deletion increases the susceptibility to fibrosis in mice⁵⁵. Interestingly, CXCL9 and CXCL10 might have opposing roles in fibrogenesis, with CXCL9 being antifibrogenic and CXCL10 being profibrogenic, both through effects on HSCs, as described earlier^55,140. In addition to traditional αβ T cells, γδ T cells were also shown to suppress fibrosis through CCL20–CCR6 interactions. Additionally, CCL20 and CCR6 were shown to be upregulated in the livers of patients with cirrhosis and in mice after chronic CCl₄ injury¹⁴¹. CCR6-knockout mice showed increased fibrosis after CCl₄ injection, which was rescued by the adoptive transfer of wild-type mouse γδ T cells¹⁴¹.

Also worthy of discussion is CXCL16, which is a survival and maturation factor for hepatic NKT cells. CXCL16 is produced early in the fibrotic process by macrophages, resulting in an accumulation of NKT cells at sites of injury¹⁴². Knockout of CXCR6 (the CXCL16 receptor) in mice resulted in attenuated inflammation and fibrosis in models of fibrosis¹⁴². CX₃C chemokines have also been studied in fibrogenesis. CX₃CL1 is widely expressed on immune and non-immune cells in the body and its serum level is elevated in a mouse model of fibrosis¹⁴³. Although CX₃CL1–CX₃CR1 signalling has been shown to promote macrophage survival, it might help to attenuate inflammation by skewing macrophages to an anti-inflammatory phenotype and CX₃CR1-knockout mice demonstrated increased fibrosis after chronic CCl₄ injury^143,144.

Hepatocellular carcinoma

HCC is the most common primary liver cancer. It primarily arises from cirrhotic livers and is a significant contributor to mortality in patients with cirrhosis¹⁴⁵. Chemokines have important roles in the pathogenesis of HCC. Chemokines not only contribute to the inflammatory conditions leading to tumorigenesis and regulate the stromal microenvironment of cancer cells with either pro-tumour or anti-tumour effects but they can also directly interact with chemokine receptors on cancer cells and modulate cell behaviour related to migration, invasion, growth and survival⁶. These chemokines come from a variety of cell sources. HCC cells are known to produce multiple chemokines and non-tumour cells, including immune cells, HSCs and other stromal cells, also contribute to chemokine production¹⁴⁶.

Many chemokines contribute to the pro-inflammatory environment that favours tumorigenesis, as discussed in previous sections (FIG. 3). After a tumour arises in the liver, abnormal aggregates of immune cells are often found surrounding or infiltrating tumour tissue. Chemokines are key regulators of this process. For example, the CCL2–CCR2 axis is important in the recruitment of monocytes. In addition, this axis can also recruit myeloid-derived suppressor cells, which has an inhibitory effect on anti-tumour immune surveillance¹⁴⁷. At the same time, CCL2 levels in the liver positively correlate with increased anti-tumour immune cells, including CD4⁺ T_H1 cells, CD8⁺ T cells and NK cells, illustrating the potential anti-tumour role of this chemokine¹⁴⁸. Another important chemokine that promotes the growth of HCC cells is CCL20. The CCL20–CCR6 axis recruits T_reg cells into tumour tissue, an increased number of which is correlated with poorer survival in patients with HCC¹⁴⁹. CCL17 and CCL22 might have similar effects in recruiting T_reg cells¹⁵⁰. However, increased levels of CCL20 might recruit T_H17 cells as well, thereby opposing the action of T_reg cells¹³⁹. Neutrophilic infiltration in HCC is also commonly seen. Of this process, the CXCL8 and CXCL5–CXCR2 axes are the best characterized. In patients with HCC, higher levels of CXCL1, CXCL8 and CXCL5 are associated with increased neutrophil infiltration and poorer prognosis^151–153.

Fig. 3 |. Chemokine actions in HCC.

Hepatocellular carcinoma (HCC) tumour cells produce a large number of chemokines. Among them, CXCL5 and CXCL8 are known to directly stimulate tumour cell growth. They also promote neutrophil infiltration into the liver. CCL2 promotes monocyte recruitment, a subpopulation of which give rise to myeloid-derived suppressor cells (MDSCs) that suppress the anti-tumour immune response. CCL20 can promote the recruitment of T helper cells. These chemokines also have important roles in angiogenesis: multiple chemokines, such as CXCL5, CXCL8 and CXCL12, promote angiogenesis, whereas CXCL9 suppresses it. CXCL5 and CXCL12 also promote tumour invasion and metastasis. These specific enhancers provide a unique angle with which to study the activation of transcriptional responses to signals such as tumour necrosis factor (TNF) in liver diseases^32,41. EC, endothelial cell; VEGF, vascular endothelial growth factor.

HCC is a hypervascularized tumour that relies on angiogenesis for its growth and survival¹⁵⁴. Many chemokine pathways are directly involved in this process. According to preclinical studies, the CXCL12–CXCR4/CXCR7 pathway has pro-angiogenic effects on HCC¹⁵⁵. The inhibition of CXCR4 decreases angiogenesis in animal models of HCC and HCC can be stimulated to produce vascular endothelial growth factor (VEGF) in the presence of CXCL12 (REFS^156,157). CXCL8 is another chemokine that has pro-angiogenetic effects. The neutralization of CXCL8 can interfere with tumour endothelial cell capillary formation in vitro¹⁵⁸. However, not all chemokines have pro-angiogenic effects and CXCL9 is a prominent example of a chemokine with anti-angiogenic effects that can be produced by HCC cells¹⁵⁹.

Chemokines can also influence HCC growth by directly binding to tumour cells. CXCL12 can act on human HCC cells and promote cell proliferation, survival and invasion in vitro¹⁶⁰. Clinically, the presence of CXCR4 (the receptor of CXCL12) correlates with an increased risk of metastasis and poor prognosis¹⁶¹. CXCR7 is also highly involved in the migration and invasion of tumour cells — its knockdown decreases tumour growth in vitro¹⁵⁶.

The role of the gut microbiota has also been recognized in anti-tumour immunity in relation to the action of chemokines. Intestinal flora is involved in the process of bile conjugation and in converting primary bile acids into secondary bile acids. Secondary bile acids have long been shown to have important tumorigenic properties¹⁶². Work by Ma et al. showed in a mouse model that an accumulation of specific gut microorganisms, such as Clostridium species, leads to increased secondary bile acid production and the suppression of these microorganisms with antibiotics leads to an increase in hepatic CXCR6⁺ NKT cells mediated by increased CXCL16 secreted by LSECs¹⁶³. The CXCL16–CXCR6 axis has also been implicated in tumorigenesis through the removal of senescent cells. Adoptive transfer experiments by Mossanen et al. demonstrated that CXCR6⁺ NKT and CD4⁺ T cells are involved in the removal of senescent hepatocytes, which prevents tumorigenesis¹⁶⁴. This interaction again showcases the importance of chemokines in orchestrating the anti-tumorigenic immune response and illustrates that chemokines are mediators that enable other organs to directly affect liver anti-tumorigenic immunity^146,165,166.

Viral hepatitis and chemokines

Chronic viral hepatitis due to HCV or hepatitis B virus (HBV) infections are important causes of liver disease around the world. Cytokines and chemokines have vital roles in the clearance or persistence of hepatitis viruses and these viruses can modulate the expression of chemokines and their receptors, which in turn hampers effective host immune responses¹⁶⁷. Research on hepatitis viruses is hindered by the lack of widely available animal models due to the species-specific tropism of human hepatitis viruses. Chimpanzees, woodchucks, ducks, and humanized or transgenic mice have been used to study hepatitis virus infections, each with its own limitations^168,169. Despite these difficulties, research on hepatitis viruses, particularly HCV and HBV, has made tremendous progress and has greatly advanced our understanding of liver antiviral immunity. Upon infection with HCV, infected hepatocytes and sentinel cells of the immune system, including dendritic cells and macrophages, are stimulated to produce cytokines and chemokines by the activation of pattern recognition receptor pathways¹⁶⁷. Dendritic cells and macrophages also serve as antigen-presenting cells to activate the adaptive immune system¹⁶⁷. Strong CD4⁺ T_H1 helper and CD8⁺ effector T cell responses have been found to be critical in successful HCV resolution in humans, whereas virus-specific T cell responses are suppressed in chronic HCV infection¹⁷⁰. A myriad of pro-inflammatory chemokines are produced in response to IFNγ and TNF and have important roles in HCV immune responses. For example, the chemokines CXCL9, CXCL10 and CXCL11 are produced by infected liver cells and activate CXCR3 on T helper cells¹⁴⁰ and CCL3, CCL4 and CCL5 are similarly upregulated and bind to CCR5 on cytotoxic CD8⁺ T cells, promoting a T_H1 immune response¹⁷¹. These chemokines are found to be substantially elevated in patients with acute HCV infections compared with healthy individuals and several chemokines have been suggested to be useful biomarkers. In particular, CXCL10 levels were associated with HCV-induced liver fibrosis as well as with fibrosis after liver transplantation^172,173. The innate immune effector cells NK and NKT cells are also important in the HCV immune defence. These cells exert non-MHC-restricted cytotoxicity and produce abundant IFNγ and TNF¹⁶⁷. However, in patients with chronic HCV infection, NK cell dysfunction can occur, with impairments in both degranulation and IFNγ production¹⁶⁷. In the age of direct-acting antiviral agents, increasing numbers of patients are treated and have achieved sustained viral response or virological cure. After achieving virological cure, cytokine and chemokine levels are significantly decreased, although some data suggest that the clearance of HCV might not result in complete immunological restitution, as certain cytokine levels might remain elevated compared with healthy controls and CD8⁺ T cell dysfunction might persist after clearance of the virus^174–176. Many chemokines are also significantly elevated in patients with HBV infection compared with healthy individuals, including CXCL9, CXCL10, CXCL11, CXCL13 and CCL4 (REFS^177,178). However, the role of chemokines in disease clearance is less clear. In animal models, the neutralization of chemokines (such as CXCL9 and CXCL10) resulted in a reduction of liver injury but not in changes in HBV-specific CD8⁺ T cell activity¹⁷⁹. Given that HBV-specific CD8⁺ T cells are thought to be most crucial in the clearance of the virus¹⁸⁰, it seems that the chemokine-mediated inflammatory response contributes to liver injury but perhaps not to viral clearance in HBV. Consistent with this theory, serum levels of CXCL8, CXCL9 and CXCL10 are increased in patients with chronic HBV flare without changes in levels of HBV-specific CD8⁺ T cells¹⁸¹.

IRI and chemokines

Ischaemia is an important mechanism of liver injury, particularly during surgical procedures such as liver resection or transplantation. Cellular injury incurred during periods of ischaemia is amplified when perfusion is restored, in a process termed IRI¹⁸². In liver transplantation, minimizing IRI is important for allograft survival, not only in the immediate post-operative period but also for long-term graft function¹⁸³. IRI has also been shown to correlate with the likelihood of transplant rejection¹⁸⁴. Cytokines and chemokines are critical in the pathogenesis of IRI. The initial ischaemic insult causes the release of endogenous danger-associated molecular patterns from damaged cells that activate pattern recognition receptor pathways in liver-resident cells, particularly Kupffer cells, in the reperfusion phase. Kupffer cells produce large amounts of pro-inflammatory molecules, including cytokines and chemokines, which can be elevated by hundred-fold to thousand-fold above baseline¹⁸⁵. This chemokine rush produces a very robust chemotactic gradient for the recruitment of immune cells, especially neutrophils, into injured tissue, leading to immune cell-mediated toxicity¹⁸⁶. In humans, CXC chemokines are among the most important chemokines upregulated in IRI¹⁸⁶. In mice, the blockade of CXCR2 with a pharmacological antagonist attenuates liver injury in a partial hepatic ischaemia model and, in rats, the neutralization of the homologue of human CXCL5 and CXCL6 resulted in decreased neutrophil sequestration and serum ALT levels^187,188. Interestingly, in mice models, there appears to be a temporal pattern to the upregulation of certain chemokines. For example, CXCL2 appears to be elevated later in the course of injury and was also elevated in the non-ischaemic lobe¹⁸⁹. The upregulation of these chemokines is inducible by TNF as shown in experimental models of TNF or lipopolysaccharide accentuating liver injury in the setting of IRI¹⁸². Therefore, it is possible that allografts that have sustained a prior insult that activated the TNF signalling pathway, resulting in elevated pro-inflammatory cytokines, might be more at risk for IRI as is the case for steatotic allografts¹⁹⁰.

Therapeutic opportunities

Chemokines are instrumental in organizing immune cell movement and recruitment through both spatial and temporal differential expression¹⁹⁰. As described earlier, different chemokines can have pro-angiogenic or anti-angiogenic and proliferative properties, making them potential therapeutic targets for immunotherapy against liver disease and cancers^58,191 (TABLE 3; FiGS 1–3). Some chemokines might also act as biomarkers for liver diseases. Intrahepatic and serum levels of CXCL10 and its receptor, CXCR3, positively correlate with the severity of HCV-related liver disease in patients^192–194. Similarly, CXCL10, along with CXCL8 and CXCL9, has shown potential as a biomarker for diagnosing HBV in patients^177,195. CXCL1 and CCL2 levels have been shown to correlate with the disease process in NAFLD and ALD; however, the relevance of these observations needs to be further tested in larger studies^196,197. Chemokines can provide insight into liver cancer activity as well. CXCL8 levels have been associated with increased tumour burden and disease aggressiveness in patients with HCC¹⁵¹. Multiple chemokines have been implicated in the activation of angiogenesis in HCC via neutrophil aggregation and can provide information about tumour invasion and progression¹⁹⁸. The overexpression of CXCR2 and CXCL7 has been shown, in a small study (n = 51), to be correlated with shorter disease-free (15.4 ± 2.0 versus 31.8 ± 2.6 months, P = 0.002) and overall (29.8 ± 1.7 versus 65.4 ± 3.3 months, P = 0.012) survival in patients with liver metastasis¹⁹⁹.

Table 3 |.

List of chemokine and chemokine receptor inhibitors in clinical trials

Chemokines and receptors	Drugs	Current status
Ligands
CXCL1	–	Important role in pathophysiology of multiple stress and inflammation-associated liver diseases, making it a potential target229
CCL2	mNOS-E36	Shown to be of importance in alcohol-associated liver disease, a potential therapeutic target; compound shown to have benefit against inflammation in murine models of NASH and CCl4–induced hepatic injury207
CCL5	–	Antagonism shown to have benefit against liver fibrosis in animal models and is a potential new target122
CCL20	–	miR-590–5p-associated downregulation of CCL20 reduced fibrosis in NASH animal models, making it a novel therapeutic target203
CXCL10	NI-0801 (mAb)	Phase IIa studies show a good safety profile in patients with primary biliary cholangitis but failed to show clinical benefit204
Receptors
CCR2	CCX872	Demonstrated good safety profile in phase II trials; these studies are based on promising effects in NASH murine models205,206
CXCR1/CXCR2	Reparixin	After demonstration of good safety profile in phase II studies, trials are under way for ischaemia-reperfusion injury prevention in liver transplant recipients208
CCR2/CCR5	Cenicriviroc	Multicentre phase II trials showed safety and efficacy signals; reduction of fibrosis at year 1 and maintenance at year 2 were noted in the phase IIb CENTAUR trial in patients with NASH compared with placebo; follow-up phase III trial is under way in patients with NASH and results are eagerly awaited209,210 (AURORA; NCT03028740)

‘–’ in the Drugs column denotes no current named compounds in trials. CCl₄, carbon tetrachloride; mAb, monoclonal antibody; NASH, nonalcoholic steatohepatitis.

Preclinical studies

Serum levels of CCL2 are increased in patients with alcoholic hepatitis, which is thought to be involved in the pathogenesis of ALD. A study showed that CCL2 levels are higher in patients with severe alcoholic hepatitis and biopsy-proven ALD compared with healthy individuals as controls. It also showed that patients who consumed the same amount of alcohol were more likely to develop severe alcoholic hepatitis with the G allele for the −2518A>G CCL2 polymorphism^99,200,201. Studies in mouse models have previously also shown that CCL2 deficiency protects against alcohol-induced liver injury, independent of CCR2, possibly by the inhibition of pro-inflammatory cytokines⁷⁸.

Similarly, CCR5 haplotypes and mRNA expression in livers were found to be associated with severe liver fibrosis in patients. CCR5 was also shown to be expressed on the surface of HSCs^136,202. CCR5 antagonism has since been shown to reduce fibrosis in multiple animal models of fibrosis¹²². This observation carries potential therapeutic implications for NASH, alcoholic hepatitis and chronic HCV, among others.

CXCL1 is another chemokine that could be a potential therapeutic target, especially for primary inflammatory conditions such as alcoholic hepatitis and liver fibrosis. An earlier study, from 2016, showed that CXCL1 was released by hepatocytes in mouse liver^5,47. Later evidence, including our own data, suggests that endothelial cells are one of the main sources of this chemokine in the human liver⁴⁷. Its role is also being investigated in similar pathophysiological conditions such as lung fibrosis and atherosclerosis.

CCL20 ligand levels have been shown to correlate with worsening severity of liver histology in patients with NAFLD. A potential therapeutic benefit of exogenous miR-590-5p, which functionally interacts with the CCL20 3′ UTR region to downregulate its expression, has also been demonstrated. This observation points to a novel mechanism of CCL20 and HSC interactions promoting fibrogenesis in NAFLD²⁰³.

Clinical trials

Encouraged by the positive results of in vitro studies, an open-label phase IIa trial was conducted for NI-0801, a selective fully human monoclonal antibody against CXCL10, in ursodeoxycholic acid non-responder patients with primary biliary cholangitis. Even though the drug demonstrated a good safety profile, it was unfortunately unable to provide clinical benefit to study participants²⁰⁴.

The selective inhibition of CCR2 was found to be effective in NASH. The drug CCX872 has demonstrated a good safety profile in phase I studies as well as improvement in NASH animal models²⁰⁵. It has also been tested in a phase II trial as a combination therapy for pancreatic cancer, again demonstrating a good safety profile as well as some improvement in overall survival²⁰⁶. Another way to disrupt CCR2 signalling is to inhibit its ligand, CCL2. The CCL2 inhibitor mNOX-E36 was shown to reduce intrahepatic macrophage accumulation and steatosis development in NASH and toxic CCl₄ mouse models²⁰⁷.

After showing a good tolerability and safety profile in multiple phase I and II studies, a small molecular inhibitor of CXCR1 and CXCR2, reparixin, is currently undergoing a phase II trial for IRI in liver transplantation and early allograft dysfunction²⁰⁸.

Cenicriviroc, an oral and dual CCR2–CCR5 antagonist, has been tested in multiple phase II studies (NCT01092104, NCT01338883) conducted in patients infected with HIV and has shown CCR2 blockade (with a reciprocal increase in CCL2 levels) and CCR5 blockade (as demonstrated by a reduction in HIV1 RNA levels). It has also demonstrated a good safety and tolerability profile. Notably, lower levels of soluble CD14, which is a circulating biomarker of monocyte activation, were observed in adults with NASH and liver fibrosis in the recent CENTAUR study after 48–52 weeks of therapy (n = 289; a randomized double-blinded phase IIb study)²⁰⁹. This was further analysed and aspartate aminotransferase-to-platelet ratio index (APRI) and fibrosis-4 index (FiB-4) were lower at 48 weeks, indicating a reduction in fibrosis. Preliminary analysis (1 year) of the 2-year CENTAUR study affirmed the safety of cenicriviroc and resulted in double the number of individuals achieving at least a one-stage fibrosis reduction compared with placebo. Furthermore, they noted twice the number of individuals showing no steatohepatitis worsening versus placebo. The final analysis of the CENTAUR study, after 2 years of therapy, has shown that, even though cenicriviroc is not associated with further improvement in fibrosis after year 1, the durability of the fibrosis benefit was higher compared with the placebo group. This was most remarkable in the advanced (stage 3) fibrosis group. The safety profile of the drug was well maintained throughout year 2 (REF.²¹⁰). Cenicriviroc is currently undergoing a phase III evaluation for the treatment of liver fibrosis in adults with NASH (AURORA; NCT03028740). At the time of writing this Review, cenicriviroc is regarded as one of the most promising therapeutics under investigation for NASH.

One major drawback in limiting the translational potential of chemokine and chemokine receptor inhibitors is the redundancy in the function of inflammatory chemokines. Rather than targeting any specific chemokine, a more efficient approach might be using chemokine receptor antagonists with specificity for multiple targets or modulating the epigenetic regulation of many chemokine targets such as using a suppressor of chemokine super-enhancers. One study demonstrated that a pharmacological inhibitor of BET proteins, GSK620, attenuated liver steatosis and inflammation in a mouse model of NAFLD along with decreased CCL2 expression²¹¹. As discussed earlier, BET proteins, such as BRD4, are required for the function of super-enhancers and pharmacological BET inhibitors modulate super-enhancer-regulated genes, such as those encoding chemokines, with high specificity and provide a novel approach to modulating the inflammatory response^212,213. The pan-BET inhibitors are effective in mouse models of hepatic steatosis and fibrosis^211,212. Even though pan-BET inhibitors have shown clinical effectiveness in humans, they are not close to being used in clinical practice owing to their potential adverse effects. Phase I trials have noted grade 1/2 adverse effects, such as dysgeusia and vomiting, as well as grade 3/4 adverse effects, including elevated liver enzymes and bilirubin^214,215. Studies currently aim to delineate the functions of the BD1 and BD2 bromodomains that are present in each of the four human BET proteins and they are leading to the development of next-generation epigenetic inhibitors for more specific inflammation or cancer BET-targeting therapies²¹².

To further advance chemokines as therapeutics, there needs to be an impetus towards translating animal studies to human clinical trials. Challenges include a mismatch in functions of chemokines between different species and an ineffectiveness to determine the correct dose to stay within the safety window but still impart therapeutic effect. Another limitation of current clinical trials is the standalone use of chemokine inhibitors to elicit a meaningful response in patients with liver diseases. As advanced liver disease is characterized not only by inflammation but also by fibrosis and other hallmark features, future clinical trials can be designed to test combination therapies of chemokines with other pathophysiological targets of the specific type of liver disease.

Conclusions

Chemokines are expressed in a temporal and spatial manner in the pathogenesis of several acute and chronic liver diseases and have many different functions (pro-inflammatory, anti-inflammatory, angiogenic, angiostatic, liver remodelling). Many cell types can express and secrete chemokines in damaged livers, including most liver-resident cells and infiltrating immune cells. To clarify what cell types are the main sources of chemokines in human or animal models is challenging. In addition, their interactions with G-protein-coupled receptors are highly relevant to the clinic. Likewise, many liver and immune cells also express different chemokine receptors. Altogether, the local concentrations of chemokines and their interactions with receptors determine the progression and outcome of associated liver damage.

With the rapid development of next-generation sequencing and the development of 3D chromatin capture technology and single-cell sequencing, more light will be shed on the multiple cell type-specific sources and roles of chemokines in the liver (BOX 1). For instance, a 2019 study profiled the transcriptomes of more than 100,000 single human cells in healthy and cirrhotic human liver by scRNA-seq⁷⁷. This study identified a scar-associated TREM2⁺CD9⁺ subpopulation of macrophages that expands in liver fibrosis and defined ACKR1⁺ and PLVAP⁺ endothelial cells²¹⁶ that are topographically restricted to the fibrotic niche in cirrhosis and enhance the transmigration of leukocytes. ACKR1 is an atypical receptor for CXCL2, which has been indicated to be critical for unidirectional paracellular neutrophil trans-endothelium migration in vivo⁶⁶. All of these studies support the notion that different CXCLs can act in a sequential manner to guide neutrophils through blood vessel walls as regulated by their distinct cellular sources. Nevertheless, more new insights and discoveries on chemokines and their interactions with receptors are warranted and many more studies are necessary before those new findings can be translated into clinical targets. Altogether, it is conceivable that this work, by interfering with the production of pro-inflammatory or angiogenic chemokines and/or their receptors, might effectively reduce or eliminate chronic inflammation and angiogenesis in the treatment of liver diseases.

Box 1. Open questions and future directions.

• It is critical to clarify the cellular sources of chemokines and their target receptors. With the application of single-cell RNA sequencing technology in the field of liver diseases, we will be able to answer this question with increased precision in the future.

• The single-cell technique has also been adopted in genomic functional studies such as single-cell Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) and single-cell ChIP–seq, which hold tremendous potential. These techniques will enable the investigation of gene transcriptional regulation with single-cell precision, which can help refine our understanding of cell type-specific and context-specific gene expression.

• The majority of current therapeutics being studied in clinical trials are receptor inhibitors, ligand antagonists or neutralizing antibodies directed to either the chemokine or its receptor. More specific approaches to affect chemokine-mediated signalling pathways using CRISPR–dCas delivery or with extracellular vesicles as drug loading carriers have the potential to transform the field.

• In addition to treating liver diseases by targeting specific chemokines or their receptors, chemokines might be important biomarkers to enable the early detection of liver damage in certain liver diseases.

• Chemokines and their receptors have long been identified as promising pharmacological targetable molecules. With the fast-paced developments in single-cell technology and 3D chromatin genomic functional studies, we will be ready to embrace these bioinformatics-driven technologies and new insights will be gained from them in the understanding and treatment of liver diseases.

Key points.

• Chemokines are secreted mediators that regulate the infiltration of immune cells into the liver and modulate the activation and proliferation of almost all liver cell types.

• Individual chemokines can interact with their corresponding receptors based on their structural characteristics and local context and can contribute to liver homeostasis by interacting with immune and non-immune liver cells.

• Tumour necrosis factor (TNF), IL-6 and many other cytokines can rapidly stimulate the transcription of pro-inflammatory chemokines in a robust and co-regulated manner through the epigenetically primed chromatin 3D conformation of enhancer–promoter interactions.

• Chemokines are associated with liver disease and it is critical to identify different major players that provide the similarities and dissimilarities between mouse and human genomic arrangement.

• Current drug development based on chemokines and receptors in liver diseases is limited and can be greatly improved with novel therapeutic strategies.