In DILI, necrotic hepatocytes show many damage-associated molecular patterns (referred to as DAMPs), such as high-mobility group box-1 protein, DNA fragments, and heat shock proteins[12], and these factors can be identified by Toll-like receptors (commonly known as TLRs) on innate immune cells. Then, proinflammatory factors recruit inflammatory immune cells into the liver, activating them to remove necrotic cell debris[13].

Liver macrophages mainly include two cell types: resident Kupffer cells and infiltrating monocyte-derived macrophages. Although of different cellular origins, both types of macrophages can phagocytose microorganisms and metabolic waste in liver sinusoids. The numbers of liver macrophages become greatly increased in any type of liver injury, due to the self-renewal ability of Kupffer cells and the infiltration of monocyte-derived macrophages[14]. In the early stages of liver injury, Kupffer cells recognize DAMPs derived from damaged hepatocytes and then secrete several proinflammatory cytokines and chemokines to attract neutrophils, NK cells, and bone-marrow-derived monocytes to the regions of inflammation[15]. Ly6C⁺ monocytes and Ly6C^- monocytes exist in the blood of mice, and can differentiate into hepatic macrophages. Infiltrating Ly6C⁺ monocytes promote inflammation and induce organ impairment, but eventually maturate with the downregulation of Ly6C expression, at which point they acquire the ability to restore liver integrity[16].

Neutrophils are the first-line immune cells that have the fastest response when inflammation occurs. However, uncontrolled neutrophil infiltration and activation lead to excessive inflammation in chemically-induced liver injuries. The expression levels of C-X-C motif chemokine ligand (referred to as CXCL) 1, IL-6, TNF-α, and monocyte chemoattractant protein-1 in the injured liver are significantly increased to regulate the infiltration and activation of neutrophils[17]. Tissue-resident phagocytes, including macrophages and DCs, release a variety of proinflammatory mediators and establish a chemoattractant gradient, triggering neutrophil recruitment into tissues. Neutrophils express receptors (G protein-coupled receptor, Fc-receptors, adhesion molecules, TLRs, C-type lectins) that can recognize these signals and then release granules (myeloperoxidase), generate reactive oxygen species, and form neutrophil extracellular traps[18,19].

DCs in ALF engage in the innate immune response involving macrophages and neutrophils, with antigen recognition by pattern-recognition receptors. However, the most important effect of DCs is initiation of the adaptive immune response. DCs reside in organs such as the liver as immature cells, which are very effective at antigen recognition, capture and processing, and then circulate in the blood or lymph fluid to peripheral immune organs where they can achieve terminal maturation with the ability of efficient antigen presentation as well as activation of T cells[20].

In the healthy state, hepatocytes express MHC-I, which binds to inhibitory receptors on NK cells, preventing NK cell activation[21]. In contrast, infected hepatocytes lacking MHC-I can be recognized and eliminated by NK cells[22].

Contrary to innate immune responses, which induce acute liver injury (ALI) in experimental animal models[13], adaptive immune responses play an undefined secondary role in DILI[23]. In homeostasis, liver sinusoidal endothelial cells and Kupffer cells constitutively express IL-10, prostaglandins, TNF-α, and transforming growth factor-beta (TGF-β) to expand Tregs, attenuate T cell activation, and induce liver immune tolerance[12]. In some ALI models induced by some special types of drugs or chemicals, T cells are important. Yu et al[24] found that IL-1β is upregulated in the AAGL (Agrocybe aegerita galectin) model and is crucial to recruit T cells from peripheral blood into the injured liver; treatment with IL-1β antibody can significantly alleviate hepatocyte damage. The possible mechanism may be inhibition of p38 or nuclear factor-kappa B (NF-κB) signaling pathways and subsequently reduced infiltration of T cells into the liver. Heymann et al[25] applied a concanavalin A (Con A)-induced liver injury model to mimic immune reactions observed in humans, trigger an inflammatory cascade by activating resident Kupffer cells, initiate neutrophil infiltration, and increase CD4⁺ T cell infiltration and activation[26]. Although Tiegs et al[27] showed that CD4⁺ T cells play a more critical role than CD8⁺ T cells in Con A-induced liver injury in wild-type mice, CD8⁺ T cells played an important role in T cell-transferred Rag2-knockout mouse (in which T cells cannot mature) challenged with Con A. Both IL-33 released by the injured liver and perforin secreted by CD8⁺ T cells were crucial components in their study.

Quiescent MSCs display immune homeostatic features biased towards suppression. When MSCs are induced by various proinflammatory cytokines, these immuno-suppressive properties can be considerably enhanced, resulting in polarization to immunosuppressive phenotypes of MSCs. IDO and inducible nitric oxide synthase (referred to as iNOS) are the key to the immune regulatory functions of MSCs, with a series of potential complementary suppressor pathways, including heme oxygenase-1, soluble human leukocyte antigen-G5, TGF-β, PGE2, galectin, and TSG-6[28]. MSCs possess a pro-inflammatory or anti-inflammatory phenotype by contacting immune cell responses in different situations, and regulate the immune response by secreting soluble factors or direct cell contact[6].

MSCs can regulate the proliferation and activation of Kupffer cells, macrophages, DCs, neutrophils, and NK cells (Figure 2). MSCs reportedly transfer to injury sites in response to large amounts of inflammatory factors, such as IL-6 and TNF-α secreted by activated Kupffer cells[29]. MSCs, in turn, inhibit the phenotype transition of activated Kupffer cells to M1 and stimulation of the NF-κB pathway in lipopolysaccharide (commonly known as LPS)-treated Kupffer cells[30]. Several studies have revealed the mechanism by which MSCs cause immunosuppression via the interaction of macrophages. MSCs can interact directly and physically with innate immune cells. Upregulated CD54 on human MSCs (referred to here as hMSCs) co-cultured with M1 macrophages in an in vitro co-culture system increased IDO activity and inhibited the proliferation of T cells[31]. Similarly, upregulated CD200 on mouse bone marrow stromal cells (BMSCs) in contact with M1 macrophages can also enhance the immunotherapeutic effects of MSCs to reprogram proinflammatory macrophages[32]. On the other hand, soluble factors secreted by MSCs can contribute to the immunoregulatory properties of MSCs. Corneal-derived MSCs can secrete pigment epithelium-derived factor and then modulate the immunophenotype and angiogenic function of macrophages[33]. TSG-6 and PGE2 secreted by MSCs have also been widely studied, due to their immunoregulatory effects on MSCs and macrophages[32,34].

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Mesenchymal stromal cells regulate innate and adaptive immune cells. Mesenchymal stromal cells (MSCs) regulate innate and adaptive immune cells through soluble factors and direct cell-to-cell contact. Breg: Regulatory B cell; CXCL2: C-X-C motif chemokine ligand 2; CXCR2: C-X-C motif chemokine receptor 2; IDO: Indoleamine 2,3-dioxygenase; IL: Interleukin; PEDF: Pigment epithelium-derived factor; PGE2: Prostaglandin E2; Treg: Regulatory T cell; TSG-6: Tumor necrosis factor-alpha-stimulated gene-6.

In addition, MSC-derived exosomes, which are rich in proteins, mRNAs, and microRNAs (designated as miRs), have been used as a therapy for liver diseases in recent years. In a study of experimental autoimmune hepatitis, BMSC-derived exosomes, which are rich in miR-223, effectively alleviated liver injury by downregulating formation of the NLR family pyrin domain containing 3 (referred to as NLRP3) inflammasome[35]. In another study, miR-17 derived from adipose tissue-derived MSC (AMSC)-derived exosomes was shown to ameliorate LPS/D-galactosamine-induced ALI by inhibiting activation of the TXNIP/NLRP3 inflammasome of macrophages[36]. MSCs can limit neutrophil recruitment or infiltration and inhibit neutrophil activation to prevent an excessive inflammatory response. MSCs ameliorate the hepatic inflammatory response by reducing the release of neutrophil chemoattractant CXCL2 and attenuating neutrophil chemotaxis via downregulation of C-X-C motif chemokine receptor 2 expression in neutrophils[37]. In a septic mouse model, MSCs optimally balanced the distribution of circulating and tissue-infiltrated neutrophils, maximizing bacterial killing and minimizing liver injury[34].

Many studies have focused on the effects of MSCs on DCs in vitro, especially through soluble factors. One showed that TSG-6 secreted by mouse BMSCs suppressed the maturation and activation of DCs by inactivating mitogen-activated protein kinase and NF-κB signaling pathways[38]. In addition, IL-6 reportedly participates in the immune regulation mechanism mediated by the murine MSC line via inhibition of DCs[39]. Regarding hMSCs, Spaggiari et al[40] demonstrated that human BMSCs can secrete PGE2 to inhibit differentiation of monocyte-derived DCs. Selleri et al[41] showed that human umbilical-cord-derived MSCs can secrete lactate to induce granulocyte-macrophage colony-stimulating factor/IL-4-treated monocytes to differentiate into M2 macrophages rather than DCs by metabolic reprogramming. A study by Liu et al[42] showed that MSCs derived from mouse embryonic fibroblasts can induce a type of novel regulatory DCs that express low levels of CD11c and Ia and are phenotypically different from immature and mature DCs via IL-10.

In addition, direct cell contacts are as important as soluble molecules in MSC/DC interactions. hMSCs can inhibit the proliferation ability and differentiation capability of CD34⁺ hemopoietic progenitor cells into interstitial DCs but cannot inhibit maturation of CD34⁺ hemopoietic progenitor cell-derived DCs, and the inhibitory effect is associated with the Notch pathway[43]. Cahill et al[44] demonstrated that mouse MSC induction of functional tolerogenic DCs that can induce Tregs in vitro requires Notch signaling. This hypothesis was confirmed in an animal model treated with Jagged-1 knockdown MSCs. In another study, mature DCs cocultured with MSCs expressing Jagged-2 acquired tolerogenic properties[45]. MSCs can influence the proliferation capacity, cytokine release, phenotypic conversion, and cytotoxicity of IL-2-induced NK cells[46]. The mechanism may include TGF-β[47], PGE2[48], IDO[49] and exosomes[50]. Moreover, NK cells can stimulate MSC recruitment by secreting neutrophil-activating peptide 2[51], and activated NK cells can efficiently lyse MSCs[46].

MSCs can inhibit proliferation, activation, and differentiation of T cells, induce apoptosis of T cells and induce recruitment of Tregs. MSCs can also induce cell cycle arrest by downregulating cyclin D2 and upregulating p27kip1 in T cells, resulting in division anergy of activated T cells[52]. Of note, MSCs can inhibit T cell function by inducing apoptosis of activated T cells. Plumas et al[53] showed that this apoptosis can be associated with transformation of tryptophan to kynurenine by IDO expressed by MSCs in the presence of IFN-γ. Akiyama et al[54] revealed that BMSCs may trigger apoptosis of transient T cells through the Fas ligand (FasL)-dependent pathway, and that apoptotic T cells can induce production of TGF-β in macrophages, thereby upregulating Tregs. In some experiments in vitro, MSCs were shown to inhibit allogeneic T cell responses in mixed lymphocyte reactions by IDO after MSCs were activated by IFN-γ[55]. Recently, MSCs were shown to inhibit activation of CD4⁺ T cells and reduce secretion of IL-2 via PD-1 ligands[56].

Several investigators have highlighted that MSCs can effectively inhibit Th17 differentiation. Duffy et al[57] demonstrated that this inhibition requires cyclooxygenase-2 induction, which is dependent on cell contact, leading to direct Th17 inhibition by PGE2. Qu et al[58] also showed that MSCs inhibit Th17 cell differentiation, and suggested that increased secretion of IL-10 may play a role. One important aspect of the immunomodulatory effect of MSCs is the recruitment and influence of Tregs[59]. MSCs can reinforce the regulatory function of CD8⁺CD28^- Treg cells by increasing expression of IL-10 and FasL[60]. They can also induce cell cycle arrest in the G0/G1 phase instead of induction of B-cell apoptosis in a soluble factor-dependent manner[61]. Furthermore, MSCs inhibit proliferation and activation of B cells by modifying the phosphorylation levels of the extracellular signal-related kinase 1/2 or p38 pathways[62]. Rafei et al[63] clarified that MSC-derived chemokine (C-C motif) ligand 2 (referred to as CCL2) can suppress secretion of immunoglobulin (referred to as Ig) in plasma cells and induce proliferation of plasmablasts, leading to IL-10-mediated blockade via inactivation of signal transducer and activator of transcription 3 (referred to as STAT3) and induction of paired box 5 in vitro. MSCs can enhance the survival and proliferation rates of CD5⁺ Bregs in an IDO-dependent manner, increasing IL-10 expression and ameliorating refractory, chronic GVHD[10]. Similarly, AMSCs reduce plasmablast formation or induce IL-10-producing CD19⁺CD24^highCD38^high B cells[64]. Park et al[65] elucidated the effect of human AMSCs on the proliferation of Bregs in an animal model of systemic lupus erythematosus.