Source and characterization of hepatic macrophages in acetaminophen-induced acute liver failure in humans

Authors

  • Charalambos Gustav Antoniades,

    Corresponding author
    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    2. Section of Hepatology, St. Mary's Hospital, Imperial College London, London, United Kingdom
    • Section of Hepatology, St. Mary's Hospital, Imperial College London, 10th Floor, QEQM Building, South Wharf Road, London W2 1NY, UK
    Search for more papers by this author
    • fax: (44)-20-7724-9369

    • These authors contributed equally to this work. Diego Vergani, Mark R. Thursz, and Julia Wendon are joint senior authors.

  • Alberto Quaglia,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
    • These authors contributed equally to this work. Diego Vergani, Mark R. Thursz, and Julia Wendon are joint senior authors.

  • Leonie S. Taams,

    1. Centre for Molecular andCellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
    • These authors contributed equally to this work. Diego Vergani, Mark R. Thursz, and Julia Wendon are joint senior authors.

  • Ragai R. Mitry,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Munther Hussain,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Robin Abeles,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Lucia A. Possamai,

    1. Section of Hepatology, St. Mary's Hospital, Imperial College London, London, United Kingdom
    Search for more papers by this author
  • Matthew Bruce,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Mark McPhail,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    2. Section of Hepatology, St. Mary's Hospital, Imperial College London, London, United Kingdom
    Search for more papers by this author
  • Christopher Starling,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Bart Wagner,

    1. Electron Microscopy Unit, Histopathology Department, Northern General Hospital, Sheffield, United Kingdom
    Search for more papers by this author
  • Adrian Barnardo,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Sabine Pomplun,

    1. Department of Histopathology, King's College Hospital, London, United Kingdom
    Search for more papers by this author
  • Georg Auzinger,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • William Bernal,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Nigel Heaton,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Diego Vergani,

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
  • Mark R. Thursz,

    1. Section of Hepatology, St. Mary's Hospital, Imperial College London, London, United Kingdom
    Search for more papers by this author
  • Julia Wendon

    1. Institute of Liver Studies and the Cellular Biology of Inflammation, King's College London, London, United Kingdom
    Search for more papers by this author
    • These authors contributed equally to this work. Diego Vergani, Mark R. Thursz, and Julia Wendon are joint senior authors.


  • Potential conflict of interest: Dr. Heaton is on the speakers' bureau of Astellas. Dr. Auzinger advises Pfizer.

  • Supported by a Medical Research Foundation/Medical Research Council Liver Transplant Grant Award. Charalambos Gustav Antoniades is supported by the National Institute of Health Research and EASL Sheila Sherlock Physician Scientist Fellowship.

Abstract

Acetaminophen-induced acute liver failure (AALF) is associated with innate immunity activation, which contributes to the severity of hepatic injury and clinical outcome. A marked increase in hepatic macrophages (h-mϕ) is observed in experimental models of AALF, but controversy exists regarding their role, implicating h-mϕ in both aggravation and resolution of liver injury. The role of h-mϕ in human AALF is virtually unexplored. We sought to investigate the role of chemokine (C-C motif) ligand 2 (CCL2) in the recruitment of circulating monocytes to the inflamed liver and to determine how the h-mϕ infiltrate and liver microenvironment may contribute to tissue repair versus inflammation in AALF. We evaluated circulating monocytes, their chemokine (C-C motif) receptor 2 (CCR2) expression, and serum CCL2 levels in patients with AALF. Cell subsets and numbers of circulation-derived (MAC387+) or resident proliferating (CD68/Ki67+) h-mϕ in hepatic immune infiltrates were determined by immunohistochemistry. Inflammatory cytokine levels were determined in whole and laser microdissected liver tissue by proteome array. In AALF, circulating monocytes were depleted, with the lowest levels observed in patients with adverse outcomes. CCL2 levels were high in AALF serum and hepatic tissue, and circulating monocyte subsets expressed CCR2, suggesting CCL2-dependent hepatic monocyte recruitment. Significant numbers of both MAC387+ and CD68+ h-mϕ were found in AALF compared with control liver tissue with a high proportion expressing the proliferation marker Ki67. Levels of CCL2, CCL3, interleukin (IL)-6, IL-10, and transforming growth factor-β1 were significantly elevated in AALF liver tissue relative to chronic liver disease controls. Conclusion: In AALF, the h-mϕ population is expanded in areas of necrosis, both through proliferation of resident cells and CCL2-dependent recruitment of circulating monocytes. The presence of h-mϕ within an anti-inflammatory/regenerative microenvironment indicates that they are implicated in resolution of inflammation/tissue repair processes during AALF. (HEPATOLOGY 2012)

Acetaminophen-induced acute liver failure (AALF) is a devastating clinical syndrome characterized by overwhelming hepatocyte death and activation of systemic inflammatory responses resulting in rapid and progressive multiple organ dysfunction and death.1-3 The uncontrolled activation of innate immune responses is central to the pathogenesis of AALF and determines the severity of acute hepatic injury and clinical outcome of AALF.1, 4 Monocytes/macrophages are key effector cells in innate immune responses and could be involved in the initiation, propagation, and resolution of hepatic inflammation during AALF. Our previous work showed that a numerical reduction and dysfunction of circulating monocytes are strongly correlated with activation of systemic anti-inflammatory responses and predicts adverse outcome in AALF.5, 6

During experimental tissue injury, expansion in macrophage numbers occurs via proliferation of the resident population that characterizes the later phases of inflammatory response when tissue repair and regenerative responses prevail.7-10 In addition, circulating monocytes may be recruited to inflamed tissue and differentiate into macrophages.11 A marked increase in hepatic macrophages (h-mϕ) is consistently observed in rodent models of acetaminophen-induced liver injury (APAP), but controversy exists regarding their role. Some studies have demonstrated that h-mϕ contribute to aggravation of liver injury, whereas others suggest a role in resolution of inflammation and tissue repair processes through recruitment of bone marrow–derived circulating monocytes.12-18 Similar to other inflammatory models, these divergent findings may be due to macrophages acquiring distinct and functionally opposing roles that are influenced by the nature, time course, and inflammatory microenvironment following a given acute hepatic insult.19-23 The role of monocytes/macrophages in human AALF is virtually unexplored.

Chemokine (C-C motif) ligand 2 (CCL2), also known as monocyte chemoattractant protein-1, acts on the chemokine (C-C motif) receptor 2 (CCR2), which plays a role in the recruitment of monocytes, natural killer cells, and T cells in a wide range of inflammatory conditions.24 CCL2 has been shown to be raised in patients with non–acetaminophen-induced acute liver failure,25 and in experimental models is a pivotal mediator promoting the mobilization of monocytes from the bone marrow into the circulation and their subsequent recruitment to areas of hepatic necrosis.13, 15, 26-28

In this study, we sought to (1) determine the relative contribution of both the resident and bone marrow–derived macrophages to the h-mϕ population and (2) analyze the liver inflammatory microenvironment and the h-mϕ population within areas of hepatic necrosis and gain insight into their functional capabilities during AALF.

Abbreviations

AALF, acetaminophen-induced acute liver failure; AALF-D, AALF patients who died; AALF-O, AALF patients who underwent transplantation; AALF-S, AALF patients who survived with medical management; APAP, acetaminophen-induced liver injury; CCL2, chemokine (C-C motif) ligand 2; CCL3, chemokine (C-C motif) ligand 3; CCR2, chemokine (C-C motif) receptor 2; CLD, chronic liver disease; HLA-DR, human leukocyte antigen DR; h-mϕ, hepatic macrophages; hpf, high-powered fields; IL, interleukin; INR, international normalized ratio; IQR, interquartile range; KC, Kupffer cell; OLT, orthotopic liver transplantation; TGF-β1, transforming growth factor-β1; TNF-α, tumor necrosis factor-α.

Patients and Methods

Patient Recruitment.

Thirty-eight consecutive patients admitted to the liver intensive care unit were recruited. AALF patients were divided into those who died (n = 8), those who received a liver graft (n = 14), and those who survived with medical management (n = 16). Fifteen consecutive in-patients with chronic liver disease (CLD) undergoing transplantation assessment and 10 healthy volunteers served as pathological and healthy controls, respectively. Exclusion criteria were age <18 years or >65 years, neoplasia, previous or concurrent immunosuppressive therapy, and clinical or microbiological evidence of sepsis on admission. AALF patients were identified for emergency transplantation according to King's College Hospital criteria. The study was approved by the King's College Hospital Ethics Committee (LREC 04LG03). Consent was obtained by the patients' nominated next of kin if they were unable to give informed consent themselves.

Clinical, Hematological, and Biochemical Parameters.

White cell count (monocyte, neutrophil, lymphocyte [count ×109/L]) was determined in AALF patients from day 1-4 following admission to the liver intensive care unit using a hematological analyzer (Siemens-Advia 2120 Berks, UK). International normalized ratio (INR), liver and renal function tests, lactate, and clinical and physiological variables were prospectively entered into a database.

Serum Cytokine Assays.

Blood was collected at the same time as initial blood sampling and was centrifuged at 2,000g for 10 minutes at 4°C, and the serum obtained was stored at −80°C. Levels of CCL2, tumor necrosis factor-α (TNF-α), interleukin (IL)-6, and IL-10 were measured by enzyme-linked immunosorbent assay (R&D Systems Europe, Abingdon, UK).

Flow Cytometry.

Monoclonal antibodies against CD14, CD16, and CCR2 (BD Biosciences, Oxford, UK) were used to determine CCR2 expression on monocyte subsets from peripheral blood mononuclear cells from healthy controls and AALF patients (blood obtained within 24-48 hours of admission) (Supporting Information, section 1.1).

Bone Marrow Sections.

Bone marrow trephine biopsies from three AALF patients obtained prior to transplantation were examined as part of a further ethically approved study evaluating the role of bone marrow progenitors in acute liver failure.

Liver Samples.

Explanted liver tissue was obtained in 10 patients undergoing orthotopic liver transplantation (OLT) due to AALF. Tissue samples were taken for diagnostic histological examination and were formalin-fixed and paraffin-embedded. Snap-frozen liver sections were concomitantly obtained and stored in liquid nitrogen. Liver tissue obtained during the resection of hepatic malignancies (n = 5), from patients transplanted for hepatitis C cirrhosis (n = 3) and chronic cholangiopathy (n = 2) and from biopsies of three healthy living related donors served as pathological control tissue and normal control liver tissue.

Immunohistochemistry.

The immune cell infiltrate in liver tissue was studied using single-stain immunohistochemistry from formalin-fixed, paraffin-embedded tissue for CD3-, CD68-, MAC387-, CD56-, and FOXP3 cell expression as described.29 The number of human leukocyte antigen DR (HLA-DR)+ macrophages (CD68+/HLA-DR+), proliferating macrophages (MAC387/Ki67+, CD68+/Ki67+), biliary epithelial cells (CK19+/Ki67+) and hepatocytes (HEP-PAR1+/Ki67+) were studied using double-staining immunohistochemistry. Presence of monocyte/macrophage progenitors was assessed in bone marrow trephine biopsies using single-staining for CD68 and lysozyme (Supporting Information, section 1.2).

Immunostains were analyzed by a liver histopathologist (A. Q.) who was blinded to the clinical data. A cell count was performed using an eyepiece graticule (Datasights limited, Middlesex, UK) as described by Going30 (Supporting Information, section 1.3).

Electron Microscopy.

Transmission electron microscopy was performed on liver tissue from three AALF explants as described in the Supporting Information (section 1.4).

Inflammatory Protein Array of Whole and Laser Capture–Microdissected Liver Samples.

Areas of necrotic and viable parenchyma were obtained from snap-frozen liver tissue samples using laser capture microdissection (Supporting Information, section 1.5). Tissue lysate was prepared using protein lysate buffer according to the protocol developed by Mustafa et al.31 (supplementary section 1.6). Protein array of tissue lysate was performed by Aushon Biosystems (Billerica, Boston, MA;USA) as described in supplementary section 1.7. Results are expressed as pg/mL.

Statistical Analysis.

To identify differences between groups, nonparametric analysis was used (Mann-Whitney U test, Kruskal-Wallis test, Wilcoxon rank test). Correlations were analyzed using Spearman's rank test. Results are expressed as the median and interquartile range (IQR). Changes in white blood cell counts were analyzed using one-way analysis of variance.

Results

Patient Characteristics.

There was no significant difference in median ages of AALF patients (34 years [IQR, 27-43]) when compared with healthy controls (33.5 years [IQR, 29-40]; P = 0.8), whereas CLD patients were significantly older (50.0 years [IQR, 44.61]; P < 0.05). The mean number of circulating monocytes was significantly reduced in all AALF patients when compared with CLD patients (0.42 × 109/L [0.53] versus 0.63 × 109/L [0.29]; P = 0.002).

AALF Subgroup Outcome Analysis.

Table 1 shows the clinical and biochemical indices and circulating inflammatory cytokine levels in the AALF patients categorized according to clinical outcome. AALF patients were divided into those who survived with conservative medical management (AALF-S), underwent emergency OLT (AALF-O), and died without undergoing OLT (AALF-D). Compared with the AALF-S group, AALF-O and AALF-D patients had significantly lower arterial pH and significantly greater derangement of physiology as evidenced by higher INR, arterial blood lactate, level of encephalopathy, vasopressor and hemofiltration requirements, MELD score, and circulating levels of proinflammatory (TNF-α, IL-6) and anti-inflammatory (IL-10) cytokines. As has been described, serum levels of TNF-α, IL-6, and IL-10 were significantly higher in AALF patients compared with CLD patients and healthy controls (data not shown).5

Table 1. Comparison of Indices of Liver Dysfunction
ParameterAALF-S (n = 16)AALF-O (n = 14)AALF-D (n = 8)P*
  • Inflammatory cytokine expression in AALF patients was subdivided according to whether the patient survived (AALF-S), required OLT (AALF-O), or died (AALF-D). All clinical, hematological, biochemical, and physiological parameters were measured at time of sampling for circulating cytokine determination. Data are expressed as the median (IQR).

  • Abbreviations: MELD, model of end-stage liver disease; NS, not significant.

  • *

    Statistically significant differences between the AALF-S group and the AALF-O and/or AALF-D patient groups.

  • Hemofiltration: 0 = no hemofiltration, 1 = continuous veno-venous hemofiltration.

  • 0 = none, 1 = <0.1 μg/kg/minute, 2 = >0.1 μg/kg/minute.

Total white cell count (×109/L)8.1 (5.4-11.2)10.1 (6.4-13.5)11.5 (7.1-17.4)NS
Bilirubin (μmol/L)89 (49-129.5)67 (54-87)103 (39-154)NS
INR2.87 (2.04-4.3)9.3 (6.9-10.2)11.4 (4.1-13.5)<0.0001
Aspartate aminotransferase (IU/L)1,463 (460-4,753)7,413 (1,453-10,332)5,266 (735-8,089)NS
Lactate (mmol/L)1.9 (1.35-3.1)5.8 (3.3-13)6.8 (3-10)<0.001
pH7.4 (7.40-7.44)7.23 (7.15-7.317.10 (7.1-7.4)<0.001
Hemofiltration1 (0-1)1 (1-1)1 (1-1)<0.01
Encephalopathy2 (0-3)3 (3-4)3.5 (3-4)<0.001
Norepinephrine dosage0 (0-1)2 (1.25-2)2 (0.25-2)<0.001
MELD33 (31-35)50 (46-52)52 (45-55)<0.0001
TNF-α (pg/mL)0 (0-80)42 (21-283)372 (0-1,355)<0.001
IL-6 (pg/mL)52 (9.7-173)425 (169-585)600 (188-1,450)<0.01
IL-10 (pg/mL)46 (20.5-126)365.5 (138-603)324.5 (210-618)<0.001

Elevated Circulating Levels of CCL2 Are Associated with Monocytopenia and Reflect Disease Severity.

The number of circulating monocytes was significantly reduced in AALF-D (median, 0.04 × 109/L [range, 0.01-0.22]) and AALF-O (median, 0.145 × 109/L [range, 0.0-1.07]) compared with AALF-S (median, 0.54 × 109/L [range, 0.1-1.05]; both P = 0.0004) at 24 hours following admission. Numbers of circulating monocytes remained significantly reduced in AALF-D and AALF-O patients up to day 4 following admission (Fig. 1A). In contrast, no significant reduction in neutrophils was observed in AALF-D and AALF-O patients (Fig. 1B). Lymphocytes were also not reduced except on day 3 following admission (Fig. 1C).

Figure 1.

Circulating white blood cell subsets and CCR-2 expression in acetaminophen-induced ALF (AALF). Median (range) counts of circulating monocytes (A), neutrophils (B), and lymphocytes (C) were obtained by hematological analysis on day 1-4 following admission to the liver intensive care unit (D1-D4) in AALF-D patients (black bars, n = 8), AALF-O patients (gray bars, n = 14), or AALF-S patients (white bars, n = 16). §Cell count ×109/L. CCR2 expression on CD14high/CD16- (D), CD14high/CD16+ (E), and CD14low/CD16+ monocytes (F) in peripheral blood mononuclear cells from 20 AALF patients (AALF-D/AALF-O, n = 9; AALF-S, n = 11) and 20 healthy controls (HC) was assessed by flow cytometry. *Statistically significant differences (one-way analysis of variance; P < 0.01) between AALF groups. MFI, mean fluorescence intensity; ns, not significant.

Circulating levels of CCL2 were markedly elevated in all AALF patients (n = 38) when compared with healthy controls (median, 3,995 pg/mL [IQR, 1,355-11,620] versus median, 98.5 pg/mL [IQR, 92-124]; P < 0.0001). When subdivided, CCL2 levels were significantly higher in AALF-D patients (median, 13,000 pg/mL [IQR, 4,446-22,060], n = 8) and AALF-O patients (median, 6,925 pg/mL [IQR, 3,959-13,270], n = 14) compared with AALF-S patients (median, 945 pg/mL [IQR, 370-2,234], n = 16) (P < 0.0005 for both) (Supporting Information, Results section; Supporting Fig. 1).

In the AALF patient group as a whole, CCL2 correlated negatively with circulating monocyte count (r = −0.76; P < 0.001) and arterial pH (r = −0.61; P < 0.001), and positively correlated with INR (r = 0.71; P < 0.001), arterial blood lactate (r = 0.70; P < 0.001), aspartate aminotransferase (r = 0.56; P < 0.01), degree of encephalopathy (r = 0.55; P < 0.01), and vasopressor dose requirements (r = 0.46; P = 0.01).

We assessed the expression of the CCL2 receptor CCR2 on the three major peripheral blood monocyte subsets (CD14high/CD16−, CD14high/CD16+, and CD14low/CD16+ cells) (Supporting Fig. 2) in a further 20 AALF patients (AALF-D/AALF-O [n = 9], AALF-S [n = 11]) and 20 healthy controls. CCR2 was expressed on all three monocyte subsets, with significantly elevated mean fluorescence intensity in the CD14high/CD16+ monocytes in AALF patients compared with healthy controls (Fig. 1D-F). There was no significant difference in CCR2 expression on the CD14high/CD16+ subset between the AALF-D/AALF-O and AALF-S subgroups (median, 4,918 [IQR, 954-7,823] versus median, 3,338 [IQR, 1,365-4,716]; P = 0.40).

Presence of Monocyte Progenitors in Bone Marrow.

To investigate whether the peripheral monocytopenia could be due to a reduced production by the bone marrow, the bone marrow of three AALF patients who had undergone bone marrow trephine biopsy prior to transplantation was examined. Reactive trilineage hematopoiesis was observed with the myeloid islands demonstrating strong lysozyme expression, indicative of progenitor cell differentiation toward a monocytic cell lineage, and an increase in the number of mature (CD68+) stromal foamy macrophages (Fig. 2A-D). These data suggest that the depletion of circulating monocytes is not attributable to lack of bone marrow–derived monocyte precursors.

Figure 2.

Bone marrow progenitors in AALF. Hematoxylin and eosin–stained sections showing (A) reactive trilineage hematopoiesis, (B) myeloid islands, (C) lysozyme-positive cells within myeloid islands indicating differentiation along a monocytic pathway, and (D) CD68-positive cells indicating the presence of stromal foamy macrophages (magnification ×400).

Liver Tissue Immune Cell Infiltrate and Inflammatory Microenvironment.

The pattern of liver injury in all cases was typical of APAP (Supporting Information, Results section; Supporting Fig. 3). H-mϕ were the predominant inflammatory cell population in areas of necrosis, whereas lymphocytes and neutrophils were concentrated in the portal tracts (Fig. 3A-E). The number of cells per 10 high-powered fields (hpf) for the different immune cell populations is shown in Fig. 3F. Higher numbers of MPO+, CD3+, and CD56+ cells were detected in AALF compared with pathological control liver tissue (Supporting Information, Results section; Supporting Fig. 4).

Figure 3.

Characterization of the immune cell infiltrate in explanted liver from an AALF patient. (A-E) Representative patterns of expression of (A) CD68+ macrophages, (B) MPO+ granulocytes, (C) CD3+ lymphocytes, (D) CD56+ lymphocytes, and (e) FOXP3+ lymphocytes (magnification ×40). (F) Quantification (median [IQR]) of the immune cell infiltrate within areas of necrosis and portal tracts from explanted liver tissue from 10 AALF patients. For each section, 10 randomly chosen portal tracts or areas of necrosis were assessed at high magnification (×400), and the cumulative number of positive cells in 10 hpf was recorded. ns, not significant.

H-mϕ were abundant and concentrated within areas of centrilobular necrosis compared with pathological control liver tissue (median, 530 cells/10 hpf [IQR, 480-725] versus median, 330 cells/10 hpf [IQR, 240-442]; P = 0.01; n = 10 AALF patients, n = 6 pathological controls) (Fig. 4A,D). Immunohistochemistry for MAC387 was used to identify infiltrating macrophages (Fig. 4B,E).27, 32-35 The number of MAC387+ cells was significantly elevated in AALF compared with pathological control liver tissue (median, 95 cells/10 hpf [IQR, 50-182] versus median, 20 cells/10 hpf [IQR, 20-35]; P = 0.001; n = 8 AALF patients, n = 8 pathological controls).

Figure 4.

Characterization of recruited and resident hepatic macrophages (h-mϕ) in AALF liver tissue. (A) Representative example of CD68+ h-mϕ in normal liver (left panel, magnification ×40), and explanted AALF liver tissue (right panel, magnification ×40). (B) Newly recruited infiltrating MAC387+ h-mϕ in normal liver (left panel, magnification ×100) and explanted AALF liver tissue (right panel, magnification ×100). (C) Proliferating Ki67+CD68+ h-mϕ in normal liver (left panel, magnification ×100) and explanted AALF liver tissue (right panel, magnification ×400) where a proliferating h-mϕ (arrow) and proliferating hepatocyte (arrowhead) are shown; (D-F) Cumulative number of CD68+ and MAC387+ per 10 hpf and proportion of Ki67-positive CD68+ macrophages (per 10 hpf) in areas of hepatic necrosis in AALF explanted liver specimens (n = 10) and in pathological control liver tissue specimens (n = 6).

The percentage of resident proliferating h-mϕ (defined by coexpression of CD68 and Ki67) (Fig. 4C,F) was significantly increased within areas of hepatic necrosis compared with equivalent anatomical locations in pathological controls (median, 19.5% [IQR, 13%-25%] versus median, 0% [IQR, 0%-1.5%]; P = 0.003; n = 10 AALF patients, n = 6 pathological controls), whereas the median percentage of proliferating infiltrating (MAC387/Ki67+) h-mϕ was 1% (IQR, 0.1%-2.5%) (Supporting Information, Results section; Supporting Fig. 5). A trend toward a higher number of resident proliferating h-mϕ and a lower number of infiltrating h-mϕ was observed in patients who underwent transplantation later in their clinical course compared with those who received a graft earlier (Supporting Information, Results section; Supporting Table 1).

In all AALF cases in which the numbers of proliferating h-mϕ were assessed, evidence of hepatocellular (HEP-PAR1/Ki67+) and ductular epithelial (CK19/Ki67+) regenerative activity was concomitantly confirmed (representative images in the Supporting Information, Results section; Supporting Fig. 6).

Immunohistochemical and Ultrastructural Analysis of h-mϕ.

To investigate the phenotype of the h-mϕ population, we assessed HLA-DR expression. Single immunostaining showed that the pattern of distribution of HLA-DR+ cells was similar to that of CD68+ macrophages—that is, largely confined to necrotic areas (Fig. 5A,B). Double-immunostaining for CD68 and HLA-DR (Fig. 5C,D) revealed that CD68/HLA-DR+ macrophages were particularly prominent at the periphery of the areas of centrilobular necrosis. In central areas of necrosis and perivenular regions, most CD68-expressing macrophages did not coexpress the HLA-DR molecule.

Figure 5.

Characterization of hepatic macrophages (h-mϕ) in AALF. (A,B) Representative single immunostaining demonstrating similar distribution of (A) CD68- and (B) HLA-DR–expressing cells in areas of centrilobular necrosis from explanted AALF liver tissue (magnification ×40 for both). (C,D) Double-immunostaining for HLA-DR and CD68+ h-mϕ demonstrating CD68 (brown) and HLA-DR (pink) colocalization within an area of centrilobular necrosis. Colocalization and HLA-DR expression is prominent in a rim around the area of centrilobular necrosis (black arrows) adjacent to the portal tract (PT). Within the central areas of necrosis (N), there is less colocalization of HLA-DR and CD68 (C, magnification ×200; D, magnification ×400 in an area of centrilobular necrosis). (E,F) Transmission electron microscopy of explanted AALF tissue showing a single h-mϕ (E) within an area of necrosis surrounded by extracellular collagenous filaments (labeled A) containing phagocytosed lipidic and necrotic debris (labeled B and C) and (F) a h-mϕ (labeled A) with cytoplasmic extensions (labeled B) in the process of phagocytosis of cellular debris (labeled C).

Electron microscopy performed on liver tissue of three AALF patients revealed that portal and periportal macrophages were markedly swollen and contained numerous lysosomes and lipolysosomes, whereas those within necrotic/perivenular areas contained large amounts of phagocytosed cellular/extracellular debris (Fig. 5E,F).

Inflammatory Microenvironment in AALF.

The hepatic inflammatory microenvironment was tested using protein microarray analysis in AALF patients (n = 10) and in pathological controls (n = 8). The whole liver tissue concentrations of anti-inflammatory/tissue repair cytokines were significantly elevated in AALF compared with pathological control liver tissue: IL-6 (median, 158 pg/mL [IQR, 66-284] versus median, 24 pg/mL [IQR, 12.5-108]; P = 0.02), IL-10 (median, 2.0 pg/mL [1.1-2.6] versus median, 0.6 pg/mL [IQR, 0.4-1.8]; P = 0.03) and transforming growth factor-β1 (TGF-β1) (median, 3,364 pg/mL [IQR, 2,416-3,485] versus 681 pg/mL [IQR, 162-1,575]; P < 0.0001). Concentrations of CCL2 (median, 8,383 pg/mL [IQR, 5,747-45,647] versus median, 329 pg/mL [IQR, 132-3,678]; P = 0.001) and CCL3 (median, 755 pg/mL [IQR, 272-1,997] versus median, 77 pg/mL [IQR, 77-175]; P = 0.0009) followed the same pattern. No significant differences in proinflammatory cytokines (IL-1β, IL-4, IL-12p70, IL-17, IL-23, interferon-γ, and TNF-α) were detected between AALF and pathological control liver tissue (Supporting Information, Results section; Supporting Table 2).

The concentrations of TNF-α, IL-6, IL-10, CCL2, and CCL3 within necrotic areas in all AALF cases were higher than in the nonnecrotic areas (Fig. 6), whereas levels of TGF-β1 (median, 292 [IQR, 101-420] versus median, 211 [IQR, 63-514]; P = 1.0) and IL-18 (median, 111 [IQR, 65-229] versus median, 84 [IQR, 35-220]; P = 0.4) were similar. Levels of IL-12p70 (median, <0.2 pg/mL [IQR, 0-0]) and IL-1β (median, 0.25 pg/mL [IQR, 0-0.7]) were barely detectable/undetectable within necrotic areas and those of IL-8 were consistently lower (median, 43 [IQR, 38-74] versus median, 432 [IQR, 189-903]; P = 0.001) in necrotic than nonnecrotic areas (Fig. 6D).

Figure 6.

Comparison of key inflammatory cytokines and monocyte chemoattractant levels within necrotic and nonnecrotic areas of the hepatic lobule in AALF tissue. Necrotic and nonnecrotic areas were isolated by laser capture microdissection from six to nine separate AALF liver tissue sections and IL-6 (A), TNF-α (B), IL-10 (C), IL-8 (D), CCL2 (E), and CCL3 (F) concentrations determined by protein array.

Regional Concentrations of Inflammatory Cytokines.

We measured regional levels of TNF-α and IL-10 in five AALF patients at the time of OLT. In four out of five patients sampled, a transhepatic gradient was demonstrated where higher concentrations of IL-10 (715 versus 581 pg/mL; 903 versus 235 pg/mL; 600 versus 554 pg/mL; 349 versus 201 pg/mL; 1,054 versus 1,081 pg/mL) were detected in the hepatic vein than in the portal vein, whereas no discernible transhepatic TNF-α concentration gradient was observed in all five AALF patients sampled (93 versus 97 pg/mL; 25 versus 98 pg/mL; 20 versus 19 pg/mL; 27 versus 45 pg/mL; 24 versus 23 pg/mL).

Discussion

We demonstrate a marked expansion of the h-mϕ population in AALF. Our data suggest that this increased macrophage infiltrate is derived both from circulating monocytes and from the proliferation of the resident Kupffer cell (KC) population. In an APAP rodent model, Holt et al.14 identified a subpopulation of hepatic macrophages, distinct from resident KCs and derived from circulating monocytes, that infiltrated liver tissue within 12 hours and persisted until the resolution of hepatic injury up to 5 days later. Our data are suggestive of a similar process occurring in AALF. This hypothesis is supported by the presence of macrophages in areas of liver necrosis, positive for MAC387, an intracytoplasmic protein maximally expressed at an early stage of monocyte/macrophage differentiation that distinguishes circulation-derived newly infiltrating macrophages from the resident KC population.27, 32, 35, 36 We found MAC387 expression to be highest in patients transplanted sooner following acetaminophen ingestion, which could suggest that the influx of monocyte-derived macrophages to inflammatory foci occurs in the earlier phases of liver injury.14, 27

Experimental models demonstrate that the interaction between CCL2 and its receptor CCR2 promotes efflux of CCR2-expressing monocytes from the bone marrow into the circulation.24, 37, 38 Our data demonstrate that despite reactive monocyte progenitor hematopoiesis and markedly elevated circulating CCL2 levels, there is a profound reduction in the absolute number of circulating monocytes that is proportional to the severity of acute liver injury (Figs. 1 and 2). This suggests that circulating monocytes are being recruited to the inflamed liver at a rate that exceeds bone marrow production resulting in a reduction in their numbers in the circulation. However, our data do not exclude the possibility that the depletion of circulating monocytes may also be attributed to apoptosis39 or recruitment to other tissues.

Consistent with the previously published experimental APAP models12-14, 18 and human studies of AALF,25, 27 our data support the role of CCL2 in recruitment of circulating monocytes to the liver during AALF. In Fig. 6, we show that necrotic liver tissue may act as a source of CCL2 secretion, as evidenced by the significantly elevated levels of monocyte chemoattractants (CCL2, CCL3) in whole liver tissue, the chemokine gradient from necrotic to nonnecrotic tissue, and elevations in circulating levels of this chemoattractant. We also found that all three circulating monocyte subsets express CCR2, suggesting that all three populations could be recruited to the inflamed liver. Our study, however, does not exclude the involvement of other chemokines in recruiting monocytes to the liver, and further studies are warranted to assess this.

We observed marked proliferation of the resident KC population within areas of necrosis (Fig. 4); this finding is in contrast to monocyte-derived infiltrating macrophages, where less than 1% were proliferating. Previous reports support the existence of two macrophage populations with distinct functional capabilities and self-renewal characteristics during steady state and inflammation. One population derived from circulating monocytes with little self-renewal potential is rapidly recruited to inflammatory sites, giving rise to the classical inflammatory macrophages that cause tissue destruction and necrosis.40 There is a second resident population with self-renewal capabilities that characterize later phases of inflammatory insult when tissue repair and regenerative responses prevail.7-10 Recently, the anti-inflammatory cytokine IL-4 has been shown to be a pivotal driver of macrophage self-renewal and tissue repair during experimental tissue injury.9 Although our data do not reveal elevated concentrations of IL-4 within the liver microenvironment, further work is required to delineate the mechanisms that trigger h-mϕ proliferation and their role during AALF.

Based on the data presented here, we hypothesize that the macrophage infiltrate following AALF is derived from an early wave of bone marrow–derived circulating monocytes followed by an expansion of the resident proliferating KC population that is implicated in the resolution of inflammation and tissue repair processes.

The protein array analysis reveals an inflammatory microenvironment favoring tissue repair processes in AALF at the time of transplantation. In whole liver tissue, levels of IL-6, IL-10, and TGF-β1 are elevated in AALF compared with pathological controls, whereas concentrations of proinflammatory mediators (IL-1β, IL-12, IL-17, TNF-α, interferon-γ) remain similar. Higher concentrations of monocyte chemoattractants (CCL2, CCL3) and immunoregulatory cytokines (TNF-α, IL-6, IL-10) are detected in necrotic compared with nonnecrotic areas. These findings concur with data from experimental liver injury models, where CCL2 skews the inflammatory microenvironment to augment levels of anti-inflammatory/hepatoprotective mediators (IL-6, IL-10, TGF-β1).13, 26, 28 In other inflammatory models, CCL2 induces an IL-10–skewed cytokine profile in murine polymicrobial sepsis and contributes to the recruitment of IL-10–producing, monocyte-derived macrophages during experimental colitis.41-44 Therefore, our data demonstrating increased expression of CCR2 on CD14+CD16+ circulating monocytes may indicate that CCL2 possesses immunoregulatory capabilities in AALF through recruitment of different circulating monocyte subsets.45 Further studies evaluating monocyte subsets at different stages of liver injury are required to address this question.

Our data indicate that h-mϕ represent the predominant cell population in the inflamed AALF liver and are avidly proliferating within areas of necrosis, a finding that is associated with hepatic regenerative responses in experimental liver models.7 They express markers associated with enhanced scavenger functions (CD68+HLA-DR+),46-48 preferentially expressed at the resolution stages of experimental liver injury,12, 28 and contain phagocytosed cellular and extracellular debris. Akin to other inflammatory conditions, phagocytosis may be the microenvironmental switch that triggers h-mϕ to secrete immunoregulatory mediators (CCL2, TNF-α, IL-6, IL-10) that are present at higher concentrations within areas of hepatic necrosis.49-51 Equally, the proportion of h-mϕ not expressing HLA-DR in central areas of necrosis could indicate that they are functionally modulated by their microenvironment. Functional and phenotypic analyses of freshly isolated h-mϕ are warranted to explore these observations.

These findings also pose further questions as to whether these intra-hepatic events impact on circulating monocyte phenotype and function. It is possible that “spill-over” of immunoregulatory cytokines, such as IL-10, from the perivenular areas, into the circulation, could offer a mechanistic explanation for the profound depletion of HLA-DR expressing monocytes previously reported in AALF and the increased predisposition to sepsis encountered in this condition.5 It is important to highlight that we were only able to examine h-mϕ at a single time point and in a cohort of patients with severe and advanced acute liver injury and therefore the data cannot be assumed to represent events at earlier time points in the evolution of AALF.

Figure 7 summarizes the postulated mechanisms that result in the expansion and functional differentiation of h-mϕ during AALF. Taken together, the increase in both circulation-derived and resident h-mϕ within areas of necrosis and the preferential expression of anti-inflammatory/regenerative cytokines such as IL-6, IL-10, and TGF-β1 suggests that h-mϕ are modulated by, or themselves alter the inflammatory microenvironment in favour of tissue repair processes. Functional and phenotypic analysis of freshly extracted h-mϕ will establish their role in resolution of inflammation and tissue repair processes in AALF.

Figure 7.

Postulated role of monocytes and h-mϕ in human AALF. Acetaminophen leads to centrilobular hepatic necrosis (1) that triggers the release of hepatically derived CCL2, leading to high circulating levels (2-3). CCL2 stimulates the proliferation, differentiation, and egress of monocytes from bone marrow (4). Despite reactive bone marrow production and egress of monocytes, their avid recruitment following overwhelming liver injury results in a net reduction in circulating monocytes resulting in monocytopenia (5). Influx of circulatory-derived (MAC387+) infiltrating macrophages and the proliferation of the resident KC population contribute to the expanded macrophage population observed within areas of hepatic necrosis (6-7). Following exposure to local microenvironmental cues and differentiation, hepatic macrophages are functionally modulated to favor the production of anti-inflammatory cytokines, phagocytose necrotic debris ultimately leading to the initiation of resolution of inflammation and tissue repair processes (8-10). However, the observed spill-over of these immunoregulatory cytokines into the systemic circulation may contribute to the observed monocyte deactivation that renders patients susceptible to sepsis (11-12).

Acknowledgements

We gratefully acknowledge the Imperial National Institute of Health Research Biomedical Research Centre for infrastructure support and the King's College Hospital Research & Development Department and Institute of Liver Studies Histopathology Department for ongoing support. Charalambos Gustav Antoniades personally acknowledges Dr. G. Antoniades for help and support.

Ancillary