Article Text

PDF

Original article
Inhibition of glutamine synthetase in monocytes from patients with acute-on-chronic liver failure resuscitates their antibacterial and inflammatory capacity
  1. Hannelie Korf1,
  2. Johannie du Plessis1,2,
  3. Jos van Pelt3,
  4. Sofie De Groote1,
  5. David Cassiman1,4,
  6. Len Verbeke1,4,
  7. Bart Ghesquière5,
  8. Sarah-Maria Fendt6,7,
  9. Matthew J Bird1,5,
  10. Ali Talebi8,
  11. Matthias Van Haele9,
  12. Rita Feio-Azevedo1,
  13. Lore Meelberghs1,
  14. Tania Roskams9,
  15. Rajeshwar P Mookerjee10,
  16. Gautam Mehta10,
  17. Rajiv Jalan10,
  18. Thierry Gustot11,
  19. Wim Laleman1,4,
  20. Frederik Nevens1,4,
  21. Schalk Willem van der Merwe1,4
  1. 1Department of Chronic Diseases, Metabolism and Ageing (CHROMETA), KU Leuven, Leuven, Belgium
  2. 2Department of Immunology, University of Pretoria, Pretoria, South Africa
  3. 3Department of Oncology, KU Leuven, and Leuven Cancer Institute (LKI), Leuven, Belgium
  4. 4Department of Gastroenterology and Hepatology, UZ Leuven, Leuven, Belgium
  5. 5Metabolomics Expertise Centrum, VIB-KU Leuven Center for Cancer Biology, KU Leuven, Leuven, Belgium
  6. 6Laboratory of Cellular Metabolism and Metabolic Regulation, VIB-KU Leuven Center for Cancer Biology, KU Leuven, Leuven, Belgium
  7. 7Department of Oncology, Laboratory of Cellular Metabolism and Metabolic Regulation, KU Leuven and Leuven Cancer Institute (LKI), Leuven, Belgium
  8. 8Department of Oncology, Laboratory of Lipid Metabolism and Cancer, KU Leuven and Leuven Cancer Institute (LKI), KU Leuven, Leuven, Belgium
  9. 9Department of Imaging and Pathology, Translational Cell and Tissue Research, KU Leuven and University Hospitals Leuven, Leuven, Belgium
  10. 10Liver Failure Group, Institute for Liver Disease Health, University College London, London, UK
  11. 11Department of Gastroenterology and Hepato-Pancreatology, Hôpital Erasme, Université Libre de Bruxelles (ULB), Brussels, Belgium
  1. Correspondence to Hannelie Korf; hannelie.korf{at}kuleuven.be and Professor Schalk Willem van der Merwe, Laboratory of Hepatology, Department of Chronic Diseases, Metabolism and Ageing (CHROMETA), KU Leuven, Leuven 3000, Belgium; schalk.vandermerwe{at}uzleuven.be

Abstract

Objective Acute-on-chronic liver failure (ACLF) is associated with dysfunctional circulating monocytes whereby patients become highly susceptible to bacterial infections. Here, we identify the pathways underlying monocyte dysfunction in ACLF and we investigate whether metabolic rewiring reinstates their phagocytic and inflammatory capacity.

Design Following phenotypic characterisation, we performed RNA sequencing on CD14+CD16 monocytes from patients with ACLF and decompensated alcoholic cirrhosis. Additionally, an in vitro model mimicking ACLF patient-derived features was implemented to investigate the efficacy of metabolic regulators on monocyte function.

Results Monocytes from patients with ACLF featured elevated frequencies of interleukin (IL)-10-producing cells, reduced human leucocyte antigen DR isotype (HLA-DR) expression and impaired phagocytic and oxidative burst capacity. Transcriptional profiling of isolated CD14+CD16 monocytes in ACLF revealed upregulation of an array of immunosuppressive parameters and compromised antibacterial and antigen presentation machinery. In contrast, monocytes in decompensated cirrhosis showed intact capacity to respond to inflammatory triggers. Culturing healthy monocytes in ACLF plasma mimicked the immunosuppressive characteristics observed in patients, inducing a blunted phagocytic response and metabolic program associated with a tolerant state. Metabolic rewiring of the cells using a pharmacological inhibitor of glutamine synthetase, partially restored the phagocytic and inflammatory capacity of in vitro generated- as well as ACLF patient-derived monocytes. Highlighting its biological relevance, the glutamine synthetase/glutaminase ratio of ACLF patient-derived monocytes positively correlated with disease severity scores.

Conclusion In ACLF, monocytes feature a distinct transcriptional profile, polarised towards an immunotolerant state and altered metabolism. We demonstrated that metabolic rewiring of ACLF monocytes partially revives their function, opening up new options for therapeutic targeting in these patients.

  • acute liver failure
  • alcoholic liver disease
  • immunology in hepatology
  • macrophages
  • bacterial infection

Statistics from Altmetric.com

Significance of this study

What is already known on this subject?

  • Monocyte dysfunction during acute-on-chronic liver failure (ACLF) is a well-described phenomenon.

  • Singular factors such as MerTK or prostaglandin E2 have been implicated as potential mechanisms responsible for the suppressive status of monocyte during ACLF syndrome; however, the overarching pathways that drive and sustain these disease-associated defects are not fully understood.

What are the new findings?

  • Extensive characterisation of monocytes from patients with ACLF support a transcriptional, functional and metabolic switch towards an immunotolerant state.

  • The transcriptional signature of ACLF patient-derived monocytes was distinct from that obtained from decompensated cirrhosis patients.

  • Feeding glutamine into the tricarboxylic acid cycle by using a pharmacological inhibitor of glutamine synthetase (GLUL), restored the phagocytic and inflammatory capacity of monocytes from patients with ACLF.

  • Underscoring the biological relevance of this finding, we detected a positive correlation between GLUL/glutaminase ratio and the  model for end-stage liver disease (MELD) severity scores.

Significance of this study

How might it impact on clinical practice in the foreseeable future?

  • This study demonstrates that pharmacological regulation of metabolic programs partially restored the dysregulated monocyte function in ACLF.

  • These results may open new avenues for the development of therapeutic strategies to restore monocyte function in ACLF.

Introduction

Cirrhosis is the end result of chronic liver disease where persistent inflammation and stellate cell activation leads to collagen deposition, fibrosis and the development of portal hypertension.1 Initially, the disease course may be asymptomatic, but if the insult persists, it may progress to decompensated cirrhosis with development of ascites, variceal bleeding and encephalopathy. However, at any moment during the disease course a precipitating event can trigger rapid deterioration of liver function and organ failure, leading to acute-on-chronic liver failure (ACLF) with high mortality rates of up to 70%.2–6 Importantly, ACLF is often precipitated by bacterial infections, which in turn may initiate a cascade of events resulting in multiple organ failure, irreversible septic shock and death.7 8 The reason why the immune system in cirrhosis is defective and why patients become prone to infections is only partly understood. One reason may be that immune paralysis observed in cirrhosis is the result of exhaustion of circulating innate immune cells through continuous exposure to damage-associated molecular patterns (DAMPs) released from the necrotic liver and from bacterial products translocation from the gut.9 10

Circulating monocytes are central players in the host innate immune defence against invading pathogens since they can interpret the extent of the microbial threat and raise an appropriate inflammatory response to contain the infection.11 12 Following clearance, it is equally important that regulatory mechanisms are initiated to counteract excessive inflammation. Macrophages, for instance, are known for their plasticity in adopting either an inflammatory or regulatory phenotype.13–15 During the course of ACLF, however, it is unclear whether patients succumb to infection due to a failure of monocytes/macrophages to sufficiently dampen the production of proinflammatory mediators leading to septic shock and organ damage or whether they trigger a prolonged anti-inflammatory response rendering the patient incapable to respond to secondary infections. Nevertheless, studies taking a snapshot of ACLF monocyte function at a given time, have confirmed the presence of a suppressive monocyte phenotype with reduced human leucocyte antigen DR isotype (HLA-DR) surface expression, antigen presentation capacity as well as impaired ability to secrete proinflammatory cytokines in response to bacterial components.16–19 Although singular factors such as MerTK or prostaglandin E2 have been implicated as potential mechanisms responsible for monocyte dysfunction during ACLF syndrome,18 19 the overarching pathways that drive and sustain these disease-associated defects are not fully understood.

The functional phenotype of monocytes/macrophages is highly regulated at both transcriptional and metabolic levels.20 21 To exert their immunological functions, they can metabolise a variety of carbon substrates,22 23 and the nature of the metabolic program involved is critically associated with their activation status.24–26 For example, proinflammatory macrophages consume glucose and heavily rely on glycolysis for ATP generation. Additionally, they exhibit a ‘broken’ tricarboxylic acid (TCA) cycle, allowing accumulation of citrate and succinate. On the other hand, anti-inflammatory monocytes/macrophages maintain an intact TCA cycle and favour fatty acid oxidation, as a mode of ATP production.26–28 In this study, the aim was to obtain a comprehensive view of the transcriptional profile of monocytes from patients with ACLF and to investigate whether their molecular signature correlated with a functional and metabolic switch towards a suppressive phenotype. More importantly, we investigated whether metabolic rewiring of the cells using pharmacological inhibitors that channel glutamine into the TCA cycle, could restore ACLF monocyte dysfunction. These results may provide new insights into fundamental disease mechanisms.

Materials and methods

Patient characteristics

For this study, we recruited healthy controls, patients with decompensated alcoholic cirrhosis and ACLF diagnosed at the University hospital of Leuven between July 2013 and May 2017. Patients with alcoholic liver disease were identified and prospectively included at first contact in the emergency room (ER), the liver ward or the medical intensive care unit. Patients that used antibiotics or corticosteroid therapy during the 6 weeks preceding admission were not considered for inclusion in the study. Patients with concomitant other liver diseases including viral hepatitis were excluded from participation. Blood was collected for study purposes before initiating other therapy. Blood and urine cultures were obtained, ascites fluid collected for analysis and a chest X-rays performed to exclude infections and pulmonary infiltrates as per standard of care. Patients with ACLF were classified according to the Consortium on Chronic Liver Failure–Sequential Organ Failure Assessment (CLIF-SOFA) classification.4 Written informed consent was obtained from all patients or their designated family members.

Plasma cytokine measurements

Custom Meso Scale Discovery V-plex assays (Gaithersburg, Maryland, USA) were used to determine plasma cytokine (interleukin (IL)-6, IL-8, IL-10 and tumour necrosis factor alpha (TNFα) and chemokine (CCL2 and CCL3) levels. All measurements were performed in duplicate according to manufacturer’s instructions.

Phenotypic analysis of monocyte subsets and intracellular cytokine determination

Peripheral blood mononuclear cells (PBMCs) were isolated and stained with the following antibodies: CD3, CD19, CD56, CD14, CD16, HLA-DR (eBioscience, San Diego, California, USA) and matching isotype controls. Flow cytometric data acquisition was performed on a Gallios flow cytometer (Beckman Coulter, Analis, Belgium) and analysed using FlowJo software.29 For intracellular staining, PBMCs were stimulated with lipopolysaccharides (LPS) and brefeldin A (eBioscience), for 18 hours and then stained with the same surface antibody cocktail as described above followed by the addition of Cytofix/Cytoperm (eBioscience) and anti-human IL-10 (BD Biosciences, Erembodegem, Belgium).

Monocyte isolation and culture

Blood was collected from healthy donors and patients with ACLF in heparin-coated tubes (BD Biosciences). Immediately after collection, CD14+ monocytes were isolated from the PBMC fraction using a negative selection procedure according to the manufacturer’s specifications (Dynabeads untouched human monocyte kit, Invitrogen, Lennik, Belgium). Monocytes were cultured in RPMI medium containing antibiotics and further supplemented either with plasma from healthy donors or plasma from patients with ACLF at a final concentration of 20% by volume. Patient plasma used in these experiments represent a pool of four donors with ACLF grade 2. Plasma was heat inactivated (56°C for 30 min) and passed through a 0.22 µM filter prior to use. Following a 16 hours incubation time (37°C, 5% CO2), the cells were harvested for transcriptional analysis or functional assessment of their phagocytic capacity as described below.

Monocyte phagocytic and oxidative burst capacity following exposure to Escherichia coli

The phagocytic and oxidative burst capacity of human monocytes were assessed as previously described.30 Briefly, 100 µL heparinised peripheral blood was incubated for 1 hour at 37°C with pH-rodoRed-labelled E. coli bacteria (Invitrogen), previously opsonised with E. coli BioParticles opsonizing reagent (Invitrogen). This was followed by the addition of th (20 min at 37°C) and staining for human monocyte markers. Flow cytometric data acquisition was performed as described above.

RNA isolation and RNA sequencing analysis

RNA was isolated from freshly isolated monocytes and from monocytes cultured in the presence of ACLF- or normal plasma for 16 hours. RNA quantity and quality was measured, and samples meeting RNA integrity criteria were used for next generation RNA sequencing (NGS) analysis using Illumina NextSeq instrument (detailed description in online supplementary file). Sequencing and initial processing of the raw data was performed by the Nucleomics Core Facility, VIB, Leuven. Sequencing data are available at the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE93265.

Data analysis

After preprocessing, reads were aligned to the reference genome of Homo sapiens (GRCh37.73) and a generalised linear model was fitted (described in the online supplementary file). Two types of analysis were performed: (1) unpaired analysis between CD14+ monocytes from patients with ACLF versus decompensated- or healthy controls (reflecting the ex vivo transcriptional profile) and (2) paired analysis of CD14+ monocytes cultured in the presence of ACLF- or normal plasma for 16 hours (reflecting an in vitro simulation of acute ACLF). The individual results were combined, the average calculated and the resulting p values were computed. A gene was considered differentially expressed if a log2 fold change greater than +1 or less than -−1 and a corrected p<0.05 (see online supplementary text for more details). We analysed GO:Biological Processes (http://geneontology.org/) to explore the ex vivo patient samples. Alternatively, KEGG pathways were used to explore the pathways involved following in vitro exposure of CD14+ monocytes to ACLF serum (http://www.genome.jp/kegg/pathway.html).

Live-cell real-time phagocytosis assay

Isolated CD14+CD16 monocytes were plated onto 96-well clear flat bottom polystyrene tissue culture treated microplates and allowed to adhere. Monocytes were cultured in RPMI 1640 medium containing antibiotics and further supplemented either with plasma from healthy donors or plasma from patients with ACLF at a final concentration of 20% by volume (see above). In some instances, the cells were treated with methionine sulfoximine at the indicated concentrations. pHrodo pathogen bioparticles were added at indicated concentrations and the plates were transferred into a humidified incubator (37°C, 5%, CO2) and measured in real time using an IncuCyte Zoom imager (Satorius). Four images per well from three technical replicates were taken every 2 hours using a 10× objective lens and then analysed using the IncuCyte Basic Software.

Real-time quantitative PCR (qPCR)

cDNA was synthesised using SuperScript II reverse transcriptase and random hexamer primers (Invitrogen/Life Technologies, USA). The PCR reaction was carried out in a mixture that contained appropriate sense and antisense primers and a TaqMan MGB probe in TaqMan Universal PCR Master Mixture (Applied Biosystems, Foster City, USA). Beta-2-microglobulin was used as housekeeping gene. qRT-PCR amplification and data analysis was performed using the Lightcycler 96 (Roche Applied Science, Penzberg, Germany). Each sample was assayed in duplicate. The ΔΔCq method was used to determine relative gene expression levels.

Statistical analysis

Group comparisons were performed using Kruskal-Wallis with Dunn’s correction for multiple testing or Mann-Whitney-Wilcoxon rank sum tests where appropriate. Spearman correlation was used to determine associations between variables. Statistical analyses were performed using JMP V.11.0.0 (SAS Institute, Cary, North Carolina, USA) and SigmaPlot V.12.0 (Systat Software, San Jose, USA). Two-sided p value <0.05 was considered statistically significant.

Results

ACLF monocyte subset distribution, phenotype and cytokine response differ from patients with decompensated cirrhosis

Patients with ALCF are highly susceptible to infections and monocytes represent the first line of defense against pathogens entering the systemic circulation. Failure of monocytes to respond to danger signals may be due to an overproduction of immunosuppressive factors and/or the induction of a tolerant state. To identify the mechanisms that play a role in monocyte dysfunction during ACLF, we evaluated the expression of characteristic immunosuppressive markers such as intracellular IL-10 or surface expression of HLA-DR as readout for their antigen presentation capacity. We specifically investigated whether these features from monocytes of patients with ACLF were different from those observed in patients with decompensated cirrhosis. The clinical characteristics as well as circulating biochemical and immunological parameters of the patient cohort are documented in the online supplementary table 1 and supplementary figure 1. Flow cytometric analysis of monocyte distribution showed a significant decrease in the classical monocyte subset and expansion of the intermediate monocyte population (figure 1A). Further, indicative of a decreased activation status, all monocyte subsets expressed lower levels of surface HLA-DR. The latter observation was most pronounced in monocytes from patients with ACLF (figure 1B). Additionally, we assayed intracellular IL-10 production following LPS exposure and detected elevated numbers of IL-10 producing monocytes within the intermediate monocyte as well as the classical monocyte subsets from patients with ACLF, suggesting an elevated immunosuppressive function (figure 1C).

Figure 1

Phenotypic and functional analysis of monocyte subsets within patients with decompensated cirrhosis or acute-on-chronic liver failure (ACLF). Peripheral blood mononuclear cells (PBMCs) were isolated and stained for monocyte specific surface markers. (A) Flow cytometric analysis of classical, intermediate and non-classical monocyte frequencies in patients with ACLF (n=15), decompensated cirrhosis (Decomp) (n=7) and healthy controls (n=10) (mean±SEM; *p<0.05; **p<0.01). (B) The quantitative median fluorescence intensity (MFI) of HLA-DR surface expression on classical, intermediate and non-classical monocytes obtained from ACLF (n=15), Decomp (n=7) and healthy controls (n=10) (mean±SEM) (*p<0.05; **p<0.01compared to controls). (C) Frequency of intracellular IL-10 positive classical, intermediate and non-classical monocytes in patients with ACLF (n=4), Decomp (n=4) and healthy controls (n=12). The data show mean±SEM (*p<0.05).

ACLF monocytes feature a distinctive immunosuppressive transcriptional profile

To obtain an in-depth understanding of the molecular signatures responsible for this dysfunctional response, we performed a transcriptomic analysis of classical monocytes as the most abundant monocyte subset. CD14+ cells from patients with ACLF (n=9), decompensated cirrhosis (n=4) and healthy controls (n=5) were freshly isolated with a negative selection strategy whereby CD16+ cells were additionally depleted along with all other unwanted cell types to avoid granulocyte contamination (see the Materials and methods section for full description). Notably, since the disease status can be very dynamic only ACLF grade 2 patients were included for this purpose to ensure homogeneity within this experimental group (see online supplementary table 2 for patient characteristics). Gene ontology analysis to assess the biological processes involved revealed that the upregulated genes from decompensated cirrhosis monocytes were predominantly associated with immune response or leucocyte activation (figure 2A). Conversely, biological pathways promoting immunological processes and cell activation were associated with the downregulated genes in monocytes of patient with ACLF (figure 2B). Although there was an overlap of differentially regulated genes between these two conditions (not shown), the comparison of monocyte profiles of patients with ACLF to that of decompensated cirrhosis, unveiled differences unique to ACLF as disease entity. In this regard, the immunosuppressive nature in ACLF is further highlighted since the majority of the downregulated genes involve immune response processes (figure 2C).

Figure 2

Immunosuppressive transcriptional profile of monocytes from patients with acute-on-chronic liver failure (ACLF). RNA sequencing was performed on isolated monocytes from patients with ACLF (n=9), patients with decompensated cirrhosis (Decomp) (n=4) and healthy controls (n=5). Differentially expressed genes with a log2 fold change greater than +1 or less than −1 and false discovery rate (FDR) <0.05 were computed to explore patient subgroups. Volcano plots indicate all the upregulated (red) and downregulated (green) genes within monocytes from Decomp compared with controls (A) or patients with ACLF compared with controls (B). (C) The comparison of patients with ACLF versus Decomp. In addition, the top 10 biological processes for the differentially expressed genes were identified and these are indicated as circular charts surrounding the volcano plots. The biological processes that were upregulated are illustrated in a red colour on the right side of the volcano plots, while the downregulated processes are depicted in a green colour to the left of the volcano plots. The size of the circle charts correlates with the number of genes per process that were regulated (the number thereof is indicated in the middle of the chart). The colour intensity of the circle charts indicates the significance whereby biological processes were upregulated (dark red=most significant) or downregulated (dark green=most significant). Heat maps of proinflammatory M1 markers (E) or anti-inflammatory M2 markers (E) selected from the RNA sequencing data shows the close clustering of the monocyte samples within each experimental group. The scale of gene expression is indicated by a colour ranging from low (green) to high (red).

Focusing specifically on subsets of immune-related genes illustrated that patient as well as healthy monocytes cluster together, indicative of their distinct gene expression profiles. Heat maps of markers characteristically expressed by proinflammatory monocytes/macrophages (M1-like) were predominantly repressed (figure 2D). These include cytokines and their receptors (TNFa, IL15, IL15R, IL23A, IL1B), chemokines (CCL4, CXCL9, CXCL10), costimulators of antigen presentation (CD80, CD83) as well as transcription factors (STAT1). In sharp contrast, anti-inflammatory markers (M2-like) were largely overexpressed in monocytes from patients with ACLF (figure 2E). We observed a number of scavenger receptors (CD163, MRC1, CD36, MARCO), growth factors (HGF), suppressive cytokines (IL10), chemokines (CCL22) as well molecules involved in phagocytosis of apoptotic cells (MERTK, TGM2) and markers of a M2-like surface phenotype (MS4A4A) to be upregulated. In line with previous reports,31 32 we detected an increased frequency of MerTK-positive cells with a macrophage-like morphology in the liver tissue of patients with ACLF compared with patients with decompensated cirrhosis (online supplementary figure 3). Furthermore, ACLF monocytes featured dampened expression of a battery of antigen presentation molecules (HLA), as well as coactivator (CIITA), molecular scaffold (TAP2) and molecules needed for proper folding and trafficking of major histocompatability complex class II (CD74) (online supplementary figure 2). Conversely, a number of heat-shock proteins (HSPA1A, HSPA1B, HSPA1L, HSPA6), which act as chaperones for proper folding and transport of newly synthesised polypeptides, were specifically upregulated within ACLF monocytes (online supplementary figure 2). Combined, our data reveal that the extent of the immunosuppressive status of ACLF monocytes goes beyond what has been documented before and implicate major defects in their potential to mount proinflammatory responses and to raise T cell reactivity against pathogens (figure 1B).

Defective antibacterial response of ACLF monocytes

We next aimed to evaluate whether immune dysfunction in ACLF translates into a defective functional capacity of monocytes to detect, engulf and respond to infection. To investigate monocyte ability to detect and recognise bacteria, we evaluated basal surface expression of both TLR2 and TLR4 on classical monocytes from the different patient groups. Interestingly, patients with decompensated cirrhosis and patients with ACLF portrayed dampened expression levels of these markers (figure 3A,B). We further assessed the ex vivo phagocytic and oxidative burst capacity of these cells following exposure to E. coli. In line with defective bacterial recognition, monocytes isolated from patients with decompensated cirrhosis and patients with ACLF featured a significantly impaired phagocytic capacity, although this defect was most prominent in patients with ACLF (figure 3C). Monocytes from patients with ACLF also featured a clear defective oxidative burst response, following interaction with E. coli (figure 3D). To further identify the underlying pathways leading to this functional defect, we investigated differentially regulated genes related to the antibacterial and oxidative burst response of monocytes originating from patients with ACLF (see schematic representation of the components involved within figure 3E). Notably, we detected dampened expression of IRF8, a prominent regulator of monocyte/macrophage proinflammatory and antibacterial function (figure 3F_. Dampened expression levels of IRF8 may implicate a defective ability to activate transcription of constituents of the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase complex, a key component of the oxidative burst response.33 Nevertheless, the activity of the subunits within the NADPH oxidase complex can be regulated in different ways including at the level of post-translational modification. In this regard, we observed significantly lower levels of protein kinase C (PRKCE) within ACLF monocytes, suggesting that the defective oxidative burst may be partially related to inadequate phosphorylation of NADPH oxidase subunits (figure 3G). Furthermore, we detected the upregulation of Rnf145, a E3 ubiquitin ligase that negatively regulates gp91phox steady-state protein levels (figure 3H). Combined, our data highlight a number of mechanisms that may explain the functional defect of patients with ACLF to respond to and eradicate infections.

Figure 3

Defective antibacterial mechanisms of monocytes from patients with acute-on-chronic liver failure (ACLF). Flow cytometric analysis of (A) TLR2 and (B) TLR4 (B) expression on the surface of CD14+CD16-− monocytes. Flow cytometry data derived from a whole blood phagocytosis and oxidative burst assay following Escherichia coli challenge and surface staining for monocyte specific surface markers in patients with ACLF (n=13), decompensated cirrhosis (Decomp) (n=12) and healthy controls (n=15). Graphs show the percentage of CD14+CD16 classical monocytes that have internalised E. coli bacteria (C) or have produced reactive oxygen species (D) (mean±SEM; *p<0.05, **p<0.01, ***p<0.001). (E) A schematic representation of components involved in monocyte oxidative burst response. The figure indicates potential checkpoints in the activation or regulation of events in this pathway. Expression levels of IRF8 (F), PRKCE (G), RNF145 (H) within monocytes from patients with ACLF (n=8), decompensated cirrhosis (Decomp) (n=4) or healthy controls (n=5) (mean±SEM; *p<0.05).

Monocytes cultured with ACLF plasma mimic functional and transcriptional disease characteristics

In order to explore the possibilities of reversing monocyte dysfunction, we devised an ACLF disease-mimicking in vitro model whereby freshly isolated CD14+ monocytes from healthy donors were cultured for 16 hours either in the presence of pooled plasma (20% v/v) from ACLF grade 2 patients or normal human plasma (figure 4A). We first evaluated whether freshly isolated healthy CD14+ monocytes exposed to pooled ACLF grade 2 plasma in vitro exhibit a transcriptional profile similar to that observed in monocytes obtained from patients with ACLF. Indeed, the transcriptional signature obtained mimicked that of monocytes obtained from patients with ACLF showing a clear elevation of anti-inflammatory (M2-like) markers, but only a partial dampening of inflammatory (M1-like) markers at 16 hours after exposure (illustrated by the heat maps within figure 4B,C). In line with the previous data, qPCR and multiplex immune assay showed that ACLF plasma triggered a clear induction of anti-inflammatory parameters (IL-10 and/or MerTK), while a tendency to dampen proinflammatory mediators (TNFα, IL-8) was observed in cultured monocytes (figure 4D–G). Finally, to verify whether ACLF plasma-conditioned monocytes also portrayed a defective phagocytic activity, we evaluated their ability to engulf gram-negative pHrodo-labelled E. coli particles over time using the IncuCyte Zoom system. The data indicate that ACLF plasma blunted the phagocytic capacity of healthy donor-derived monocytes (figure 4H). Of note, this defect in bacterial uptake only became apparent after 8–16 hours following conditioning of the cells in ACLF plasma. Interestingly, monocytes cultured in the presence of ACLF plasma featured a similar defective capacity to engulf gram-positive pHrodo-labelled Staphylococcus aureus particles (online supplementary figure 4). Combined, this observation strongly supports the hypothesis that circulating plasma derived factor(s) in ACLF plasma induce an ACLF-like phenotype in healthy donor derived monocytes.

Figure 4

In vitro model to mimic monocyte dysfunction in acute-on-chronic liver failure (ACLF). (A) Freshly isolated CD14+CD16 monocytes from healthy donors were cultured in the presence of normal human plasma or plasma from patients with ACLF (20% v/v). RNA sequencing was performed after a culture period of 16 hours to identify differentially expressed genes. Hierarchic clustering of monocytes following conditioning with ACLF plasma (ACLF) or normal human plasma (Normal) was performed using the expression of a subset of (B) proinflammatory M1 and (C) anti-inflammatory M2 genes. The scale is indicated by a colour ranging from low (green) to high (red). Quantification of proinflammatory and anti-inflammatory parameters from monocytes of healthy donors following condition with Normal or ACLF plasma for 16 hours. Relative mRNA levels of (D) proinflammatory or (E) anti-inflammatory parameters were determined by means of real-time quantitative PCR at the end of the culture period (mean±SEM; n=6). Similarly, protein levels of a selection of (F) proinflammatory and (G) anti-inflammatory markers were determined after 16 hours of culture with a high-sensitive multiplex immunoassay. Finally, healthy monocytes were cultured in the presence of ACLF or normal plasma (20% v/v) and challenged with low doses pHrodo-labelled E. coli to measure engulfment with the IncuCyte system over a period of 24 hours (H).

Metabolic rewiring and promoting TCA cycle metabolism, restores the phagocytic capacity of in vitro generated as well as monocytes from patients with ACLF

Mounting evidence suggests that immune-suppressive cells exhibit unique metabolic requirements.33 34 In particular, they maintain an intact TCA cycle and favour oxidative metabolism, especially fatty acid oxidation, as a mode of ATP production. Similar to what has been described for M2 macrophages, ACLF-conditioned monocytes also featured low glycolysis activity with lower expression of glycolytic enzymes (HK1, PGK1, PFKM, ENO2) (figure 5A & online supplementary figure 5). Glutamine, on the other hand, can be catabolised thereby fueling the TCA cycle (figure 5A).35 We next argued that continued glutamine anabolism through glutamine synthetase (GLUL) may be responsible for the sustained suppressive phenotype of ACLF monocytes and we therefore implemented a strategy to redirect glutamine into the TCA cycle by inhibiting its enzyme activity. Hereto, we tested the ability of the inhibitor of GLUL and methionine sulfoximine (MSO) to restore the phagocytic capacity of ACLF-plasma-conditioned monocytes (the concept thereof can be visualised in figure 5B). Strikingly, treatment of monocytes cultured in the presence of ACLF plasma with MSO triggered a dose-dependent increase in their ability to phagocytose E. coli bioparticles (figure 5C). Considering the importance of the above findings, we reanalysed the expression levels of parameters involved in glutamine metabolism within monocytes of patient with ACLF. More specifically, we calculated the ratio of glutamine anabolism over catabolism by assessing GLUL and glutaminase (GLS)1 expression levels. Interestingly, the data clearly indicate an elevated GLUL/GLS ratio for monocyte originating from patients with ACLF compared with healthy counterparts (figure 5D). To improve our understanding of the significance of these findings, we investigated possible correlations of the GLUL/GLS ratio with survival or model for end-stage liver disease (MELD) scores of the patients. Importantly, our data show that the GLUL/GLS ratio positively correlated with disease severity in patients with ACLF (figure 5E). Treatment of ACLF patient-derived monocytes with MSO similarly improved their capacity to recognise and engulf bacteria (figure 5F). Finally, treatment of patient monocytes with MSO inhibited IL-10 production while promoting the production of the proinflammatory cytokine, TNFα (figure 5G,H).

Figure 5

Metabolic reprograming of acute-on-chronic liver failure (ACLF) monocytes. (A) Schematic representation of glucose and glutamine metabolism programs. (B) Blocking glutamine synthetase (GLUL) with the pharmacological inhibitor, methionine sulfoximine (MSO) fuels the tricarboxylic acid (TCA) cycle (schematic representation). The effect of MSO on healthy monocyte’s ability to phagocytose bacteria under normal plasma or ACLF plasma conditioning. Monocytes were cultured in the presence of ACLF or normal plasma (20% v/v) and treated with MSO (2 mM). Low doses of pHrodo-labelled Escherichia coli was added to the cells and engulfment was measured with the IncuCyte system. (C) The results depict the number of pHrodo-positive monocytes over time. (D) The GLUL/glutaminase (GLS) ratio in monocytes from patients with ACLF compared with monocytes from healthy controls. (E) The correlation of GLUL/GLS ratio with the model for end-stage liver disease (MELD) scores of patients with ACLF. (F) Monocytes obtained from patient with ACLF were treated with or without MSO and challenged with low doses of pHrodo-labelled E. coli before measuring engulfment with the IncuCyte system. Monocytes obtained from patients with ACLF were treated with or without MSO and cultured under normal conditions for 16 hours before assessment of (G) interleukin (IL)-10 and (H) tumour necrosis factor alpha (TNFα) production.

Discussion

Bacterial infection is one of the most common precipitating events in ACLF that may initiate multiorgan failure. The mechanisms and pathways responsible for failure of the host innate immune system to respond to bacterial infections in ACLF are only partly understood. Here we demonstrate that monocytes from patients with ACLF show a broad range of innate immune defects, which extend beyond what has been previously documented. This monocyte immunosuppressive signature in ACLF is clearly different from decompensated cirrhosis supporting ACLF as a distinct clinical entity. We further demonstrate the importance of metabolic rewiring in establishing the monocyte phenotype in ACLF and showed that promoting glutamine fuelled TCA cycle metabolism within monocytes, improved their defective phagocytic and inflammatory function.

In our study, we observed that ACLF monocytes displayed profound defects in ex vivo phagocytic and oxidative burst capacity following exposure to E. coli (figure 3E) that could be recapitulated by exposing healthy monocytes to ACLF plasma (figure 4G) suggesting a specific defect in antibacterial function. We specifically focused on constituents of the Nox2 NADPH oxidase complex, which mediates the generation of reactive oxygen species to kill invading pathogens. Interestingly, although transcription of constituents of the Nox2 NADPH complex were not severely affected, we observed the upregulation of Rnf145, a negative regulator of the Nox2 complex. Rnf145 has recently been implicated in the proteostasis of the Nox2 complex by endoplasmic reticulum-associated degradation.36 Graham et al, very elegantly demonstrated the importance of this factor as negative regulator by showing that ablation of Rnf145 in murine macrophages enhance bacterial clearance and rescued the oxidative burst defects associated with Ncf4 haploinsufficiency.36 A further important observation was that ACLF monocytes featured a dampened expression of IRF8, a transcriptional activator of the oxidative burst response. IRF8 along with IRF1 and their downstream targets have been shown to be specifically required for the protection against infection.37 In addition, IRF8 plays a critical role in monocyte/macrophage polarisation to an inflammatory phenotype and as regulator of genes involved in antigen presentation and T cell activation.37 38 Interestingly, the impaired IRF8 expression within ACLF monocytes was associated with lower levels of molecules involved in antigen presentation (HLA, CD74) and T cell costimulation (CD80, CD86) as well as characteristic markers for classically activated immunostimulatory or proinflammatory M1 monocytes/macrophages (CXCL9, CXCL10, IL-15 and ITGAL). It is therefore tempting to speculate that an IRF8-dependent mechanism may be involved in the failure of ACLF monocytes to raise an appropriate oxidative burst and inflammatory response. However, besides hampered proinflammatory and antibacterial defence mechanisms, ACLF monocytes also exhibited elevated expression of markers associated with alternatively activated anti-inflammatory/tolerogenic M2 macrophages including IL10, MERTK, CCL22, IL4R, CD36, MARCO and CD163. This result is in line with previous studies showing upregulation of singular anti-inflammatory parameters in monocytes from patients with ACLF.18 19 Combined, our data outline the extent of immune dysfunction within ACLF monocytes and provides possible insights in the underlying mechanism governing the susceptibility to infections, as a characteristic precipitating event during this syndrome.

Interestingly, similar immunosuppressive features along with metabolic and epigenetic reprogramming have been observed in circulating monocytes from patients with late stage sepsis.39 40 This is in sharp contrast to the characteristic overt upregulation of proinflammatory parameters during acute stages of sepsis that resembled more the expression pattern observed in monocytes from patients with decompensated cirrhosis.41 Considering the fact that sepsis is a major cause of death in patients with ACLF, future investigations aimed at understanding the mechanisms underlying immune dysfunction, as well as similarities between sepsis-induced and ACLF-induced immunosuppression, may improve therapeutic strategies.42 Another potential interesting and important line of investigation is how the source of infection may affect circulating and peripheral innate immune cells. Intestinal barrier failure and bacterial translocation through the portal vein has recently been shown to directly affect liver residing myeloid cells rendering them incapable of clearing infection.43–45 It will therefore be intriguing to determine how ACLF monocyte function correlates with the presence of bacteria or its components in the circulation as well as with clinical patient data regarding past and ongoing infections.

Monocyte/macrophage function is not only controlled at the transcriptional and posttranscriptional level but their metabolic program can also govern their phenotype.33 46 In particular M2 macrophages maintain an intact TCA cycle and favour oxidative metabolism, especially fatty acid oxidation, as mode of ATP production. Additionally, glutamine anabolism is particularly elevated in M2 macrophages for the generation of UDP-GlcNAc and glycosylation of many characteristic scavenger receptor markers.25 35 We demonstrate here for the first time that monocytes conditioned with ACLF plasma portrayed a clear downregulation of key parameters in the glycolytic pathway (including PFKM, PGK1 and ENO2) as well as modulation of parameters involved in fatty acid metabolism. Strikingly, blocking GLUL using an inhibitor thereof improved the capacity to clear bacteria both in ACLF-plasma conditioned monocytes and monocytes derived from patients with ACLF. Furthermore, we demonstrate that monocytes derived from patients with ACLF feature elevated expression of enzymes regulating glutamine anabolism (GLUL) and dampened expression of enzymes regulating glutamine catabolism (GLS). Underscoring the biological relevance of this finding, we detected a positive correlation between GLUL/GLS ratio and MELD disease severity scores. Interestingly, in the cancer research field such metabolic rewiring of tumour-associated macrophages, through the inhibition of GLUL, has been shown to revert M2 macrophages towards an M1-like phenotype, that promoted immunostimulatory and antiangiogenic effects that prevented the development of metastasis.47 Our results are in line with these findings showing decreased IL-10 and elevated TNFα production by patient monocytes following inhibition of GLUL. GLUL therefore may represent an important checkpoint in the regulation of the immunological function of monocytes/macrophages both in cancer as well as in response to bacterial infections.

In addition to characterising the full extent of innate immune dysfunction of monocytes from patients with ACLF, we highlight a number of possible mechanisms that may explain increased susceptibility to infections. Our data also demonstrates that metabolic programs can be manipulated to rescue defective monocyte phagocytic functions. Finally, this work highlights the importance of metabolic immunotherapeutic strategies in the treatment of ACLF, and it will be intriguing for future work to further fine tune and develop this approach as potential interventional strategy.

Acknowledgments

The authors wish to acknowledge the contributions of the VIB-Nucleomics Core for excellent assistance in the RNA sequencing experiments. We also wish to thank Petra Windmolders, Ingrid Vander Elst and Elien de Smidt for the excellent technical assistance.

References

View Abstract

Footnotes

  • Patient consent for publication Not required.

  • Contributors SvdM, HK and JvP conceptualised and planned the study. HK, JdP and SvdM wrote the protocol. HK and JdP performed the flow cytometry, monocyte functional studies, gene expression and multiplex cytokine assays. AT, RFA and SDG assisted with the Incucyte experiments. LM documented the clinical patient information. MVH and TR performed immunohistochemistry experiments. BG, DC, MB and S-MF provided support for the metabolic aspects of the study. JvP performed the RNA sequence pathway identification, the hierarchic clustering and statistical analysis. RJ, GM, RPM, TG, DC, FN, WL, LV and SvdM recruited and cared for the patients. HK and SvdM wrote the manuscript.

  • Funding This work was supported by internal funding from the UZ Leuven (KOOR) and KU Leuven (C1) as well as by the FWO and Gilead Sciences.

  • Competing interests SvdM, FN and DC are recipients of Flanders fund for scientific research (FWO fundamenteel-klinisch mandaat). RJ has research collaborations with Yaqrit and Takeda. RJ is the inventor of OPA, which has been patented by UCL and licensed to Mallinckrodt. He is also the founder of a UCL spin out, Yaqrit Limited, Ammun Limited and Cyberliver Limited.

  • Ethics approval Medical Ethics Committee (KU Leuven/UZ Leuven; S54588).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The RNA sequencing data will become available at the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE93265.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.