Expression of AXL receptor tyrosine kinase relates to monocyte dysfunction and severity of cirrhosis

Immune dysfunction determines morbidity and mortality in liver cirrhosis. Distinct AXL-expressing circulating monocytes, which regulate antimicrobial responses, expand with progression of the disease.


Introduction
Patients with cirrhosis are at increased risk of infection and consequent acute decompensation (AD) with substantially elevated morbidity and mortality (1). Compared with the overall rate of infections in hospitalised patients (5-7%), bacterial infections occur significantly more frequently in patients with cirrhosis (32-34%) (2,3).
Similarly, infections account for more than 50% of hospitalisations in cirrhotic patients, are the main precipitant for AD without and with organ failure (acute-on-chronic liver failure [ACLF]) (4,5), and implicate a high mortality (2,6). Infection susceptibility in cirrhosis has been attributed to a state of immuneparesis, defined by inadequate immune responses to microbial challenge (7,8,9).
The pathophysiology of immuneparesis in cirrhosis is highly complex and remains incompletely understood, involving diverse defects in immune cell function, including monocytes, and soluble factors in multiple compartments (7). Circulating monocytes from patients with AD and ACLF compared with stable cirrhosis demonstrated reduced expression of HLA-DR and attenuated production of TNF-α/IL-6 in response to lipopolysaccharide (LPS), which has previously been linked to adverse outcome (10,11,12). Moreover, the role of bacterial translocation in the pathogenesis of immune dysfunction and infection susceptibility has been highlighted (13,14).
TAM receptors (TYRO3, AXL, and MERTK) belong to the family of receptor tyrosine kinases. Among immune cells, they are expressed on monocytes, macrophages, dendritic cells, and glial cells, and additionally on epithelial cells of the reproductive system, the retina, and tumour cells (15). TAM receptors are important regulators of innate immune homeostasis, acting by inhibition of TLR signalling pathways through a signal transducer and activator of transcription 1 (STAT1)-and suppressors of cytokine signalling (SOCS1/3)-dependent mechanism (15,16) and by promotion of phagocytic removal of apoptotic cells (efferocytosis) (16). Their activation succeeds ligand binding (growth arrest-specific gene-6 [GAS6], PROTEIN S) and interaction with phosphatidylserine on apoptotic cells (15,16,17). In murine dendritic cells, activation required interaction with the type I interferon receptor (IFNAR) (16).
We recently identified the expansion of MERTK-expressing monocytes and macrophages in diverse compartments in patients with ACLF that dampened innate immune responses to microbial challenge and conferred disease severity and adverse outcomes (18). The expansion of MERTK-expressing monocytes and macrophages was moreover detected in acute liver failure (18) and characterised by both suppressed immune responses and enhanced efferocytic capacities (19). Another immune-suppressive population, expanded in the circulation of patients with ACLF, was monocytic myeloidderived suppressor cells (M-MDSC) that suppressed T cell activation, innate immune responses, and pathogen uptake (20).
It is not clear when and/or under which circumstances immuneparesis and monocyte dysfunction occurs with an associated susceptibility to infection during the clinical course of cirrhosis and portal hypertension, before the onset of AD. The main emphasis of this study is to detail the expression of TAM receptors on monocytes in relation to monocyte function and disease severity of patients with cirrhosis in the absence of AD using patients with AD, chronic liver disease (CLD) without cirrhosis and healthy controls (HCs) as comparators. We hereby seek to better understand the pathophysiology of immuneparesis development in patients with cirrhosis prior to AD and identify candidates for biomarkers and future immunotherapeutic targets that may preserve innate immune responses.

Patient characteristics
Patients with cirrhosis were distinguished between Child-Pugh A, B, and C and compared with AD, CLD without cirrhosis, and HC. The cohort was characterised by disease severity scores, aetiologies, and diverse clinical parameters (Tables S1 and S2). In patients with cirrhosis without AD, 1-yr mortality rate was 4.5% and rising with Child-Pugh stage: A (0%), B (5.7%), and C (11.8%). 1-yr mortality rate for those with AD was 75%, with two of eight patients dying within 28 d of enrolment. N = 9 patients deceased from cirrhosis-related complications during follow-up of 1 yr (secondary infections [n = 4], ACLF with multiorgan failure [n = 3], hepatocellular carcinoma [HCC] [n = 1], hypovolemic shock due to variceal bleeding [n = 1]), and the cause of one death was unknown. Current infections at hospital admission were seen in 62.5% of patients with AD. Within 4 wk following inclusion into the study, 5.2% of patients (Child B: 5.7%, Child C: 5.9%, and AD 25%) developed infectious complications, adapted from the definition by Bajaj et al (9). Episodes of AD developed in 10.2% of cirrhotic patients (Child B: 17.1% and Child C: 17.6%) within 4 mo following inclusion (Table S1).
Innate immune responses are impaired in patients with cirrhosis and parallel the expansion of an AXL-expressing circulating monocyte population In patients with AD/ACLF, we recently described impaired inflammatory cytokine production of circulating monocytes to microbial challenges (18,20). Attenuated responses were also seen in stable cirrhotic patients (18,20). It, however, remained unknown when and to what extent circulating monocytes develop immune dysfunction over the time course of disease progression. We measured ex vivo inflammatory cytokine production upon LPS treatment of circulating monocytes from patients with cirrhosis at different stages of disease. TNF-α and IL-6 production was reduced in cirrhosis compared with HC and incrementally decreased from Child A to C, and AD but remained preserved in patients with CLD without cirrhosis (Fig 1A and B).
In parallel with increased disease severity and the decline of inflammatory cytokine production in response to LPS, we demonstrated the expansion of an AXL-expressing monocyte population ex vivo in the circulation of patients with cirrhosis (Figs 1C and S1A). The occurrence of AXL-expressing monocytes was independent of the underlying aetiology and other potential confounders (inpatient treatment, current infection, antimicrobial treatment, immunosuppressive therapy, and non-metastatic malignancies; Fig S1B and D). Within monocyte subsets, the expression of AXL was highest in but not restricted to the intermediate subset (cluster of differentiation [CD]14 ++ CD16 + ) ( Fig S2A). AXL expression on monocytes of patients with CLD without cirrhosis was low; a similar pattern was also seen in AD ( Fig 1C). Other immune cells such as lymphocytes and granulocytes barely expressed AXL ( Fig  S2B). Longitudinal follow-up data showed an increase in AXL expression after re-compensation of AD episodes and a change in AXL expression paralleling the evolution of disease severity after 1 yr (Fig S1E and F). Recently, we described a MERTK-expressing monocyte population that was expanded in the circulation of patients with AD/ACLF (18), which was again confirmed in this cohort ( Fig 1D). In CLD with and without compensated cirrhosis, however, MERTK and TYRO3 expressions were sparse (Figs 1D and E, and S1A). Circulatory plasma levels of the AXL ligand GAS6 were significantly elevated in cirrhosis compared with HC, independent of the aetiology. GAS6 increased from Child A to C and correlated with AXLexpressing monocytes (Figs 1F and S1C).
Circulating AXL-expressing monocytes in patients with advanced cirrhosis indicate diseases severity, complications, and poor outcome We next assessed the expansion of AXL-expressing monocytes in relation to clinical parameters, disease severity scores, indicators of complications, and outcome. The proportion of AXLexpressing monocytes strongly correlated with Child-Pugh and model for end-stage liver disease (MELD) scores and the classification of cirrhosis established by D'Amico et al (21) (Fig 2A). AXL-expressing monocytes also correlated with soluble AXL (sAXL) plasma levels. sAXL was significantly elevated in cirrhosis compared with controls, independent of the underlying aetiology, correlated with Child-Pugh and MELD, and predicted the onset of AD episodes within 4 mo following study inclusion (Fig S3A-E).
AXL on monocytes predicted 1-yr mortality with a sensitivity of 80% and specificity of 79.2% for the criterion median fluorescence intensity (MFI) > 440. AXL moreover predicted the onset of AD within 4 mo following inclusion, and the development of infection over the next 4 wk. AXL was also associated with C-reactive protein (CRP) (Figs 2B and S4A, and B). Furthermore, AXL-expressing monocytes were associated with manifestations of portal hypertension (ascites, hepatic venous pressure gradient, varices, hepatic encephalopathy, and renal dysfunction) and correlated with individual parameters of liver function (bilirubin, international normalised ratio [INR], albumin) (Fig 2C and D). High AXL expression on monocytes may albeit small numbers also predict the need for transplantation, transplantation-free 1-yr survival, and development of HCC within 1 yr (Fig S4C-E).
Phenotype of circulating monocytes in patients with cirrhosis and the AXL-expressing monocyte population Monocytes from patients with CLD without cirrhosis did not differ phenotypically from HC. Monocytes from patients with cirrhosis, however, showed an HLA-DR low phenotype with decreased expression of Fcγ-and homing receptors (CD32 low CX3CR1 low CCR7 low ). HLA-DR expression significantly decreased from Child A to C ( Fig  S5A). Within this entire population, the expanded subset of AXLexpressing monocytes (AXL + ) ( Fig 3A) were CD14 + CD16 high HLA-DR high indicating a mature monocyte subpopulation with augmented expression of Fcγ-receptor CD32, TLR4, and homing/chemokine receptors (CCR5, CCR7, and CX3CR1) ( Fig 3B). There was no difference in viability between AXL + -and AXL − -monocytes ( Fig 3C).
Importantly, the CD14 + HLA-DR + AXL + immune cell subset detailed here has to be distinguished from the recently identified immunosuppressive M-MDSCs in patients with cirrhosis and ACLF (20), which we observed expanding from Child A to C in our cohort. M-MDSCs were CD14 + CD15 − CD11b + HLA-DR low/neg , as previously defined (22), and expressed lower levels of AXL in comparison with CD14 + HLA-DR + monocytes (Fig S6A-C).
AXL-expressing circulating monocytes contribute to impaired innate immune responses and suppression of T cell proliferation while retaining phagocytic capabilities for bacteria ex vivo To investigate the effect of the expanded CD14 + HLA-DR + AXL + subset on the impaired inflammatory cytokine responses observed in monocytes from patients with cirrhosis we assessed the functional properties of AXL + monocytes ex vivo.
Detailed analyses of the distinct subsets revealed that TNF-α/ IL-6 production in response to LPS was decreased in both CD14 + HLA-DR + AXL + monocytes and M-MDSCs when compared with CD14 + HLA-DR + AXL − monocytes from patients with cirrhosis and HC. In detail, TNF-α production decreased from 59.2% (13.8)  , and B). The CD14 + HLA-DR + AXL − population represented the majority of monocytes in HC, indicating it may be regarded "functionally intact" but was sequentially lost in the circulation of patients with progression of cirrhosis ( Fig S6B). In line with previous data detailing TAM receptor signalling pathways (16), we observed higher mRNA levels of SOCS1/SOCS3 in monocytes of Child B/C patients, compared with HC ( Fig 4B). Our data thus reveals the determination of functional roles of monocyte subsets in a pathophysiological context such as cirrhosis.
We further revealed that AXL expression on monocytes in cirrhosis was associated with inhibition of T cell proliferation, when tested in an allogeneic mixed lymphocyte reaction ( Fig 4C).
Ex vivo phagocytic capacity of Escherichia coli (E. coli) bioparticles and live GFP-containing E. coli by circulating monocytes did not differ between cirrhotic patients and HC. AXL + monocytes showed preserved phagocytosis capacities for E. coli bioparticles and live E. coli bacteria, whereas M-MDSCs revealed reduced phagocytosis, when compared with CD14 + HLA-DR + AXL − monocytes from patients with cirrhosis and HC (Figs 4D and S7C).
Considering these observations, the expanded CD14 + HLA-DR + AXL + monocyte population in the circulation of patients with cirrhosis (notably, not existing in healthy subjects) remained functionally phagocytic, but prevented T cell proliferation and inflammation (low TNF-α/IL-6 production) in a presumably SOCS1/3-dependent manner, representing an immune-regulatory "homeostatic" monocyte population expanding during cirrhosis progression.

AXL overexpression in THP-1 cells attenuates LPS-induced inflammatory cytokine production in vitro
As proof-of-concept for the observations developed above, in vitro, we overexpressed AXL in the monocytic THP-1 cell line using a retroviral system (Fig 5A). Following transduction, AXL mRNA expression (2.2 ± 0.3-fold; Fig S8A) and protein levels (88% THP-1-AXL + cells; Fig 5A and B) were increased. Phenotypic characterisation of the THP-1-AXL + model cell line is illustrated in Fig S8B. Consistent with the observations in patients with cirrhosis ex vivo, AXLexpressing THP-1 cells produced less TNF-α and IL-6 in response to LPS when compared with non-transduced THP-1 cells (Fig 5C).
Pathogen-associated molecular patterns (PAMPs), cytokines, bacterial uptake, and efferocytosis induce AXL up-regulation on monocytes Next, we sought to understand the mechanisms leading to the expansion of the described immune-regulatory AXL-expressing monocyte population. Pathophysiologically, cirrhosis progression involves development of portal hypertension and subsequent pathologic bacterial translocation facilitates microbial products accessing the systemic circulation (8,13,14,23). Hence, we tested PAMPs and damage-associated molecular patterns (DAMPs) for their ability to modify monocyte differentiation and AXL expression in vitro. Stimulation with bacterial products such as TLR ligands (Pam3SK4, LPS, CpG, and poly (I:C)) significantly up-regulated AXL expression in vitro. Similarly, pro-inflammatory factors (IFN-α and TNF-α) induced AXL up-regulation. In contrast, the DAMP highmobility group protein B1 (HMGB1), TGF-β, and AXL ligand GAS6 did not induce AXL expression. LPS-induced up-regulation of AXL was time dependent (Fig 6A), and those monocytes produced significantly less TNF-α/IL-6 upon LPS when compared with monocytes without prior LPS exposure ( Fig S9A). Notably, monocytes incubated with 25% plasma of cirrhosis patients did not show biologically relevant changes in AXL expression (Fig 6B), suggesting that additional factors are required to generate this subset in vivo.
Phagocytosis is required for efficient clearance of pathogenic microorganisms and initiation of various immune responses. AXL + monocytes exhibited preserved phagocytosis capacities (Figs 4D and S7C) and AXL expression significantly increased on monocytes after phagocytosis of E. coli and Staphylococcus aureus (S. aureus) bioparticles as well as of live GFP-E. coli bacteria (Fig 6C). Phagocytosis capacity of E. coli bioparticles positively correlated with the degree of AXL expression and concurrently negatively with TNF-α production (Fig S7D-E).
Within inflammatory milieus, as prevalent in different compartments of patients with cirrhosis (8), efferocytosis is required to maintain immune homeostasis (24,25). We, therefore, co-cultured healthy monocytes with neutrophils and HepG2 cells previously labelled with a cytoplasmic cell-tracker and then induced to apoptosis (19,26). AXL expression on monocytes was up-regulated following co-culture with apoptotic cells. Monocytes that engulfed apoptotic cells (efferocytosing) were characterised by higher AXL expression, compared with monocytes that did not (resting). AXL + monocytes showed higher efferocytosis capacity than AXL − monocytes when co-cultured with apoptotic neutrophils (Fig 6D); following efferocytosis of neutrophils, AXL + monocytes produced less TNFα/IL-6 upon LPS than AXL − monocytes (Fig S9B), supporting the accumulation of AXL + immune-regulatory monocytes in an inflammatory environment.

AXL inhibitor BGB324 and metformin restore innate immune responses of monocytes from patients with cirrhosis ex vivo
Given the distinct immune-regulatory functions of the AXL + monocyte population in patients with cirrhosis and its association with disease severity and infection, we questioned whether inhibition or downregulation of AXL would reverse the anti-inflammatory properties. BGB324 is a selective small molecule inhibitor of AXL previously tested in clinical studies (27). Metformin, a well-known antidiabetic drug, was previously described to down-regulate AXL expression (28) and to regulate the AXL signalling cascade in the context of cancer (29,30). Metformin-induced down-regulation of AXL was confirmed in monocytes from patients with cirrhosis ex vivo here. Treatment with BGB324 did not affect AXL expression (Figs 7A and S10A). Both, BGB324 (1 μM) and metformin (10 mM) treatment restored LPS-induced TNF-α production of monocytes from patients with cirrhosis ex vivo. When comparing AXL + with AXL − monocyte populations from patients with cirrhosis following metformin treatment, cytokine production was enhanced in AXL + but not AXL − cells ( Fig 7B). Viability of monocytes after metformin treatment was marginally reduced (Fig S10B). Whereas phagocytosis capacity of E. coli bioparticles was preserved after BGB324 administration, it decreased upon metformin treatment ( Fig 7C).

Discussion
In this work, we detail the characteristics of circulating monocytes in patients suffering from cirrhosis at different stages of disease without signs of AD. We newly describe the expansion of an AXLexpressing immune-regulatory monocyte subset (CD14 + HLA-DR + AXL + ) along cirrhosis progression and its close association with disease severity, infection susceptibility, development of AD, and prognosis. The AXL-expressing monocyte generation was linked to the abundance of PAMPs and cytokines, phagocytosis, and efferocytosis in the context of recurrent inflammation. Our findings substantially add to the understanding of the pathophysiology of immuneparesis in cirrhosis and identify a potential biomarker and immunotherapeutic target.
Although we and others have previously focused on impaired innate immune responses after the onset of AD/ACLF, when infection susceptibility is highest (10,12,18,20), it is barely known when and under which circumstances innate immune dysfunction occurs and infection susceptibility emerges during progression of cirrhosis and portal hypertension. Previous studies addressing phenotype and function of classical (CD14 + CD16 − ) and nonclassical (CD14 + CD16 + ) monocyte subsets in CLD showed inconsistent data regarding cytokine production and phagocytosis (31,32,33). We previously showed that inflammatory cytokine production was depressed not only in AD/ACLF but also in stable cirrhosis (18), whereas the underlying mechanism remained unexplained.
Here, we demonstrate the accumulation of circulating CD14 + HLA-DR + AXL + monocytes with attenuated innate immune functions, that is, decreased inflammatory cytokine production (TNF-α/IL-6) and T cell activation, along disease progression of cirrhosis in compensated and chronically decompensated patients. Although dysfunctional monocytes were rarely encountered in Child A, their number substantially expanded in advanced stages (Child B/C) and displaced functionally intact monocytes. The findings were irrespective of the underlying aetiology. Monocyte functions were preserved in patients with CLD without cirrhosis (F ≤ 3). Follow-up data of individual patients showed an evolution of AXL-expressing monocytes in parallel with disease severity scores. However, this requires further evaluation in larger prospective longitudinal studies.
AXL is a member of TAM receptors that mainly function as inhibitors of TLR-and cytokine receptor-mediated monocyte/macrophage activation and promoters of apoptotic cell removal (15,16). Loss of AXL expression on antigen-presenting cells has been linked to autoimmunity (15). TAM receptors are differentially expressed and exhibit distinct expression, regulation, and activity under specific conditions (34,35). Although CD14 + HLA-DR + AXL + cells accumulated with worsening stages of cirrhosis, they disappeared upon acute hepatic decompensation. By contrast, CD14 + HLA-DR + MERTK + monocytes remained undetectable in stable cirrhosis and emerged upon AD (18). In addition, CD14 + HLA-DR + MERTK + cells were abundant in the circulation and the liver in acute liver failure where they were characterised as resolution-type monocytes/macrophages (19). This underlines the distinct and counter-regulatory roles of AXL and MERTK in different phases of cirrhosis and inflammation. Our findings in a human disease verify previous murine data that identified differential and reciprocal expression and function between AXL and MERTK on BMDMs/BMDCs (34). The authors described both receptors as phagocytic mediators in vitro, whereas MERTK expression was induced by tolerogenic stimuli and induced tolerance, and AXL was induced by inflammatory stimuli and acted in the feedback inhibition of inflammation (34). The underlying differential signalling mechanisms need to be addressed in future investigations.
Extensive characterisation of the AXL-expressing monocyte population revealed an immune-regulatory subset that emerged in the circulation likely to maintain immune homoeostasis despite rising inflammatory signals during progression of cirrhosis. AXLexpressing monocytes were characterised by increased HLA-DR, CD16 and chemokine-receptor expression, enhanced clearance of apoptotic cells, increased expression of Fcγ-receptor, and preserved phagocytosis of E. coli, but attenuated T cell activation and secretion of pro-inflammatory cytokines (TNF-α/IL-6) after microbial challenge presumably in a SOCS1/3-dependent manner, as previously proposed (16). Although phagocytosis of pathogens represents the first line of defence when an organism encounters microbes, clearance of apoptotic cells is required for immune homeostasis during inflammation. Our findings suggest that AXLexpressing monocytes may expand during cirrhosis and progressive portal hypertension in response to the uptake of pathogens and bacterial products in the setting of pathologic bacterial translocation (8,13,14,23), and to clear apoptotic cell debris accumulating in response to chronic inflammation (15,16). Concurrently, excessive systemic inflammatory responses are inhibited.
Similar to our findings, a recent study described AXL-expressing murine airway macrophages at homeostatic conditions, which increased after influenza infection, thereby preventing excessive tissue inflammation through efferocytosis (35). AXL expression was critical for functional compartmentalisation, as it was not present on interstitial lung macrophages (35). We propose that in cirrhosis, AXL may be operative in settings where the nature of injury is driven by excessive pro-inflammatory responses, as present in diverse underlying aetiologies. This newly described immune-regulatory, AXL-expressing monocyte population must be clearly distinguished from another recently discovered immunosuppressive cell subset that accumulated in the circulation in stages of AD: M-MDSCs were characterised by suppression of T cell activation, pathogen uptake, and TLR-elicited pro-inflammatory responses to microbial challenge (20). Here, we observed that M-MDSCs started to emerge also in patients with cirrhosis without signs of AD. Considering the reduced inflammatory responses to microbial challenge of both CD14 + HLA-DR + AXL + -and M-MDSCs compared with functionally intact CD14 + HLA-DR + AXL − cells in relation to their abundance in the circulation, we propose that these populations together largely explain the depressed innate immune responses of the entire monocytic population at these stages.
Having observed AXL up-regulation on circulatory monocytes in relation to portal hypertension, we hypothesized an underlying mechanism involving pathologic bacterial translocation leading to the abundance of bacterial products, PAMPs (8,13,14,23), and subsequent chronic systemic inflammation (8,23). At the same time, chronic liver injury leads to release of DAMPs (36). Indeed, we were able to generate AXL-expressing monocytes with dampened innate immune responses by stimulation with selected TLR ligands and pro-inflammatory factors in vitro. These findings coincide with previous data showing AXL up-regulation on murine BMDMs (34,35) and peritoneal macrophages (35) upon stimulation with inflammatory stimuli (34,35). In contrast to M-MDSCs, generated in vitro by culturing monocytes in ACLF plasma (20), inflammatory factors in plasma of patients with cirrhosis were necessary, but insufficient to induce AXL up-regulation alone. Our data support the hypothesis that efferocytosis and phagocytosis of bacteria in the circulation are further required to enhance AXL up-regulation on monocytes. The stimulatory effect of pathogen uptake on AXL expression is novel and may explain high AXL expression on circulating monocytes in conditions where pathogens and their products become abundant because of pathologic bacterial translocation such as cirrhosis. TAM receptor activation after efferocytosis had previously been shown on murine BMDMs/BMDCs (34).
Dissecting the complexity of differential monocyte differentiation and activation of effector pathways of particular TAM receptors on monocytes at different stages of cirrhosis will be subject to future investigations, including the use of unbiased large-scale techniques. In a multisystem disorder such as cirrhosis, additional compartments such as the liver, but also the gut, the portal circulation, the peritoneum, and potentially others and their tissuespecific immune systems play crucial roles in the pathophysiology of the underlying immuneparesis. It is the aim of our subsequent investigations to detail the differentiation and immune function of tissue-specific myeloid cells in these compartments, in particular in respect to the immune-regulatory role of TAM receptors.
Moreover, by ex vivo proof of principle experiments treating monocytes from cirrhotic patients with the highly specific AXL inhibitor BGB324 and metformin, which was previously described to target and down-regulate AXL (28), innate immune responses were significantly enhanced, suggesting AXL as potential immunotherapeutic target to augment defence against infections. Whereas BGB324 did not negatively affect phagocytic capabilities, metformin did.
BGB324 was originally developed for cancer treatment and is currently tested in clinical Phase Ib/II trials for patients with aggressive and metastatic cancers (37). Interestingly, other studies have examined BGB324 as an anti-fibrotic agent. GAS6/AXL pathways were associated with fibrogenesis in CLD (38) and idiopathic pulmonary fibrosis (39), respectively, and were reversed by BGB324. Multi-tyrosine kinase inhibitors, including AXL, are known for their diverse antitumour effects and are tested in phase III clinical trials for advanced HCC (40). Distinct AXL blockage impacts on tumour progression through immune surveillance by AXL-expressing immune cells and anti-proliferative effect on AXL-expressing tumour cells (37,40). AXL inhibition by BGB324 may, thus, represent a promising concept with anti-fibrotic, immune-stimulatory, and also anti-tumour effects.
Whereas previous studies described anti-inflammatory, presumably AXL-independent properties of metformin on myeloid cells (41), we observed enhanced immune responses of AXL-expressing monocytes from cirrhotic patients after metformin treatment. Metformin, conventionally used as anti-diabetic drug, exerts various pleiotropic effects acting via diverse downstream signalling pathways (42) and has been reported to be associated with reduced HCC incidence (43) and reduced portal hypertension in cirrhosis models (44). Further studies hint at a potential regulatory effect of metformin on the AXL cascade in the context of cancer (29,30). As an inexpensive, well-established drug, metformin may represent an interesting immunomodulatory treatment option for patients with cirrhosis and no signs of AD, when AXL-expressing monocytes are frequent and the risk for metformin-associated lactic acidosis is low. Our data are suggestive to further investigate the potential significance of metformin in this context and its underlying signalling mechanism.
As these substances were only tested ex vivo here, subsequent in vivo studies in rodent models are required to systematically investigate target-and off-target effects such as auto-immunity or uncontrolled inflammation. We showed previously that inhibition of MERTK on monocytes of AD/ACLF patients reversed innate immune dysfunction (18). Given the distinct and reciprocal expression profiles of AXL and MERTK in cirrhosis, it further needs to be addressed which receptor to target at which stage of disease and in which compartment.
Finally, strong correlations of AXL expression on monocytes with disease severity and prognosis, that is, i.e. development of infection, episodes of AD, and 1-yr mortality underline its clinical significance. Two recent studies suggested sAXL as a serum biomarker for advanced liver fibrosis, cirrhosis, and HCC (45,46). Here, we observed strong correlations of AXL-expressing monocytes with co-culture with apoptotic cells for 8 h. Apoptosis of neutrophils and HepG2 cells after 24 h. AXL expression after efferocytosis, AXL expression of resting and efferocytosing monocytes, and efferocytosis capacity for neutrophils of AXL + -/AXL − -monocytes. Bar plots showing mean/SD. *P < 0.05/**P < 0.01 (unpaired/paired t tests).  the shed receptor sAXL and also liver disease severity scores. Based on our findings, the number of AXL-expressing monocytes in blood count may represent a prognostic biomarker for immuneparesis and cirrhosis and validates further evaluation.
In conclusion, the number of AXL-expressing immune-regulatory monocytes in the circulation of patients with cirrhosis indicated disease severity, immuneparesis, infection susceptibility, AD, and mortality. CD14 + HLA-DR + AXL + monocytes were expanded upon PAMP and cytokine exposure, pathogen,-and apoptotic cell uptake and hallmarked by preserved phagocytosis and enhanced efferocytosis but reduced cytokine production and T cell activation, implying a role in immune homeostasis in a condition defined by pathologic bacterial translocation and recurrent inflammation. Immunotherapeutic modulation of AXL may represent an option deserving evaluation to augment immune responses and reduce infection susceptibility, morbidity, and mortality in cirrhosis. , respectively, categorised according to Child-Pugh and European Association for the Study of the Liver -Chronic Liver Failure (EASL-CLIF) Consortium scores (47). We included HC (n = 27), patients with CLD without cirrhosis (n = 8; Metavir F ≤ 3), and patients with AD within 24 h following hospital admission (n = 8) as comparators. Healthy volunteers from the regions St. Gallen and Basel, Switzerland, were matched by age and sex and served as HCs. Patients' assent was obtained by the patients' nominated next of kin if they were unable to provide informed consent themselves. Cirrhosis was diagnosed by liver biopsy (n = 92, 95.8%) or clinical presentation with typical ultrasound (n = 4, 4.2%). Exclusion criteria for patients were age younger than 18 yr and evidence of metastatic malignancies (including HCC). Five patients with non-metastatic malignancies were included (HCC, Barcelona Clinic Liver

Clinical, haematologic, and biochemical parameters
Routine clinical and laboratory parameters obtained by the clinician such as full blood count, CRP, INR, liver, and renal function tests and other variables were entered prospectively into a database. Differential blood count at the sites was performed using Sysmex XE differential analyser (Sysmex Europe GmbH) (Cantonal Hospital St. Gallen) and ADVIA 2120i hematology systems (Siemens Healthineers) (University Hospital Basel).

Monocyte isolation
Monocytes were isolated from PBMCs using CD14 MicroBeads or Pan Monocyte Isolation Kit (Miltenyi Biotec) as previously described (18). Purity of monocytes was assessed by flow cytometry.
Cell viability was determined using the Annexin V Apoptosis Detection Kit I (including 7-AAD) according to manufacturer's protocols (BD Biosciences).

Formula for calculating absolute numbers of TYRO3/AXL/ MERTK-expressing cells and M-MDSCs
Absolute cell numbers were calculated with the formula: Frequency of TYRO3/AXL/MERTK-expressing monocytes or M-MDSCs (% defined by flow cytometry) × monocyte count (G/L).

Mixed lymphocyte reaction
Monocytes from HCs and patients with cirrhosis expressing high (>20%/450 MFI) versus low (<7.5%/300 MFI) AXL levels were isolated using the Pan Monocyte Isolation Kit (Miltenyi Biotec) and cocultured with allogeneic CD3 + T cells from a different healthy donor, isolated using the Pan T Cell Isolation Kit (Miltenyi Biotec), in a 1:1 ratio. T cell stimulation was induced with anti-CD2/CD3/CD28 beads (T cell Activation/Expansion Kit; Miltenyi Biotec) as previously described (20). T cells were stained with carboxyfluorescein succinimidyl ester at day 0. Proliferation was assessed at day 2 of co-culture by flow cytometry.

Quantitative RT-PCR
Total RNA of isolated monocytes from patients with cirrhosis, HCs, and THP-1/THP-1-AXL + cells in cell culture was isolated using RNeasy Mini Kit (QIAGEN) and reversely transcribed into cDNA using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems/ Thermo Fisher Scientific). qRT-PCR was performed with 400 ng of cDNA using LightCycler 480 SYBR Green I Master Mix (Roche). Commercial primers for AXL (Hs_AXL_1_SG QuantiTect Primer Assay) were purchased from QIAGEN. Sequences (59-39) of primers are as follows: SOCS1 forward: CCC CTT CTG TAG GAT GGT AGC A; reverse: TGC TGT GGA GAC TGC ATT GTC and SOCS3 forward: ATG GTC ACC CAC AGC AAG TT; reverse: TCA CTG CGC TCC AGT AGA AG. GAPDH was used as endogenous control as previously described (35). qRT-PCR was performed according to the manufacturer's recommendations on QuantStudio Real-Time PCR (Applied Biosystems/Thermo Fisher Scientific).

Whole blood phagocytosis assay
Whole blood was incubated with pHrodo E. coli Red BioParticles (Phagocytosis Kit for Flow Cytometry from Invitrogen/Thermo Fisher Scientific) and live E. coli NovaBlue carrying the gfp-mut2 encoding plasmid pCD353, which expresses a prokaryotic variant of GFP controlled by a lactac promoter as described in (49) (a kind gift of Prof Dr C Dehio, University of Basel). 5 μl of E. coli Red BioParticles were added for 15 min at 37°C to 100 μl of whole blood and processed as previously described (20). Blood with bioparticles was stained with antibodies against CD14, CD16, HLA-DR, AXL, and CD15, and CD3, CD19, and CD56 (BD Biosciences) and acquired on the flow cytometer. E. coli bacteria were freshly grown on LB Agar plates supplemented with kanamycin (50 μg/ml; Sigma-Aldrich) and incubated overnight at 37°C. A single colony was picked and grown in LB medium supplemented with kanamycin (50 μg/ml) and IPTG (1 mM; Sigma-Aldrich) for GFP induction at 37°C until early logarithmic growth (OD600 = 0.5-0.6) was reached. After the incubation period, bacteria (1 × 10 9 bacteria) were centrifuged at 3000g for 5 min at 4°C, resuspended in 1 ml PBS, and used immediately. 5 × 10 7 of GFPcontaining E. coli were added for 60 min at 37°C to 100 μl of whole blood and processed as previously described (20). Blood with GFPcontaining E. coli was stained with antibodies against CD14, CD16, and AXL and acquired on the flow cytometer. The rate of phagocytosis was obtained by the proportion of GFP positive monocytes.

In vitro inhibition of AXL
A small-molecule inhibitor of AXL, BGB324 (Selleck Chemicals), and metformin (Stemcell Technologies) were used. Selectivity and mechanism of BGB324 were described previously (27,38). Metformin was previously described to suppress AXL expression at a concentration of 10 mM (28). We used 0.5 × 10 6 PBMCs from HCs and patients per well on a 48-well plate and cultured them in X-VIVO medium (Lonza) containing 10% FBS in a 37°C, 5% CO 2 environment. The cells were treated with 1 μM BGB324/10 mM metformin or dimethyl sulfoxide/PBS for 24 h, harvested, and washed two times with PBS before the assessment of inflammatory cytokine production in response to LPS (100 ng/ml, 5 h), phagocytosis capacity, and viability of monocytes by flow cytometry. For the assessment of monocyte phagocytosis from isolated PBMCs in vitro, the harvested cells were incubated with pHrodo E. coli Red BioParticles (Invitrogen/Thermo Fisher Scientific) for 60 min, processed, and assessed by flow cytometry as previously described (20). The optimal dose of 1 μM BGB324 was initially defined by a dose finding experiment assessing cytokine production of monocytes in response to LPS (100 ng/ml, 5 h). Cell viability using Annexin V (BD Biosciences) was assessed after BGB324 and metformin treatment (Fig S10B-D).
For the experiments incubating healthy monocytes in plasma from HC (n = 4) and patients with cirrhosis (n = 4), 1 × 10 6 CD14 + cells were cultured in a 24-well plate for 24 h in X-VIVO medium (Lonza) containing 25% of the indicated plasma in a 37°C, 5% CO 2 environment. Subsequently, a fraction of these cells was phenotyped and the remainder was transferred to fresh medium for the assessment of LPS-stimulated TNF-α/IL6 production and phagocytosis capacity of monocytes by flow cytometry as detailed before.
For the experiments measuring AXL-expressing monocytes after treatment of bacteria, whole blood was incubated with pHrodo E. coli Red BioParticles, pHrodo S. aureus Red BioParticles (Invitrogen/ Thermo Fisher Scientific) for 15 min, and with live GFP-containing E. coli bacteria for 60 min and processed as described above.

Efferocytosis assay
The experimental design was adapted from Zizzo et al (26) and Triantafyllou et al (19). Human neutrophils were isolated using PolymorphPrep (Axis-Shield) by density-gradient centrifugation according to the manufacturer's protocols, re-suspended at 1 × 10 6 cells/ml in RPMI-1640 (Sigma-Aldrich) containing 10% FBS (complete medium), labelled with CellTracker Violet BMQC (5 μM in serum-free medium, 45 min, 37°C, dark; Life Technologies, Thermo Fisher Scientific), and incubated for 20 h (37°C in 5% CO 2 ) in 300 μl complete RPMI-1640 in 24-well plates. HepG2 cells were seeded at 0.4 × 10 6 cells/ml in 24-well plates, labelled with CellTracker Violet BMQC as described above, and incubated with LPS (1 μg/ml) for 18 h (37°C in 5% CO 2 ) in 300 μl complete RPMI-1640. After the incubation period, percentage of apoptotic neutrophils and HepG2 cells in culture was determined using Annexin V Apoptosis Detection Kit I according to the manufacturer's protocols (BD Biosciences). Neutrophils and HepG2 cells were re-suspended in the wells and healthy monocytes were added to apoptotic cells (1:4 monocytes to apoptotic cells ratio) for 8 h in 1 ml fresh complete RPMI-1640 (37°C in 5% CO 2 ). The cells were harvested and washed two times in PBS and subjected to immunophenotyping, intracellular staining of cytokine production in response to LPS (100 ng/ml, 5 h), and viability assays as described above. The rate of efferocytosis was obtained by the proportion of CellTracker-positive monocytes.

Statistical analyses
Statistical evaluation was performed in GraphPad Prism v.7.0a (GraphPad Software). P < 0.05 values were considered statistically significant. Data are shown as box and whiskers or scatter dot plots and expressed as median with 10-90 percentile, unless otherwise specified. For data that did not follow a normal distribution, significance of differences was tested using Mann-Whitney or Wilcoxon tests. Spearman correlation coefficients and area under the receiver operating characteristic curve were calculated. Normally distributed data were compared using paired or unpaired t-tests.
1 receptor antagonist, and soluble tumour necrosis factor receptors as well as effector cell desensitisation.