Granulocytic myeloid‐derived suppressor cell population increases with the severity of alcoholic liver disease

Abstract Alcoholic liver disease (ALD) is a progressive liver disease that can cause a series of complications, including cirrhosis, liver failure and hepatocellular carcinoma. Granulocytic myeloid‐derived suppressor cell (gMDSC) populations have been observed to expand in various liver diseases and to inhibit innate and adaptive immunity in patients with liver disease. However, the characteristics of gMDSCs in patients with ALD have not been studied. We studied 24 healthy controls (HCs) and 107 patients with ALD and found an accumulation of gMDSCs in the peripheral blood of patients with alcoholic liver cirrhosis (ALC). Furthermore, ALC patients with a poor prognosis displayed a significant increase in peripheral gMDSCs and showed an increased capacity for arginase I production compared to HCs. In contrast, plasma arginase I levels in ALC patients were negatively correlated with total bilirubin and international normalized ratio, two key parameters of liver damage. Importantly, gMDSCs accumulated in the livers of ALC patients, and the frequency of liver gMDSCs significantly correlated with that of peripheral gMDSCs. In addition, gMDSC enrichment in vitro significantly inhibited the function of natural killer (NK) cells, perhaps preventing the NK‐induced apoptosis of hepatic stellate cells. In summary, increased peripheral and intrahepatic gMDSC populations are present in patients with ALC and may contribute to enhancing the severity of liver cirrhosis.

innate and adaptive immunity through the production of arginase, reactive oxygen species (ROS) and interleukin-10 (IL-10). [7][8][9] MDSC accumulation leads to the sequestration of most of the available cysteine, which is essential for certain processes. The absence of cysteine prevents the production of proteins related to T cell activation. 10 In addition, cytokines such as granulocyte colony stimulating factor and granulocyte-macrophage colony stimulating factor are increased under pathological conditions, effectively stimulating the proliferation of MDSCs. Furthermore, different MDSC subsets with various phenotypic and functional features are classified as CD33 + CD11b high CD14 + HLA − DR − CD15 − monocytic MDSCs (mMDSCs) and CD33 + CD11b high HLA − DR − CD14 − CD15 + granulocytic MDSCs (gMDSCs). 11 Accumulating evidence indicates that gMDSCs produce high levels of arginase I, IL-10 and transforming growth factor-β (TGF-β) 7,12-14 and play a vital role in regulating the immune environment. A relatively recent study showed that an increased gMDSC population inhibits T cell responses via an arginase-dependent pathway, thus sustaining HBV replication without overt liver injury. 11 However, whether gMDSCs can regulate the function of immunocytes in ALD patients remains unknown.
In the present study, we characterized gMDSCs in two groups of patients to determine the correlation between gMDSCs and the severity of ALD and found that the gMDSC population was increased in ALD patients and closely associated with disease progression.

| Patients
Blood samples were obtained from 105 individuals with ALD, including 16 AH patients and 89 ALC patients. All ALD patients were diagnosed according to existing criteria, 15,16 and individuals with concurrent HBV, HCV, autoimmune liver disease or severe systemic disease were excluded. ALC patients were further subdivided based on the Child-Pugh score, which is used to evaluate the prognosis of patients with ALC: A (5-6 points), B (7-9 points) and C (10-15 points). 17    were stained with a Wright-Giemsa kit (Solarbio, Beijing, China)

| Flow cytometric analysis
following the manufacturer's instructions.

| Cell isolation
Peripheral blood mononuclear cells were isolated from freshly heparinized blood using protocols described previously by our team. 18 gMDSCs were isolated by CD14-negative selection, followed by CD15positive selection, using anti-CD14 and anti-CD15 antibody-coated magnetic beads (Miltenyi Biotech, Bergisch Gladbach, Germany). Natural killer (NK) cells were isolated by negative selection using an NK isolation kit. All assays were performed according to the manufacturers' instructions. The purity of the separated cells ranged from 95% to 98%.
Anti-CD107a antibody was added at the beginning of the stimulation period, and monensin was added 1 hour later. To further investigate the potential suppressive mechanisms by gMDSCs, 500 μmol/L Nω-Hydroxy-nor-L-arginine, diacetate salt (nor-NOHA; Calbiochem, Darmstadt, Germany) was added in the culture system.

| Immunohistochemistry and confocal microscopy
Paraffin-embedded and formalin-fixed liver tissue sections (5 μm thick) were incubated with anti-CD15 antibody (Novus Biological, Littleton, CO, USA) and/or anti-CD66b antibody (BD Bioscience) overnight at 4°C after endogenous peroxidase activity was blocked with 0.3% Then, the colorimetric substrate 3-amino-9-ethyl-carbazole (red colour) was added, followed by counterstaining with haematoxylin for single staining according to previously described protocols. 19 Images were captured at high magnification (400×) by an Olympus CX31 microscope and Olympus FV1000 confocal microscope. 20,21

| HPLC-MS/MS for amino acid quantification
Plasma L-arginine levels were determined using an API 3200 QTRAP HPLC-LC MS/MS as described elsewhere. 11

| Enzyme-linked immunosorbent assay
Commercial kits were used for detecting arginase (Hycult, Uden, The Netherlands) according to the manufacturer's protocols. The detection limit of arginase I was 13 ng/mL.

| Statistical analysis
All data were analysed using SPSS 20.0 software. The quantitative data are presented as the mean ± SD. The Kruskal-Wallis H nonparametric test was used for multiple comparisons among various groups, and the Mann-Whitney U-test was used to compare differences between two groups. The Wilcoxon matched-pairs signedrank test was used for analyses within the same group. Correlations between variables were evaluated by Spearman's rank correlation test. For all tests, a two-sided P < 0.05 was considered to indicate statistical significance.

| gMDSC populations were increased in ALD patients
Circulating frequencies of gMDSCs were determined as the frequency of CD15 + HLA − DR − CD14 − cells derived from CD33 + CD11b + myeloid cells ( Figure 1A). Wright-Giemsa staining confirmed the granulocytic nature of the gMDSCs population analysed ( Figure 1B).
As shown in Figure 1C, the frequency of gMDSCs was significantly higher in ALD patients than in HCs. The frequency of gMDSCs in ALD patients was highly variable; in some subjects, the expanded population of gMDSCs accounted for as much as 78% of circulating myeloid cells. Consistent with the results for gMDSCs, the frequency of mMDSCs, which were defined as the CD33 + CD11b high CD14 + H-LA − DR − CD15 − population, was also significantly higher in ALD patients than in HCs (data not shown), but the proportion was much lower (mean = 2.48% mMDSCs vs 26.7% gMDSCs, P < 0.001) ( Figure S1). According to the clinical diagnosis of ALD, we divided the ALD cohorts into two groups: AH and ALC. Both AH (P < 0.001) and ALC groups (P < 0.001) have a higher frequency of gMDSCs compared with HCs, whereas there was no significant difference between AH and ALC groups (P = 0.662) ( Figure 1C).
Child-Pugh scores are used to determine the prognosis of liver cirrhosis, and higher scores predict poor prognosis. For ALC patients, Child-Pugh B and C patients have a greater population of gMDSCs than do Child-Pugh A patients ( Figure 1D). However, there was no significant difference between Child-Pugh B and C patients. These data indicate that gMDSCs have a role in ALD and suggest that the increased frequency of gMDSCs in ALD patients is closely associated with disease progression.

| Arginase + gMDSCs were increased in ALD patients
As shown in Figure 2A, gMDSCs showed stronger staining than other cell types for arginase I (Figure 2A). In the ALD and HC groups, gMDSCs showed almost 100% staining for arginase I compared to the isotype control ( Figure 2B). As shown in Figure 2C, the proportion of arginase + gMDSCs was elevated in both AH and ALC patients compared to HCs. Then, we analysed the difference in the arginase + gMDSC frequency in ALC patients based on Child-Pugh score. We noted a significant increase in the arginase + gMDSC frequency in Child-Pugh B and C patients compared with Child-Pugh A patients ( Figure 2D).

| gMDSCs were enriched in the livers of ALD patients
To determine whether gMDSCs accumulated in the liver of ALD patients, we first stained liver tissues for CD15 and CD66b, two markers of gMDSCs. As shown in Figure 3A, a few CD15-or CD66b-positive cells were found in liver tissues from HCs, while the number of such cells was considerably higher in liver tissues from ALC patients than from HCs or those with AH ( Figure 3B). Importantly, more CD15-or CD66b-positive cells were observed in the liver sinusoids in ALC patients. These data were further validated by double epitope staining and confocal microscopy ( Figure 3C). In addition, we found that ALC patients with a higher number of circulating gMDSCs had more intrahepatic gMDSCs ( Figure 3D,E). In addition, both circulating and intrahepatic gMDSCs highly expressed arginase I compared with the isotype control ( Figure 3F). These data suggest considerable accumulation of gMDSCs in the livers of ALC patients, and this infiltration of gMDSCs is closely associated with their peripheral counterparts.

| Arginase I concentration was increased in ALD patients
We further analysed plasma arginase I levels and found that corresponding with the frequency of arginase + gMDSCs, the plasma arginase level was markedly higher in ALD patients than in HCs.
Interestingly, AH patients had a higher arginase I levels than did ALC patients ( Figure 4A). In contrast with the results for arginase + gMDSCs, the arginase I concentration was significantly higher in the Child-Pugh A group than in the Child-Pugh B group ( Figure 4B).
Total bilirubin (TBIL) and international normalized ratio (INR) are important indicators of the severity of liver injury. Further analysis showed that the plasma arginase I concentration was inversely correlated with TBIL and INR in ALD patients ( Figure 4C). In addition, we found that the arginase I concentration was negatively with MELD score in ALC group ( Figure 4C). Importantly, as shown in Figure 4D, the plasma L-arginine concentration was significantly lower in ALD patients than in HCs.

| gMDSCs from ALD patients showed enhanced suppressive function
Activated NK cells play an important role in attenuating liver cirrhosis by killing hepatic stellate cells (HSCs). 22 We cultured NK cells with or without purified gMDSCs from ALD patients and found that a decreased percentage of NK cells cultured with gMDSCs expressed tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) compared to those cultured without gMDSCs. A similar result was obtained by evaluating the TRAIL mean fluorescence intensity between the two groups ( Figure 5A). Purified NK cells were stimulated with or without autologous gMDSCs. There was a lower percentage of cells expressing CD107a and a decreased capacity for IFN-γ production among NK cells cultured with gMDSCs compared to those cultured without gMDSCs in vitro ( Figure 5B). As expected, the addition of the arginase I specific inhibitor nor-NOHA significantly restored TRAIL expression, CD107a expression, and IFN-γ production in NK cells ( Figure 5A,B).
We further compared the inhibitory activity of gMDSCs from ALD patients and HCs and found that gMDSCs from ALD patients significantly inhibited CD107a expression and IFN-γ production in NK cells ( Figure 5C). These results strongly suggest that suppressive effect of gMDSCs is primarily mediated through the production of arginase in ALD patients.

| DISCUSSION
Myeloid-derived suppressor cells were first described in the early 1900s, but it was not until the late 1970s that the phenotypes, function and nomenclature of MDSCs began to be studied. 23 In recent years, researchers have become increasingly more interested in the immunoregulatory role of MDSCs. However, until now, the characteristics of MDSCs in ALD patients had not been reported. In our study, we found that the gMDSC population was increased in ALD patients and was closely associated with disease severity. We also provide evidence for the enhanced suppressive function of gMDSCs in ALD.
Myeloid-derived suppressor cells can be further divided into the mMDSC and gMDSC subsets based on phenotype and origin. In the hepatology field, mMDSCs play a role in chronic inflammatory liver disease. The frequencies of peripheral and intrahepatic mMDSCs correlate with liver inflammatory injury and viral load in chronic viral hepatitis 24,25 and are associated with disease severity and prognosis in primary biliary cholangitis. 26  There are significantly more mMDSCs in the liver in patients with non-alcoholic fatty liver disease than in HCs. 27