Inflammatory signature in acute-on-chronic liver failure includes increased expression of granulocyte genes ELANE, MPO and CD177

Acute-on-Chronic Liver Failure (ACLF) is associated with innate immune dysfunction and high short-term mortality. Neutrophils have been identified to influence prognosis in ACLF. Neutrophil biology is under-evaluated in ACLF. Therefore, we investigated neutrophil-specific genes and their association with ACLF outcomes. This is an observational study. Enriched granulocytes, containing neutrophils, isolated from study participants in three groups- ACLF(n = 10), chronic liver disease (CLD, n = 4) and healthy controls (HC, n = 4), were analysed by microarray. Differentially expressed genes were identified and validated by qRT-PCR in an independent cohort of ACLF, CLD and HC (n = 30, 15 and 15 respectively). The association of confirmed overexpressed genes with ACLF 28-day non-survivors was investigated. The protein expression of selected neutrophil genes was confirmed using flow cytometry and IHC. Differential gene expression analysis showed 1140 downregulated and 928 upregulated genes for ACLF versus CLD and 2086 downregulated and 1091 upregulated genes for ACLF versus HC. Significant upregulation of neutrophilic inflammatory signatures were found in ACLF compared to CLD and HC. Neutrophil enriched genes ELANE, MPO and CD177 were highly upregulated in ACLF and their expression was higher in ACLF 28-day non-survivors. Elevated expression of CD177 protein on neutrophil surface in ACLF was confirmed by flow cytometry. IHC analysis in archival post mortem liver biopsies showed the presence of CD177+ neutrophils in the liver tissue of ACLF patients. Granulocyte genes ELANE, MPO and CD177 are highly overexpressed in ACLF neutrophils as compared to CLD or HC. Further, this three-gene signature is highly overexpressed in ACLF 28-day non-survivors.

Sample processing. PMN were isolated from 6 ml of whole blood in EDTA vial within 2 h of collection, followed by centrifugation at 500×g for 10 min. Plasma was aseptically separated and stored at − 80 °C until further use. The remaining blood pellet used for PMN isolation as described below. Enriched PMN were used for all experiments. Overall workflow is described in Fig. 1A.
PMN isolation and enrichment analysis by flow cytometry. PMN were isolated by modified Boyum's method of double gradient centrifugation 16 . Blood pellet containing the buffy coat was diluted with 2× volume of sterile 1× PBS (VWR, USA, 97062-730), at room temperature (RT). Ficoll-Hisep (Himedia, INDIA, LSM 1077) was layered over Granulosep (Himedia, INDIA, LS004) in a 2:3 ratio to prepare a double gradient in a 15 ml centrifuge tube. Whole blood was carefully layered on top followed by centrifugation at 300×g for 30 min at RT without brakes. The following phases were formed in order (top to bottom) after centrifugation: Diluted plasma, PBMC, Hisep, PMN, Granulosep, RBC pellet. Diluted plasma was discarded; PBMC layer was separated and the lower enriched PMN layer was collected. PMN cells were resuspended in sterile 1 × PBS and washed twice by centrifugation at 500×g for 10 min. Contaminating RBCs were removed by incubating the washed pellet in 1 × RBC lysis solution for 10 min at 4 °C, followed by two 1 × PBS washes. The final cell pellet was reconstituted in 1.2 ml filtered RPMI-1640 cell culture media (Himedia, INDIA, AL171A) supplemented with 2% heat-inactivated fetal bovine serum (FBS) (Himedia, INDIA, RM10432). The enriched PMN cells were diluted 1:20 and counted by Trypan blue exclusion assay to score for live and dead cells (1 µl of reconstituted cells with 10 µl of 0.4% of Trypan blue and 9 µl of sterile 1× PBS). Total cells were counted, and 1 × 10 5 cells were stained with respective antibodies. To assess PMN enrichment, anti-human CD14 (clone M5E2, FITC conjugated, Cat No.301803, Biolegend, USA) and anti-human CD16 (clone 3G8, APC conjugated, Cat No 302011, Biolegend, USA) were used and their respective stain index calculated by antibody dilutions. 5 × 10 4 cells were acquired in BD LSR Fortessa X-20 Flow Cytometer. Unstained controls and single-stain tubes were prepared for each stained sample, and the acquisition was supported by BD FACSDiva software. PMN cells were gated using SSC versus FSC plot and single cells were gated as FSC height versus FSC area ( Supplementary Fig. 1A-D). Neutrophils were selected as CD14 − (negative) CD16 + (positive) population. The percentage enrichment of samples is listed in the Supplementary Table 1. PMN RNA isolation and microarray. RNA isolated from 1 × 10 6 enriched PMN cells was subjected to microarray to identify differentially expressed genes. Briefly, RNA was isolated from 1 × 10 6 cells using the TRIzol (Invitrogen) and the manufacturer's protocol was followed. The cell suspension was pelletized and 1 ml TRIzol was added to solubilize the cells by vortexing for 30 s. Molecular biology grade chloroform (0.2 ml) was added and mixed until milky-white appearance was formed and incubated at room temperature (RT) for 10 min. Phase separation was done by centrifugation at 12,700×g for 15 min. The upper aqueous phase was removed carefully, and RNA was precipitated using 400 µl of ice-cold Isopropanol. The precipitate was washed in 75% ethanol before being air-dried and suspended in 30 µl of RNase free water. RNA quantity and quality were checked using Bioanalyzer. www.nature.com/scientificreports/ normalization was done using Percentile and background correction was based on median. The normalized data was subject to sequential filtering: Expression filter of 20-100 window, Filter on flags of Detected/Non detected, Filter on Error on CV < 50%, SD < 0.1, SD < 0.5. Statistical analysis was proceeded with the probeset of CV < 50%. NetworkAnalyst (https:// www. netwo rkana lyst. ca/ Netwo rkAna lyst/ home. xhtml) was used for carrying out DEG analysis of the mentioned sample comparatives. NetworkAnalyst uses Limma workflow for analysing data from gene expression experiments 17 . The differentially expressed genes were then selected with threshold criteria of p value 0.05 and log FC > 1.0 and < − 0.5. The significant DEG were then tested for their biological pathway implication. All the gene sets from respective comparisons were taken individually for enrichment analysis. The enrichment analysis was done using ClueGO (a Cytoscape plugin) and Enrichr [18][19][20] . The enrichment criteria for ClueGO was set for detailed network specificity at 5% genes for clustering and pathway significance of p value 0.05. For Enrichr, the significant pathways were selected at adjusted p value 0.05. KEGG enrichment analysis was used for overlapping of the DEG in both cases. The pathways output from ClueGO was used for visualization in Cytoscape. A network centrality analysis was done for the created pathway with CytoNCA. A custom style was created wherein genes involved in various pathways were highlighted based on their expression levels and, the size of individual nodes varying to their betweenness values, which helped identify the key elements (genes/pathways) that regulate the network created. Gene set enrichment analysis (GSEA) was additionally performed using the Molecular Signatures Database or MSigDB (http:// www. gsea-msigdb. org/ gsea/ msigdb/ index. jsp). GO cellular component (GO_CC) enrichment analysis was done for DEG of ACLF versus HC and ACLF versus CLD using ClueGO, a Cytoscape plugin at p value ≤ 0.05. Additional analysis were carried out between our dataset and published eosinophil, and neutrophil datasets. The DEG profile of transcriptomics datasets were mined from literature for eosinophils (GSE65239) and neutrophils(GSE142254) , a neutrophil dataset of GSE ID-GSE153781 was also taken for analysis. The common DEG in ACLF with respect to HC and CLD and mentioned datasets were illustrated using Venn diagrams (https:// www. molbi otools. com/ listc ompare. php) 21,22 .
The DGE analysis for Neutrophil dataset (GSE153781) was carried out using GREIN, an online interactive web platform used to analyse GEO RNA-seq data (http:// www. ilincs. org/ apps/ grein/).

Microarray validation and tissue-level expression by Quantitative Real-time PCR. Total PMN
RNA was reverse transcribed into cDNA using verso cDNA synthesis kit (AB-1453A from Thermo Fisher Scientific). DNaseI treatment was inherent to this kit and prevented genomic DNA carryover into downstream reactions. PCR and qRT-PCR were performed with 500 ng of cDNA synthesized from total PMN RNA. Prior to qRT-PCR, temperature gradient standardization for all primer sets were done to select optimum annealing temperature. Validation of gene expression log fold change by qRT-PCR was performed for the top 8 differentially expressed genes in ACLF versus CLD using specific primers designed from IDT oligoanalyzer software. The upregulated genes chosen for validation were ELANE, MPO, CD177, OLFM4, and OLAH. The transcript for 18S rRNA was used as a reference gene. The genes ELANE, MPO and CD177 were selected on the basis of their reported specificity in neutrophils, since the neutrophil population is known to be expanded in ACLF, resulting in a high NLR [23][24][25] .
The primer sequence and qRT-PCR conditions are included as supplementary data as per Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines (See Supplementary data_MIQE guidelines file). All graphical representations of log fold change gene expression were done using Graphpad Prism.
Cell-surface CD177 staining and detection by flow cytometry. Peripheral blood was collected in EDTA vial, and 200 μl of blood was aliquoted for antibody staining. Briefly, 1 ml of RBC lysis buffer was added to 200 μl of whole blood, and mixed properly. A 10× red blood cell lysis buffer was prepared in-house using NH 4 Cl (0.155 M), KHCO 3 (0.01 M) and EDTA (0.1 mM). The tubes were incubated for 10 min at 4 °C, and centrifuged at 300×g for 10 min. The cell pellets were washed twice with 1× PBS and suspended in 300 μl of PBS to obtain a single cell suspension. For cell surface staining, 100 μl of cell suspension was used, and the antibodies CD16 (1:100) (clone 3G8, APC conjugated, Cat No 302011, Biolegend, USA), CD66b (1:100) (clone G10F5, FITC conjugated, Cat No 305103, Biolegend, USA) and CD177 (1.5:100) (clone MEM-166, APC/Cyanine 7 conjugated, Cat No 315809, Biolegend, USA) was used. Unstained controls, stained samples, and fluorescence minus one controls were acquired on BD-LSR Fortessa flow cytometry machine. Using the FACS DIVA software, PMN gating was done based on FSC-A v/s SSC-A plot. Enriched and activated neutrophils were gated based on CD16 + CD66b + (double positive) in a quadrant plot. CD177 + neutrophils were gated within these double positive cells.
Dual colour immunohistochemistry for CD177 and CD16 in post-mortem liver biopsy. 5-micron thick sections of formalin-fixed paraffin-embedded (FFPE) tissues were retrieved from the Department of Pathology, AIIMS New Delhi and taken on coated slides. Deparaffinization was done by dipping the slides in xylene for 5 min (2 changes), acetone for 2-3 min, alcohol for 2-3 min, and then under running/tap water. Antigen retrieval was performed with citrate buffer (pH = 6) in a microwave oven, at 100 degrees Celsius at 900 MW for 30 min. Tissue sections were then allowed to cool down to come to room temperature. The slides were washed three times with Tris buffer (pH 7.5). Endogenous peroxidase blocking was done with 4% Hydrogen peroxide in 96 ml of methanol for 20 min. Anti-CD177 (Invitrogen, pH 9, 1: 50), Rabbit anti-human antibody was incubated overnight at 2-4 degrees Celsius. Next day, three sequences of washings were given with Tris buffer (pH 7.5). Universal polymer-based secondary antibody (SkyTek Laboratories, USA) was incubated at room temperature for 30 min, and the reaction product was developed with 3, 3"-diaminobenzidine chromogen (1: 1). Appropriate positive and negative controls were used. Colour development was monitored under the microscope. Four sub- www.nature.com/scientificreports/ sequent sequences of washings with Tris buffer were given at 5 min intervals, followed by the addition of 200 µl of enhancer and incubation for 5 min at. After that the Mouse anti-CD16 antibody (Invitrogen, USA. 1: 50) was incubated at room temperature for 1 h.Three sequences of washings were given with TRIS buffer. Alkaline phosphatase tagged goat anti-mouse IgG H&L secondary antibody (Ab7069) was added in a dilution of 1:10 for 30 min at room temperature. Three sequences of washings were given in Tris buffer. A VECTOR ® Blue Alkaline Phosphatase chromogen was used to develop the colour of the reaction (Blue AP), prepared in Tris HCL, pH 8.5, with 5 min incubation at room temperature and monitoring under the microscope. The slides were then washed under running tap water for 3-5 min, counterstained with Neutral red before mounting with a glycerin solution. The dual-colour stained slides were photographed by using a BX43 Olympus microscope. The images taken at 2 × objective power were divided into multiple fields of vision (FOVs) having a diameter of 2 mm each.
The procedure for manual cell counting the biopsy core was sequentially divided into multiple non-overlapping FOVs. A systematic eyeballing and counting of the cells were performed and noted. This was done to adjust the variable number of FOV available for each case due to varying biopsy core lengths. In the case of clustering and overlap of the stained cells, only those cells were counted whose nuclei were identified. Additionally, in each case, randomly in 5 FOVs the manual counting was crosschecked by the manual tagging and counting tool of `the Image Proplus 6.1 software. Sepsis was defined on the basis of blood culture reports within 48 h of admission, and the patients who did not have sepsis were classified as having "sterile inflammation".

Statistical analysis. All statistical analyses and graphical representations were done using Graphpad
Prism software. Normally distributed continuous variables were expressed as Mean ± SD, and continuous variables with skewed distribution were expressed as Median (Interquartile Range). For two group comparisons, Unpaired T-test and Mann-Whitney Test were applied for categorical variables and non-normally distributed variables, respectively.

Results
Study participants, workflow and neutrophil enrichment. The study groups included in the microarray-based gene expression analysis were ACLF (n = 10), CLD (n =4 ); as well as Healthy controls (n = 4) ( Table 1, Fig. 1A). ACLF is a dysfunction that occurs due to an acute injury over an underlying chronic liver disease (CLD). All the ACLF patients recruited into this study had organ failure scores of above 2 (CLIF-C-OF 2 and 3) indicating the presence of multiple organ dysfunction. Chronic etiologies in ACLF patients included alcohol (n = 4), autoimmune hepatitis (AIH; n = 3), cryptogenic causes (n = 3). Acute etiologies included alcohol (n = 2), AIH (n = 2), HEV + alcohol (n = 1), cryptogenic causes (n = 5). Etiologies for CLD(Compensated -diseased controls) patients included alcohol (n = 2) and viral hepatitis (n =23) ( Table 1). Neutrophil enrichment was determined by surface labelling of cells for the markers CD14 and CD16; followed  Table 1) . RNA isolated from enriched PMN/granulocytes were subjected to microarray analysis. Differential gene expression (DGE) and pathways analysis were performed only with the samples which showed > 50% enrichment and RIN score > 6.0 ( Fig. 2A-E). The only exceptions were-1 sample of ACLF (45.2%) and 2 of CLD (45.2% and 41.1%). These were included due to their high RIN value (> 7.0) suggesting good RNA quality. Since the aim of the experiment was to describe an overall gene expression profile of PMN cells and additional validation of selected genes using qRT-PCR was incorporated, these three samples were included so that robust transcriptome profiles could be obtained from which gene expression patterns could be further extracted and validated.
The qRT-PCR validation set included ACLF (n = 30), CLD (n = 15) and Healthy controls HC (n = 15) samples. These samples were age and gender matched to minimize bias, and the baseline clinical parameters were compared between ACLF and CLD (compensated) ( Table 2). As expected, baseline parameters were significantly different from ACLF, due to the worsening of liver function and presence of organ failure in ACLF. We next estimated the Neutrophil to lymphocyte ratio (NLR) from the complete blood count (CBC) reports of these patients. The NLR median value was 8.2 in ACLF versus 1.9 in CLD patients ( Supplementary Fig. 1E).  Fig. 2A-C). CD177 was one of the highest upregulated genes in ACLF versus HC as well as ACLF versus CLD (Fig. 2C). Other genes found to be among the top 50 upregulated genes in ACLF were-ELANE, OLFM4 and OLAH ( Fig. 2C; Supplementary Tables 2 and  3). MPO gene, a classical neutrophil activation marker was found to be upregulated in ACLF versus CLD with a Log2FC of 2.4 (Supplementary Table 2). Based on the neutrophil gene expression patterns reported in literature, and log-fold values in our microarray data set, 5 differentially expressed genes from ACLF versus CLD, were selected for validation using qRT-PCR which included ELANE, MPO, CD177, OLFM4 and OLAH (Fig. 2D).
Overall trends for ACLF versus CLD were found to be conserved in the qRT-PCR analysis (Fig. 2D). ELANE, MPO and CD177 are three genes which have been linked to neutrophil pathogenicity in inflammatory disorders and were found to be significantly upregulated in ACLF PMN compared to CLD or HC (Fig. 2E).
Up regulated Since sepsis is a major etiology as well as complication in ACLF, gene expression values were stratified into sepsis versus sterile inflammation ( Supplementary Fig. 2). ELANE, MPO, CD177, OLFM4 and OLAH gene expression were found to be comparable in both the groups sepsis and sterile inflammation ( Supplementary  Fig. 2) indicating that neutrophil response to both bacterial infection and tissue injury involved upregulation of these genes.
A literature search for the functions of the above mentioned genes revealed that MPO is associated with inflammatory diseases such as cardiovascular disease and sepsis, whereas the role of the other neutrophilenriched genes such as ELANE, CD177 and OLFM4 are relatively poorly described. ELANE, MPO and CD177 were among the highest expressed genes in transcriptome meta-analyses of sepsis whole blood transcriptomes (Supplementary Table 11). MPO and CD177 but not ELANE, have been found to be associated with increased ROS production and enhanced phagocytosis in neutrophils. The precise role of CD177, OLFM4 and OLAH in granulocyte functions, is unknown. CD177 surface level expression is elevated in ACLF. CD177 is a neutrophil specific cell surface molecule which defines heterogeneous neutrophil populations. Since the CD177 gene was found to be highly overexpressed in ACLF neutrophils, we investigated the cell-surface protein level expression of CD177 in circulating neutrophils from ACLF patients. For this, the total percentage of CD16 + and CD66b + neutrophils were estimated from the whole blood of ACLF, CLD and Healthy (n = 10 per group) samples.  Figure 4A represents the gating strategy used for the assessment of CD16, CD66b and CD177. First, PMN were gated based on their characteristic position in the FSC-A/ SSC-A plot, followed by singlet gating. Singlets were assessed for two markers-CD16 (human FCγRIII) and CD66b. While CD16 is a marker present on most neutrophils, CD66b is an activation marker specific to mature neutrophils. A majority of the neutrophils in the CLD group were found to be CD66b + (almost 100%); ACLF showed a wider range and HC in our study population showed about 92% mature neutrophils (Fig. 4B). CD177 + cells were gated within the CD16 + CD66b + positive neutrophil population (lineage markers for granulocytes and neutrophils) for all samples. Percentage of CD177 + neutrophils as a proportion of total CD16 + CD66b + neutro-   www.nature.com/scientificreports/ phils were compared between the study groups. Percentage of CD177 + neutrophils in total CD16 + CD66b + neutrophils were found to be significantly higher in ACLF (~ 80%) as compared to CLD (~ 50%) (p value < 0.0001) and Healthy controls (~ 60%) (p value 0.005) (Fig. 4B,C). The values of CD16 + CD66b + enrichment and CD177 positivity for individual samples are provided in Supplementary Table 4.

Presence of CD177 + neutrophils in ACLF and CLD-AD post-mortem liver biopsies. Formalin
fixed paraffin embedded (FFPE) liver biopsies from deceased ACLF and CLD-AD (CLD with acute decompensation but without ACLF) were retrospectively retrieved and subjected to dual colour IHC for CD16 and CD177 in order to investigate the tissue localization of the CD177 + neutrophil sub-population (Fig. 5A-D). A limitation of this experiment was that "normal" or "CLD" control liver biopsies are ethically non-permissible at our centre in the absence of clinical indication for a biopsy in live patients, and therefore, cannot be included as controls. The CD177 stained neutrophils were represented by brown stain (black arrows) and the CD16 stained neutrophils, monocytes, macrophages, and Kupffer cells were represented by blue stain (blue arrows) (Fig. 5A,B). The manual counts, post validation was averaged for each cell type in a case and were expressed per 3.14 mm 2 , that is the area of each FOV of 10 × objective in this microscope (Supplementary Table 5). These data show that CD177 + neutrophils are detectable in the liver tissue of the deceased patients in both the ACLF and CLD-AD groups and CD16 + leukocytes were present at comparable levels in both groups (Fig. 5E,F; p values 0.1508 and 0.1625 respectively). Stratification of ACLF samples based on the presence or absence of sepsis demonstrated higher numbers of CD177 + neutrophils in the ACLF sepsis biopsies although CD16 + neutrophils were comparable (Fig. 5G,H, p value 0.0635 and p value 0.2222 respectively). In order to understand target cells with which CD177 + neutrophils may potentialy interact, cell specific expression data was extracted from human protein atlas for the well known CD177 binding partner PECAM1 (https:// www. prote inatl as. org/ ENSG0 00002 61371-PECAM1/ cellt ype) 26 . Among blood cells , classical and nonclassical monocytes, neutrophils, eosinophils had high expression of PECAM1 and CD4 naïve T cells, naïve and memory CD8 T cells, B cells and dendritic cells had low expression of PECAM 1 (Fig. 5I). Within the liver,
(B) 2 of 371 genes (0.54%) were common between the DEG from a recently published ACLF neutrophil dataset (GSE142254) and the eosinophil dataset revealed an overlap of 2 of 371 genes (0.54%), which was comparable.
(C) 94 genes were common in the comparative analysis of ACLF versus HC (our study) with the published ACLF neutrophil dataset (GSE142254; named as Neutrophil (ACLF vs HC)) and 62 were common between genes between ACLF versus CLD (our dataset) versus Neutrophil (ACLF vs HC). www.nature.com/scientificreports/ between recently published ACLF neutrophil dataset (GSE142254) and the SLE versus HC (GSE153781) and found that they shared 29 of 371 genes in common (Fig. 7C).

Overexpression of ELANE, MPO and CD177 in ACLF PMN.
We found significantly higher overexpression of ELANE, MPO and CD177 genes in neutrophils from ACLF in comparison to neutrophils in CLD (compensated) and healthy controls (Fig. 2E). Over-expression of ELANE, MPO and CD177 in ACLF neutrophils was associated with higher 28-day mortality (Fig. 3A-C). Among the upregulated genes in ACLF , CD177 expression was highest (Fig. 2D). Such high CD177 gene upregulation was associated with significant increase in the corresponding neutrophil cell surface CD177 protein expression in ACLF compared to CLD and HC (~ 80% neutrophils in ACLF vs ~ 50% in CLD vs ~ 55% in healthy control; One way ANOVA p = 0.0002 and ACLF vs CLD p value < 0.0001) as captured by flow cytometric evaluation (Fig. 4C ). The expression of ELANE, MPO and CD177 was not influenced by sex or the presence of bacterial infection (Supplementary Figs. 1A-C, and Supplementary Fig. 2A-C). ELANE, MPO and CD177 are among the highest expressed genes in a variety of inflammatory disease 24,25,[30][31][32][33][34][35][36] . Their high overexpression in ACLF patients as compared to controls, and their association with ACLF non-survivors suggests the involvement of neutrophil responses in ACLF pathogenesis leading to mortality. An independent study on ACLF blood immune signatures carried out in a European (French) population has also reported expansion of the CD177 + sub-population in ACLF 10 . This corroborates our observation in a genetically non-identical cohort of ACLF patients. Our study additionally shows that CD177 gene expression is associated with 28-day non-survivors. CD177 + neutrophils have been shown to be associated with increased ELANE and MPO gene expression, in agreement with our observations 37 . Therefore ELANE, MPO and CD177 may form a three-gene signature that characterize a pathogenic neutrophil subset in neutrophil mediated pathogenesis in ACLF. Quantitative gene expression values of ELANE, MPO and CD177; or quantitative flowcytometric expression of CD177 on neutrophils may provide quantitative measures for evaluating the extent of inflammation and as a prognostic biomarker for short term mortality.
Other genes found to be overexpressed in ACLF granulocytes were OLFM4 (Olfactomedin 4) and OLAH (Oleoyl-ACP hydrolase). Although the precise mechanistic role of OLFM4 in neutrophil is unknown, it has been shown to have a strong association with highly inflammatory conditions. OLFM4 defines a neutrophil subset in healthy as well as diseased individuals 38 . While OLFM4 has not been shown to directly affect neutrophil functions such as phagocytosis or tissue migration, OLFM4+ neutrophil subsets have been found to be highly expanded and associated with mortality in inflammatory diseases with high mortality and morbidity such as sepsis, intestinal ischemia/ reperfusion injury 39 and, hemorrhagic shock. Our study reports the elevation of OLFM4 mRNA. OLFM4 KO mice have been shown to have reduced ROS burst in neutrophils suggesting a role for OLFM4 in promoting ROS mediated functions 40 . However the role of OLAH in granulocyte function has not been described. OLAH (Oleoyl ACP hydrolase) catalyzes the release of fatty acids from fatty acid synthase enzyme. Among blood cells, it is neutrophil enriched (Supplementary Table 11). It can be postulated that the upregulation of OLAH in ACLF granulocytes might indicate upregulation of fatty acid release that can enter TCA cycle thereby generating high levels of ATP for neutrophil function. It has been shown that ATP generation through glycolysis as well as fatty acid metabolism through TCA cycle are important in ATP-requiring neutrophil functions such as NETosis 41 . While the precise role of OLAH in neutrophil function still needs to be deciphered, we think that this enzyme may be involved in regulating energy production to facilitate neutrophil function. 2. CD177 + Neutrophils are present in the liver tissue of deceased ACLF and CLD-AD patients. CD177 is a GPIanchored cell surface glycoprotein which has been shown to bind the cell adhesion molecule PECAM-1 on endothelial cells 42 . CD177 participates in cell adhesion, tissue transmigration and neutrophil degranulation 43 . However, its exact role in tissue transmigration is controversial and it is not clear whether CD177 + neutrophils are capable of reaching target tissue sites, or not 44,45 . We retrieved archived FFPE liver biopsies from deceased ACLF and CLD-AD patients and carried out dual-colour IHC staining for CD16 and CD177 cell surface proteins. In the liver, CD16 is expressed by all cells of myeloid origin such as neutrophils, macrophages, Kupffer cells, and NK cells and CD177 is a neutrophil specific marker. Therefore CD16 + CD177 + double positives were scored as neutrophils in the liver biopsies (Fig. 5). Liver biopsies isolated from both deceased ACLF and CLD-AD patients were found to have CD177 + neutrophils suggesting that these cells are capable of tissue transmigration (Fig. 5). We also observed that in ACLF, patients with sepsis had greater number of CD177 + neutrophils in liver tissues as compared to patients without detectable sepsis (sterile inflammation) ( Fig. 5G; p = 0.0635). Our observation is corroborated by recent studies carried out in periodontitis as a model for inflammation, that show that CD177 + neutrophils preferentially travel towards tissue sites with microbial inflammation as compared to tissue sites with sterile inflammation, although the levels of CD177 + neutrophils in circulation are elevated in both 44 . A substantial body of work suggests that the primary binding partner for CD177 on neutrophil surface is the platelet/ endothelial cell adhesion molecule 1 (PECAM1) [46][47][48] . While PECAM1 has been established as a hallmark of endothelial cells, it was not clear if PECAM1 could also be expressed by other cell types. Hence, we investigated the expression of PECAM1 in the Human Protein Atlas database. We found that while endothelial cells had high expression of PECAM1 mRNA, other cell types also expressed PECAM1 mRNA at varying levels (Fig. 5I,J) www.nature.com/scientificreports/ (Fig. 5I). Within the liver, endothelial cells, hepatic stellate cells (Ito cells), Kupffer cells, B cells and T cells were found to express PECAM1 mRNA (Fig. 5J). While this is mRNA level expression that needs to be confirmed at the level of protein expression, the analysis suggests that non-endothelial cell types including hepatic stellate cells, B and T lymphocytes might be capable of synthesizing PECAM1 and therefore, may interact with CD177 + neutrophils. These interactions may further regulate inflammation or tissue injury within the liver. This is an area that needs further investigation and validation. The precise molecular function of the CD177 + neutrophils is not yet known. However, based on its presence in the liver tissue from our study, along with its reported ability to preferentially get recruited to septic tissues, adhesion to endothelial PECAM1 and, association with increased granule enzyme genes ELANE and MPO, we hypothesize that CD177 + neutrophils expressing high levels of cytotoxic enzymes ELANE and MPO reach the liver tissue in severe ACLF. However, the precise role of CD177 + neutrophils, and association with sepsis and the extensive cellular damage that is seen in ACLF, need further investigation 44 . Our study suggests that ACLF 28-day non-survivors have significantly higher levels of CD177 + neutrophils in circulation, indicating a role of the CD177 + neutrophil sub-population in ACLF pathogenesis, that needs to be further explored. 3. Upregulation of Inflammatory Pathways in ACLF PMN. Pathways analysis of overall transcriptomic signatures in ACLF versus CLD revealed the upregulation of inflammatory modules in ACLF PMN that resembled gene expression modules in several inflammatory diseases such as SLE, RA and IBD suggesting commonalities in PMN activation in these diseases ( Supplementary Fig. 4). Neutrophil degranulation pathway was a part of the shared gene signature among the inflammatory diseases and may provide a basis for a common therapeutic target in inflammatory diseases that share this signature (Supplementary Table 10). Further, a comparative analysis of ACLF DEG from our dataset with ACLF DEG reported in the recently published study by Weiss et al., 2021 (GSE142254) revealed several shared DEG (Fig. 6C). These genes likely represent the most consistently expressed PMN genes which are differentially expressed in ACLF across study populations and ACLF definitions (Fig. 6C). We also found that our ACLF PMN dataset shared gene expression signatures with the transcriptome of enriched neutrophils from SLE patients that include the genes ELANE, MPO and CD177. This suggests that these three genes might constitute a damaging inflammatory gene expression signature that is common across diseases where inflammation mediated damage may be occurring.
A limitation of our study is the limited sample sizes included in various experiments. However, the observations made in the study, particularly in the context of ELANE, MPO and CD177, are supported by significant trends in spite of low sample sizes and therefore, cannot be ignored. We report novel associations between neutrophil-specific genes and ACLF outcomes. We also address the controversial question about the ability of CD177 + neutrophils to transmigrate to sites of tissue injury. Our study shows that they can indeed travel to the liver tissue in deceased ACLF and CLD-AD patients.
A major challenge in ACLF research is the heterogeneity of the disease in terms of variable clinical definitions of ACLF across continents, chronic and acute etiologies, presence or absence of sepsis and the involvement of multi-organ dysfunction. At the time of the acute super added liver injury, the functional liver reserve also vary which might influence the inflammatory and immune dysregulation. These variables are characteristic of ACLF and unfortunately cannot be avoided. This complicates studies on the pathogenesis of ACLF due the involvement of different kinds of PAMPS and DAMPS, interaction of different molecular and physiological processes 49 . Therefore, it is difficult to extrapolate findings from one study to another and this has been a major challenge in ACLF biomarker research 50,51 . However, our observations which are made in an Indian cohort are corroborated by published studies carried out on genetically non-identical French population, particularly in the context of overexpression of neutrophil CD177 10 . This suggests that neutrophil specific CD177 expression may be a major conserved pathogenic signature in ACLF. CD177 expression has also been found to be high in ACLF whole blood transcriptome which suggests that whole blood gene expression profiling might be a good strategy for biomarker development 10 . While further investigations are needed to understand the mechanistic relevance of these genes and pathways in ACLF; our study provides a rational basis for further exploration of the ELANE, MPO, CD177 signature as potential biomarkers in predicting ACLF outcomes; as well as highlights these neutrophil granule genes and pathways as possible points of therapeutic intervention in ACLF.