Iron-Overload triggers ADAM-17 mediated inflammation in Severe Alcoholic Hepatitis

Severe alcoholic hepatitis (SAH) is associated with iron accumulation in hepatocytes/macrophages. This possibly correlates with inflammation and stress but the exact mechanism still remains obscure. To understand the role of iron and the mechanisms of systemic iron-overload, a transcriptomic study of liver and Peripheral Blood -Mononuclear-Cells (PBMCs) was undertaken in SAH patients, with and without hepatic iron-overload. Our results show that iron-overload in hepatocytes/macrophages is due to an increased expression of iron-loading receptors and CD163 signaling cascade. Increase in labile iron pool induces expression of iron-loading, oxidative-stress and inflammatory genes along with expression of CD163 and ADAM17. Increased liver iron correlated with circulatory iron, TNF-α, macrophage activation (sCD163) and peroxide-stress in CD163+macrophages in patients who were iron-overloaded and died. Circulatory TNF-α and sCD163 levels were associated with poor outcome. Temporal iron/Fenton stress induced in healthy monocyte-derived-macrophage (MDM)/Tohoku-Hospital-Pediatrics-1(THP1) cells showed higher expression of iron-regulatory, inflammatory and oxidative-stress genes. These genes could be suppressed by iron-chelation. These results suggest that iron mediates inflammation through ADAM17 induction, resulting in macrophage activation and increased shedding of TNF-α and sCD163. These events could be inhibited with iron chelation or with ADAM17-blockade, postulating a therapeutic strategy for SAH patients with iron overload.

These observations suggest a direct association between iron-overload and induction of genes related to iron loading, oxidative-stress, inflammation and metalloproteases in the liver and PBMC of SAH patients.
SAH patients show increased CD163 and ADAM17 expression. Validation-phase: Systemic iron-overload associated genes identified in the liver and PBMC transcriptome were validated in the liver biopsy and plasma of study groups, and by in-vivo and in-vitro analyses ( Fig. 2A). A total of 16 genes were dysregulated in the liver and PBMC transcriptome of Group A versus B. These genes were linked to cellular iron homeostasis, response to hydrogen peroxide and notch receptor signaling (C-score > 2; Fig. 2B). Of these only 3 genes (CD163, ADAM17, and TMED7) were significantly linked to systemic iron-overload based on multivariate projection model analyses (Fig. 2C, Supplementary Table 8). The expression of CD163 and ADAM17 were validated on the representative liver sections of Group A and B patients, while TMED7 was excluded from analysis due to its variable expression (Supplementary Table 8). The liver section of Group A showed a significant increase in ADAM17 and CD163 expression in the sinusoidal spaces with no alteration in ADAM10 expression (p < 0.05; Fig. 2D). To establish a relationship between macrophage iron-overload and expression of CD163 and ADAM 17, the liver of Group A patients was compared with AC patients. In Group A, the expression of CD68, CD163 and ADAM17 was higher (p < 0.05, Fig. 2E).
Serum iron, sCD163, He-Hp complex and TNF-Alpha correlate with outcome of SAH patients. In our study, the levels of circulating iron, sCD163, TNF-α directly correlated with hepatic encephalopathy and inversely with disease severity scores and survival (r 2 > −0.3, p < 0.05, Supplementary Table 9). The levels of serum iron, ferritin, sCD163, TNF-α were significantly higher in SAH patients who did not survive (Fig. 3A,B, Table 1). In fact, sCD163 and TNF-α levels had high predictability for mortality (AUROC >0.90) ( Table 2). Univariate and multivariate Cox regression analyses identified a significant association of sCD163, TNF-α, and hepatic encephalopathy (≥grade 2) with mortality (HR > 1.5, p < 0.05) in SAH patients (Table 2). Since ADAM17 expression is associated with ectodomain shedding of TNF-α and sCD163 7 , our results suggest that the increase in sCD163 and TNF-α is associated with iron-overload and activation of ADAM17 in SAH patients.

CD163+ macrophages of SAH non-survivors exhibit iron linked oxidative stress and inflammation (In-vivo analysis).
Recently our group showed that CD163 +ve macrophages contribute to the circulating iron pool 12,13 . Patients of SAH who do not survive, showed higher CD163 +ve macrophages ( Supplementary Fig. 2) and higher       . Caspase dependent cell death and autophagy genes were relatively higher in survivors (Fig. 3F). These results suggest a prominent dysregulation of iron homeostasis, oxidative stress, inflammatory cytokines, induction of ADAM17 and CD163 signaling cascade in CD163 + macrophages of non-survivors.

Iron chelator reverses the effect of Iron and Fenton in healthy Monocyte Derived Macrophages (MDMs).
To understand the role of circulating iron or Fenton in the induction of metalloproteases (ADAM17), inflammation and stress, MDMs from healthy controls were treated with iron/Fenton reagent at increasing concentrations in presence or absence of iron chelator, deferiprone. Fenton was more prominent in stimulation of genes related to oxidative-response, iron-regulation, matalloproteases, ER-stress and autophagy as compared to iron alone ( Supplementary Fig. 3). Deferiprone treatment could significantly (p < 0.05) ameliorate expression of genes linked to superoxide/antioxidant response, iron-homeostasis, and inflammatory-cytokines (TNF-α, IL-7, IL-8), (Fig. 4A,B). Iron alone failed to induce other metalloproteases (ADAM10) but was able to significantly induce ADAM17 and genes associated to macrophage iron production (CD163, HMOX1) which got neutralized by deferiprone (Fig. 4C). Fenton on the other hand, induced expression of both ADAM10, ADAM17 and macrophage iron processing genes (CD163, HP, HMOX1), which also got reduced (p < 0.05) under deferiprone (Fig. 4D). Iron/Fenton significantly increased apoptosis, ER-stress and autophagy genes which got ameliorated (p < 0.05) under deferiprone (Fig. 4F,E). These results clearly suggest that iron invariably induces ADAM17 and genes linked to oxidative-stress, iron-regulation, apoptosis which get more pronounced in presence of peroxide stress but can be neutralized if iron is chelated.
Iron chelation can reverse effect of iron or Fenton reagent in THP-1 derived macrophages. The result of THP-1 stimulation assay (induction by Iron/Fenton and neutralization by deferiprone) was similar to the MDMs stimulation assay (Supplementary Fig. 3). In brief, 50 mM iron was able to (FC > 1.5, p < 0.05) induce genes linked to superoxide generation (CYBA), antioxidants (GSH), iron-homeostasis (FERROPORTIN, HEPCIDIN), and inflammation (IL-10, IL-8), which were reduced under deferiprone (Fig. 5A). Again, iron in the presence of peroxide stress (Fenton) was more potent in the induction of oxidative/superoxide response, iron-homeostasis, inflammatory-cytokines, and deferiprone could inhibit (p < 0.05) their expression (Fig. 5B). Both iron and Fenton stimulated expression of ADAM17, CD163 and linked genes (p < 0.05, FC >1.5) which were significantly reduced under deferiprone (Fig. 5C,D). Iron/Fenton stress significantly increased the protein levels of ADAM17 (Fig. 5C,D lower panel) which got reduced with chelation. Iron chelation resulted in higher cellular expression of CD163 and ADAM 10 (Fig. 5C,D) suggesting that expression of ADAM17, ADAM10 and CD163 are under the influence of iron regulation. Iron/Fenton stress also increased CASP1 and ATG5 expression which was suppressed with Deferiprone treatment (Fig. 5E,F). These observations suggest that in THP-1 macrophages, iron/Fenton significantly increases iron loading, oxidative stress, pyroptosis and autophagy. Further, the cellular level of CD163 is regulated by iron-load via activation of ADAM17 in both MDMs and THP-1 derived macrophages.
ADAM 17 inhibition decreases TNF-α and sCD163 expression in THP1 macrophages. ADAM17 inhibition was performed on THP1 macrophages under Iron/Fenton stress to establish the causality between iron-mediated activation of ADAM17, sCD163 and inflammation. Cellular expression of CD163 and TNF-α were increased, while sCD163 and TNF-α levels in the cell-supernatant were decreased (p < 0.05) under ADAM17 inhibition irrespective of iron/Fenton stress (Fig. 6A,B). Expression of inflammatory genes was also lower and iron regulatory genes were higher under ADAM17 inhibition irrespective of iron/Fenton stress (Fig. 6C) Figure 4). Similar observations were seen in response to temporal iron/Fenton stress (Supplementary Figure 5, 6, 7). The soluble level of TNF-R1 was unchanged while the levels of TNF-R2 were significantly increased under ADAM 17 (TAPI-1) inhibition (Fig. 6D) in the absence of iron stress. However, under iron stress, TNF-R1 levels were significantly lower in TAPI-1 treated cell supernatant (Fig. 6D). Our data clearly shows that TAPI-1 has a differential activity towards TNF-R1 and TNF-R2 in the presence or absence of iron. Since the activation of TNF-R1 is mainly involved in the activation of inflammation, use of TAPI-1 may prevent iron induced inflammation. ADAM 17 inhibition by TAPI-1 increased membrane TNF-α levels in the macrophages irrespective of iron/Fenton stress ( Fig. 6E and Supplementary Fig. 8). These observations suggest that ADAM 17 inhibition modulates inflammation by inhibiting TNF-α shedding in macrophages.

Discussion
Severe alcoholic hepatitis is accompanied by significant hepatocellular necrosis and in a proportion of patients, increased iron load in hepatocytes and macrophages 18 . Our results based on the RNA-Seq on liver tissue and PBMCs, clearly show that genes linked to iron-loading, oxidative-stress and inflammation, including expression of CD163, ADAM17 are over expressed in SAH patients with iron-overload and this plays a major role in progressive hepatic injury and correlates with non-survival. Furthermore, iron chelation prevented induction of ADAM17 expression and mediated an increase in expression of CD163, both at the gene and protein level. Expression of genes linked to inflammation and oxidative-stress were decreased with iron chelation providing an explanation that chelation prevents TNF-α and CD163 shedding and suppresses inflammation via reducing the expression of ADAM17. We validated these results in a large validation cohort and confirmed this by in-vivo and in-vitro studies.
In the present study, the increase in the liver iron in SAH was due to a marked increase in expression of transferrin receptor protein 1 (TFRC) and decrease in FPN1 (ferroportin) which is consistent with our previous study 13 . Liver and PBMC transcriptome showed significant increase in secondary iron loading genes (CD163, HP, HMOX1), indicating that iron overload in SAH is via activation of CD163 signaling 12 . Iron-overload in liver/ PBMC mediated iron dysregulation and significantly induced genes linked to oxidative/antioxidant activity, proteolytic cleavage of receptors, iron-homeostasis, ROS/hydrogen peroxide metabolism 18,24 . In SAH patients with iron-overload, pathways linked to TNF-α maturation and ROS detoxification were augmented. This indicates that iron accumulation promotes inflammation and oxidative damage 10,21 ; dominantly by activation of TNF-α signaling.
Liver and PBMC transcriptome identified 16 genes linked to cellular iron homeostasis, notch receptor signaling and secondary cleavage of receptors. Univariate and multivariate projection analysis identified ADAM17 and CD163 as dominantly upregulated genes related to systemic iron-overloaded in SAH. Immuno-histochemistry validated a significant increase in ADAM17 and CD163 levels with minimal alteration of ADAM10 expression in the Kupffer cells (Fig. 2D). This suggests that in SAH, iron load may play a significant role in the induction of ADAM17 and CD163 expression 10 . This observation was further confirmed in thalassemia patients who have a high iron-load (Fig. 2F).
Increase in ADAM17 expression increases cleavage of CD163 receptor 7 . Livers of patients with iron-overload, showed significantly increased levels of CD163 and ADAM17, correlating with the increase in sCD163 and TNF-α levels 11 . This could be due to activation of ADAM17 by iron (Fe ++ ) replacing zinc (zn ++ ) from its catalytic domain. In our study, the circulatory levels of TNF-α, sCD163, iron and ferritin were significantly higher along with increase in ADAM17 expression in SAH patients who died as compared to those who survived. In addition, high sCD163, TNF-α levels and presence of hepatic encephalopathy were significantly associated with mortality in SAH.
Soluble CD163 works as an anti-inflammatory molecule while TNF-α is pro-inflammatory 7,10 . Iron accumulation in SAH patients is by iron import receptor or by processing of He-Hp complex and activation of HMOX1 in CD163 + macrophage 12 . We tried to correlate the iron disbalance and transcriptomic profile with survival in patients with SAH. In our study He-Hp complex was significantly downregulated in non-survivors with concomitant upregulation of CD163, HMOX1, HP, and CD68 genes, suggesting a hyper-regulated CD163 signaling cascade in hepatic lineage macrophages. This increase in CD68 expression on circulating macrophages could have many functions and warrants further studies. This observation was validated in PBMCs where there was an increase in CD163 receptor expression. SAH patients showed increase in the intracellular ROS/ peroxide pool, which serves as a precursor for Fenton reaction 22 . We have earlier shown that increase in CD163 + cells are associated with an increase in free labile iron pool 13 .
To understand the role of CD163 + macrophages in iron processing, oxidative-stress and ADAM17 induction, gene expression analysis was performed on purified CD163 + and CD163cells from SAH patients. A significant increase in oxidative-stress, antioxidant response and iron processing genes in response to the circulatory iron stress was observed in CD163 + macrophages of non-survivors. Result of our current study show that there is an increase in labile iron pool in CD163 + macrophages of non-survivors 13 and this might be due to increase activation of DMT1 receptor and secondary iron loading. CD163 + macrophages of non-survivors documented significant increase in the expression of ADAM17 and associated genes while expression of genes linked to ER-stress, oxidative stress (NOX1, NOXA1) and inflammation (TNF-α, ADAM17). Increased expression of ADAM17 increases sCD163 and TNF-α in circulation, which correlates to mortality in SAH patients. Iron chelation and ADAM17 inhibition remarkably neutralizes these alteration and could be used as attractive therapeutic option in such patients. Pathways marked with [***] is significantly increased in SAH patients particularly in those with iron-load.
To ascertain whether iron alone or iron in presence of peroxidative stress (Fenton) is responsible for gene expression alteration observed in CD163 + macrophages, we subjected MDMs and THP-1 derived macrophages to iron/Fenton stress at increasing concentrations in the presence or absence of iron chelator, deferiprone. A temporal and dose-dependent increase in expression of genes related to oxidative-stress, iron-regulation, macrophage iron processing, and metalloproteases was seen in macrophages which were neutralized under deferiprone, suggesting a central role of iron in inflammation, oxidative-stress and metalloprotease activation 24 . Iron and Fenton both induced the protein expression of ADAM17, CD163 and ADAM10. Under iron chelation, the expression of ADAM17 decreased and the expression of CD163 increased. Blocking of ADAM 17 increased cellular expression of CD163, TNF-α and reduced the inflammation irrespective of iron/Fenton stress. These novel observations suggest that Iron/Fenton produces inflammation through ADAM17 and iron-chelation reduces ADAM17 induction and inflammation. This also signifies that ADAM17 and CD163 expression are under iron regulation 10 and inversely correlate with each other and could modulate inflammation in SAH patients.
In summary, the results of this novel study clearly show that in patients with SAH, iron overload induces a Fenton reaction and a vicious cycle of oxidative stress 21,26 . Increased circulatory iron stimulates macrophage iron loading via activation of TFRC, TFR2, DMT1 and induction of the CD163 signaling cascade. Intracellular labile iron pool generates peroxide stress and induces genes linked to iron homeostasis (hepcidin, ferritin), inflammation (TNF-α and ADAM17) and oxidative-stress. This results in increased circulatory sCD163 and TNF-α levels (Fig. 6F). Iron chelation on the other hand, reduces the expression of ADAM17 and TNF-α underlining the fact that iron plays a major role in inflammation and progression of disease to organ failure in SAH. It would be worthwhile to explore iron and ADAM17 as attractive therapeutic targets in these patients.

Methods
Patients. In this prospective study, 120 liver-biopsy proven SAH patients were enrolled between January 2013 to January 2015. Nineteen patients were excluded due to hepatocellular carcinoma (10), portal vein thrombosis (5) and previous history of plasmapheresis. SAH was diagnosed by histological criteria and Maddery's discriminant function (DF) of >32 23 . Alcoholic Cirrhosis (AC) was diagnosed on history of chronic heavy alcohol intake (with >1-month alcohol abstinence) and a combination of clinical, biochemical, endoscopic and radiological criteria 27 . Healthy controls (HC) had no evidence of present/past liver disease ( Supplementary Fig. 1). For comparison of iron-load (n = 5; Scheuer-grade ≥3+), thalassemic patients were included as positive control.
The patient groups received standard medical therapy based on their clinical status, including albumin, lactulose, bowel washes, antibiotics and pentoxifylline. At the time of patient enrolment, treatment of SAH with corticosteroid was not a part of standard of care at our center and hence, none of the enrolled patients received corticosteroids. Patients were followed-up for a period of three months or until death. Only baseline samples were analyzed and correlated with outcomes. The laboratory staff were unaware of the clinical details. The study was approved by the institutional ethical committee/ institutional review board Institute of liver and biliary science (ILBS) New Delhi, India and written informed consent was obtained in all cases. Further all the experiments were performed in accordance with relevant guidelines and regulations of the concerned ethical committee.

Methods
Discovery. RNA-Seq was performed in the initial 15 SAH patients, which were further grouped into SAH with iron load (SAH-IO; Scheuer-grade ≥1+, Group A: n = 5) and SAH with no-iron load (SAH-NIO; Group B: n = 10) based on histological evidences. One sample was excluded from the last group due to poor liver RNA quality (RIN < 7). Differentially expressed genes (DEG) identified in Gr.A versus B were cherry picked and validated in the validation cohort ( Supplementary Fig. 1).
Validation. Target genes and linked mechanisms were identified after univariate and multivariate projection analysis and were validated in the representative liver tissue from 5 SAH patients and 5 alcoholic cirrhosis using immuno-histochemistry. Circulatory protein levels were validated in 100 SAH patients and compared to 20 alcoholic cirrhosis and 20 controls (Supplementary Fig. 1). In-vivo and in-vitro experiments were performed to understand the mechanisms of iron-loading, inflammation, and stress in the macrophages of SAH patients.
Transcriptomic analysis. RNA-Seq was performed in the liver and PBMC samples from both Group A and B patients. Heat map for the differentially expressed genes (DEG) was generated on 'R' , and supervised clustering was performed only on significantly (p < 0.05) modulated genes. GO and Pathway analysis was performed using Enrichr 28 . Genes segregating Group A from B were validated for expression and mechanisms using immuno-histochemistry, ELISA or Western-blot (Supplementary methods).