Toll-like receptor polymorphisms compromise the inflammatory response against bacterial antigen translocation in cirrhosis

Bacterial translocation is associated with clinically relevant complications in cirrhosis. We evaluated the effect of toll-like receptor polymorphisms in the soluble response against these episodes. Consecutive patients with cirrhosis and ascitic fluid were distributed by TLR2 rs4696480, TLR4 rs4986790, and TLR9 rs187084 single-nucleotide polymorphisms. Lipoteichoic acid, lipopolyssaccharide, bacterial-DNA, pro-inflammatory cytokines and nitric oxide levels were quantified in serum samples. In vitro response against specific ligands in variant TLR genotypes was evaluated. One hundred and fourteen patients were included. Variant TLR-2, TLR-4 and TLR-9 SNP genotypes were associated with significantly increased serum levels of LTA, LPS and bacterial-DNA. TNF-α, IL-6 and nitric oxide serum levels were significantly decreased in all variant TLR genotyped patients. Cytokine levels were significantly less upregulated in response to specific TLR-ligands in patients with all variant vs wildtype TLR genotypes. Although in vitro gene expression levels of all wildtype and variant TLRs were similar, MyD88 and NFkB were significantly downregulated in cells from TLR-variant genotyped patients in response to their ligands. Variant TLR genotypes are associated with an increased circulating antigen burden and a decreased proinflammatory response in cirrhosis. This immunodeficiency may facilitate bacteria-related complications in cirrhosis and enhance TLR targeting for its management.

The translocation of commensal bacterial antigens into blood of patients with decompensated cirrhosis is a frequent event that triggers relevant clinical complications. Several studies have related the transient passage of bacterial products into the systemic circulation of these patients with the exacerbation of the inflammatory response [1][2][3][4] , the prediction of infections 5 , an increased risk of hemodynamic disturbances 4,6 or even death 7 .
Bacterial antigenic products are specifically recognized by a host's family of receptors, both at the membrane and the intracellular levels. The Toll Like Receptor (TLR) family comprises different receptors that specifically bind unique bacterial products, launching an inflammatory signalling cascade and the mechanisms for these bacterial products clearance 8,9 . TLR-2 is the specific receptor for lipoteichoic acid (LTA), a product derived from gram-positive bacteria, which are also able to translocate and differentiate a specific immune response 10 . TLR-4 and TLR-9 are receptors for lipopolysaccharide (LPS) 11 and bacterial DNA 12 respectively, two of the most immunogenic bacterial products. Several studies have described the deep effects of endotoxemia or the presence of bacterial DNA on the clinical outcome of decompensated cirrhosis.
Several polymorphisms in these TLR genes have been described and associated with immune alterations or clinical complications in cirrhosis and other inflammatory-related disorders. A polymorphism in the TLR-2 promoter has been linked to an increased risk for spontaneous bacterial peritonitis in cirrhosis 13 . The TLR-4 rs4986790 polymorphism has been found to predispose to infections in cirrhotic patients 14 and to an LPS hyporresponsiveness 15 . These results have brought the attention into TLRs as potential targets in chronic liver disease 16 .
On the other hand, the TLR-9 rs187084 polymorphism has been described to be functionally relevant by downregulating TLR-9 expression 17 .
Therefore, the genetic background may be a relevant potential modulator of the immune response in bacterial-derived complications of cirrhosis. The aim of the present study was to evaluate the incidence of relevant polymorphisms in TLR-2, TLR-4 and TLR-9, specific receptors of bacterial products which are frequently translocated from the gut in patients with cirrhosis.

Results
Characteristics of patients. One hundred and fourteen patients with cirrhosis and non-infected AF were included in the study. All clinical and analytical characteristics of patients are detailed in Table 1. Mean age was 60 ± 10 and 68.4% were male. Main etiology of cirrhosis was alcohol, and mean Child-Pugh score was 9.2 ± 1.8. Thirty-six patients (31.5%) were on beta-blockers and 40 patients (35%) were taking proton pump inhibitors (PPIs).
We followed up patients for 6 months. Four patients died (all due to acute on chronic liver failure) during this time and 8 patients had an infection (1 SBP and 7 non-SBP [3 urinary tract, 2 clostridium dificile cholitis and 2 pneumonia]). We couldn't find any statistically significant relationship between any TLR variant genotype and the development of infections. However, 6 out of the 8 infectious episodes were present in patients bearing 2 or more TLR variant genotypes.
TLR polymorphisms and bacterial antigen detection in patients with cirrhosis. We evaluated the incidence of 3 well-known polymorphisms in TLR-2, TLR-4 and TLR-9 in patients with cirrhosis. The distribution of genotypes and allelic frequencies in patients and controls are shown in Table 2. All variants were found to be in the Hardy-Weinberg equilibrium in the controls. Of interest, no homozygous variant genotype for TLR-4 rs4986790 was found in the cirrhotic population. Figure 1 shows the concentration of different bacterial antigens according to the polymorphisms studied in their specific receptor. As can be observed, the loss of the wild-type allele in the studied polymorphisms of TLR-2, TLR-4 and TLR-9 was associated with significantly increased serum levels of LTA, LPS and bacterial DNA, ( Fig. 1A,B and C, respectively). Patients with the homozigotic variant showed the highest load of circulating bacterial antigens in blood samples.
In the case of TLR-9 rs187084, although the variant genotype was associated with an increased amplified bacterial DNA load in patients' samples, it was not associated with an increased rate of bacterial translocation, as represented by the bars in Fig. 1C. There were not statistical differences, either, in the sequenced species from bacterial DNA between patients distributed by TLR-9 rs187084 genotype, with a similar percentage of gram-positive microorganisms in all three groups (Supplementary Table 1). Clinical and analytical characteristics of patients distributed by each TLR polymorphism genotype showed no clinical or analytical differences (data not shown).

Soluble inflammatory response is influenced by TLR polymorphisms.
Serum pro-inflammatory mediators were evaluated in all patients and compared by the genotype of studied polymorphisms. Figure 2 shows that TNF-α and IL-6 were significantly decreased in patients bearing the variant genotype of TLR-2 rs4696480 ( Fig. 2A and E). When distributing the cytokine response by the variant TLR-4 rs4986790 ( Fig. 2B and F) and by TLR-9 rs187084 ( Fig. 2C and G) genotypes, both cytokines revealed a similar behaviour to that shown by TLR-2 rs4696480 genotype, showing significantly decreased serum levels of both mediators in patients with the variant genotypes. Patients were further distributed by the presence of detectable bacterial DNA in blood. The differences in TNF-α and IL-6 levels in patients distributed by TLR-9 rs187084 genotype were, as expected, due to those with circulating bacterial DNA in blood ( Fig. 2D and H).
A potent immune modulator such as NOx showed similar results to those observed for TNF-α and IL-6. The wild-type genotypes of all TLRs were significantly associated with increased serum NOx levels ( Fig. 3A-C). As shown for the cytokine pattern, the differences in NOx levels in TLR-9 rs187084 genotyped patients were associated with circulating bacterial DNA in blood (Fig. 3D).
We further analysed the inflammatory milieu in the serum of patients grouped by the number of variant TLR genotypes. As observed in Table 3, the accumulation of variant TLR genotypes was associated with significantly decreased pro-inflammatory cytokine and NOx levels.  Immunodeficiency in decompensated cirrhosis is associated with variant TLR-2, TLR-4 and TLR-9 genotypes. We evaluated the in vitro response to specific TLR ligands to confirm the downregulation of the inflammatory milieu in patients and donors bearing variant TLR genotypes. TNF-α , IL-6 and IL-10 levels in patients' PMNs supernatants were significantly upregulated in all cases after stimulation with Pam3-Cys, LPS and CpGs compared with resting conditions. However, TNF-α and IL-6 levels were significantly lower and IL-10 was significantly higher in response to TLR-specific ligands in patients with variant TLR-2 rs4696480, TLR-4 rs4986790 and TLR-9 rs187084 genotypes compared with those bearing the wild-type alleles (Table 4). Results on donors are shown in Table 5 and, as well as for patients, the presence of variant genotypes in all studied TLRs was associated with significantly reduced pro-inflammatory cytokine levels and increased IL-10 compared with the wild-type genotypes. Interestingly, the gene expression levels of variant TLR-2 ( Fig. 4A), TLR-4 ( Fig. 4B), and TLR-9 ( Fig. 4C) were not significantly different in response to their specific ligands compared with those observed in cells from patients with the wild-type alleles. However, when evaluating the gene expression levels of downstream signalling molecules, the presence of studied SNPs in TLR-2, TLR-4 and TLR-9 were associated with a significant decrease in MyD88 (Fig. 4D-F) and NFkB (Fig. 4G-I) compared with cells from patients with wildtype alleles after stimulation with their respective specific ligands. Statistical correlations between MyD88 and NFkB gene expression levels and the concentration of in vitro secreted cytokines are shown in Table 6.

Discussion
In this study, we demonstrate that variant genotypes in TLR-2, TLR-4 and TLR-9 genes are associated with an increased bacterial antigen burden and a decreased pro-inflammatory cytokine profile in blood of patients with cirrhosis and non-infected AF, suggesting that these genetic variants may compromise bacterial antigen interaction with their specific receptors and limit the innate soluble inflammatory response in these patients. This genetically derived immunodeficiency might have consequences in bacterial antigen clearance and contribute to the clinically relevant complications that are frequently developed in patients with cirrhosis.
A compromised immunological status is one of the three main aspects classically proposed for the development of bacterial antigen translocation episodes in cirrhosis, which in turn, is proposed to be either responsible or a contributor to life-threatening complications in these patients such as SBP, hepatic encephalopathy or   haemodynamic disturbances. TLRs take part in the specific recognition of antigenic molecules and trigger early events in the immunological response. We have evaluated the effect of TLR2 rs4696480, TLR4 rs4986790 and TLR9 rs187084 polymorphisms in the bacterial antigen load and the pro-inflammatory mediator levels in the blood of a consecutive series of patients with cirrhosis and AF. We first show that TLR variant genotypes are associated with significantly increased serum levels of their specific antigenic ligands (LTA, LPS and bacterial DNA, respectively). This would be in line with previous findings showing an increased risk of SBP in patients with cirrhosis bearing TLR-2 polymorphisms 13 or the predisposition to infections in cirrhotic patients with a variant TLR-4 genotype 14 . Interestingly, although the rate of bacterial DNA translocation is not increased in patients with a variant TLR9 genotype, there is a significant increment in the amount of amplified bacterial DNA in these patients compared with those bearing the wild-type TLR-9 genotype. This is relevant, as it has been proven that the grade of soluble inflammatory response is significantly affected by bacterial DNA concentrations in patients with cirrhosis 18 .
However, an important result of the study is that evaluated TLR variant genotypes show a reduced secretion of pro-inflammatory mediators in the serum of cirrhotic patients, despite their significantly increased antigenic burden (Fig. 2). Moreover, the presence of several variant TLR polymorphisms is associated with a progressive, significantly reduced inflammatory milieu. To confirm our results, we evaluated the soluble response to TLR-specific ligands in in vitro cultured neutrophils from cirrhotic patients and donors with different TLR genotypes. As shown in vivo, although TLR stimulation resulted in significantly increased supernatant levels of pro-inflammatory cytokines compared with resting cells, the variant genotypes showed a significantly decreased ability to induce TNF-and IL-6 in response to Pam3-cys, LPS and CpGs compared with wild-type TLR cells, both in patients (Fig. 4) and donors ( Table 3), suggesting that the impact of studied polymorphisms on the immune system is independent of the pathological context.
These results would restrict findings in previous studies to wild-type TLR-genotyped patients and they would suggest either a lack of the antigenic ligand recognition by TLR or a TLR variant genotype-induced breakdown in the pro-inflammatory signalling pathway. Our results in vitro show no differences in TLR gene expression levels, suggesting a competent stimulation of these receptors despite their polymorphic regions ( Fig. 4A-C). On the contrary, downstream signalling molecules were reduced in cells from TLR variant vs wildtype genotyped patients ( Fig. 4D-I), supporting the second hypothesis. Either way, as a result, bacterial antigenic internalization and/or cellular signalling events would be compromised, leading to serum free antigen accumulation and low pro-inflammatory cytokine levels. In fact, hyporresponsiveness to LPS has been described in the presence of TLR-4 variant genotypes 15 and TLR-4 expression downregulation 19 . This resultant "tolerant state" might account for the increased risk of infections commented before.
The mechanisms by which the TLR polymorphisms may condition receptor are not fully elucidated. Several controversial examples were nicely reviewed in the recent past by Medveded A 20 . For example, the R753Q TLR-2 polymorphism does not significantly affect TLR2 expression, but induces the suppression of NF-kB activation and cytokine expression by affecting the electrostatic potential of the TIR domain 21 . Other TLR2 variants, such as P681H, do not change wild-type TLR-2 expression levels 22 , but it is associated with slower rates of internalization from the cell surface to endosomes 23 . The D299G TLR-4 variant is associated with decreased TLR-4 expression levels in response to LPS 15 . In contrast, others have reported no changes in expression levels of WT, D299G, or T399I TLR4 variants in HEK293T/CD14/MD2 transfectants, but a compromised ability to elicit NFkB activation in response to LPS in the TLR-4 variants 24,25 . In fact, despite similar TLR-4 expression levels, PMNs from individuals expressing the D299G TLR4 variant showed reduced phosphorylation of IkB-a and secretion of IL-12 p17 upon stimulation with LPS compared with cells expressing WT TLR4 26 . D299G polymorphism has also been described to inhibit LPS-induced association of TLR-4 with adapters MyD88 and TRIF, resulting in suppressed activation of the transcription factors NF-kB and IRF-3, p38 MAPK phosphorylation, and induction of MyD88and TRIF-dependent cytokines 27 .
From a clinical standpoint, TLR polymorphisms have widely been associated with human disease 28 . Particularly, bacteria-related complications and the associated inflammatory response are relevant to the TLR polymorphism evaluation, as several members of this family are specific receptors of their products. In fact, Total n = 20 TNF-α (pg/mL) IL-6 (pg/mL) IL-10 (pg/mL)   previously TLR-2 and TLR-4 polymorphism have been associated with spontaneous bacterial peritonitis or bacteremia in cirrhotic patients 13,14 and the immunological hyporesponsiveness to LPS in TLR-4 variants has been previously reported 15 . As an explanation for those associations, we now demonstrate that the circulating antigen burden is significantly increased in variant-TLR cirrhotic patients. As the translocation of bacterial products is frequent and recurrent in cirrhosis, it is likely that genetically deficient TLR signalling might be involved and, therefore, provide new targets for treating bacteria-derived complications in cirrhosis.
In summary, polymorphisms in TLR-2, TLR-4 and TLR-9 genes are associated with an increased bacterial antigen burden and a deficient pro-inflammatory cytokine response in patients with cirrhosis. Both aspects may delay bacterial antigen clearance in blood and contribute to clinically relevant bacteria-derived complications in patients with cirrhosis.

Patients and Methods
Patients and study design. We conducted a prospective observational study in patients with cirrhosis and non-infected ascitic fluid (AF) consecutively admitted or followed at our Hospital from January 2014 to January 2016.
Exclusion criteria were the presence of spontaneous bacterial peritonitis (SBP) or infections, multinodular hepatocellular carcinoma, portal thrombosis, alcoholic hepatitis, previous liver transplantation or previous transjugular intrahepatic portosystemic shunt. SBP was defined as the presence of > 250 polymorphonuclear cells/μ L in AF. The Ethics committee of Hospital General Universitario de Alicante approved the study protocol, and all patients gave informed consent to participate in the study. All methods described herein were performed in accordance with the relevant guidelines and regulations.
Blood and AF were obtained from all patients at admission and analyzed for routine biochemical and cytological studies. Blood and AF cultures were performed in all cases. Aliquots of blood and AF were inoculated under aseptic conditions in sterile, rubber-sealed Vacutainer SST II tubes (BD Diagnostics, Belgium) that were never exposed to free air.

Bacterial antigen measurement in patients' samples.
Samples and reagents were handled in an airflow chamber and processed with pyrogen-free material tested by manufacturers. To determine lipoteichoic acid (LTA), the specific ligand of TLR-2, a Human lipoteichoic acid elisa kit (Abbexa Ltd., Cambridge, UK ) was used according to manufacturer's instructions. A quantitative chromogenic limulus amebocyte lisate (LAL) test (BioWhittaker, Nottingham, UK) was followed to evaluate endotoxin levels in blood and AF samples as previously described 30 . To detect the presence of bacterial DNA fragments in blood and AF, a broad-range polymerase chain reaction (PCR) was performed according to the methodology described elsewhere 31 . PCR amplicons were loaded onto DNA Laboratory-on-chips (Agilent Technologies, Palo Alto, CA) and analyzed with an Agilent 2100 BioAnalyzer.

Inflammatory mediators quantification.
Enzyme-linked immunosorbent assays (ELISAs) for the quantitative measurement of TNF-α and IL-6 levels were carried out in serum samples of patients by handling Human Quantikine kits (R&D Systems, Minneapolis, MN), according to manufacturer's instructions. All samples were tested in triplicate and read in a microplate reader. Lower limits of detection of all cytokine assays were 5-8 pg/mL.   Standard curves were generated for each plate, and the average zero standard optical densities were subtracted from the rest of standards, controls, and samples to obtain a corrected concentration for both cytokines. The sum of the NO metabolites nitrite (NO 2 − ) and nitrate (NO 3 − ) is widely used as an index of NO generation 32 and expressed as NOx levels (nmol/ml). NOx levels were calculated by measuring conversion of NO 3 − to NO 2 − by the enzyme nitrate reductase using an ELISA assay (R&D Systems, Minneapolis, USA). All samples were tested in duplicate and values were corrected by running samples with culture media without cells to assess background NOx levels.

Statistical analysis.
Continuous variables are reported as mean ± standard deviation (or as median [25th-75th percentiles] in Figures) and categorical variables as frequency or percentages. Quantitative data were analysed using the Mann-Whitney U test for simple comparisons or the Kruskal-Wallis test followed by pairwise comparisons using the Mann-Whitney U test with the post-hoc Bonferroni correction for multiple comparisons. Differences in qualitative variables were analysed using the χ 2 test. Bivariate correlations between continuous variables were calculated using the Spearman test. All reported p values are 2-sided, and p values lower than 0.05 were considered to indicate significance. All calculations were performed using the IBM SPSS Statistics 19.