Jump to
Abstract
Objective—The goal of this study was to investigate the effects of stimulants for a nucleotide-binding domain, leucine-rich repeat-containing (NLR) protein family on human artery endothelial cells and murine arteries.
Methods and Results—Human coronary artery endothelial cells were challenged in vitro with microbial components that stimulate NLRs or Toll-like receptors. We found stimulatory effects of NLR and Toll-like receptor ligands on the adhesion molecule expression and cytokine secretion by human coronary artery endothelial cells. On the basis of these results, we examined the in vivo effects of these ligands in mice. Among them, FK565, 1 of the nucleotide-binding oligomerization domain (Nod)-1 ligands induced strong site-specific inflammation in the aortic root. Furthermore, coronary arteritis/valvulitis developed after direct oral administration or ad libitum drinking of FK565. The degree of the respective vascular inflammation was associated with persistent high expression of proinflammatory chemokine/cytokine and matrix metallopeptidase (Mmp) genes in each tissue in vivo by microarray analysis.
Conclusion—This is the first coronary arteritis animal model induced by oral administration of a pure synthetic Nod1 ligand. The present study has demonstrated an unexpected role of Nod1 in the development of site-specific vascular inflammation, especially coronary arteritis. These findings might lead to the clarification of the pathogenesis and pathophysiology of coronary artery disease in humans.
Germ-line encoded pattern-recognition receptors of the innate immune system sense exogenous microbial components and endogenous danger signals to protect the host.1–4 The pattern-recognition receptors include Toll-like receptors (TLRs), retinoic acid-inducible gene (RIG)-I-like receptors, the leucine-rich repeat-containing (NLR) protein family, and as-yet-unidentified pattern-recognition receptors that recognize double-stranded DNA.1,3 The TLR, RIG-I-like receptor, and NLR families consist of 10 (human), 3, and more than 20 members, respectively.1,3,4
In the cardiovascular system, endothelial cells are usually the first among the structural cells to sense microbial components through pattern-recognition receptors. Human endothelial cells express functional innate immune receptors, such as TLRs and NLRs.5,6 There is a line of evidence that activation of TLRs, especially TLR4 and TLR2, contributes to the development and progression of cardiovascular diseases, including atherosclerosis, cardiac dysfunction in sepsis, and congestive heart failure.7 With respect to NLRs, only a limited number of studies have shown that human endothelial cells express functional NLRs, nucleotide-binding oligomerization domain 1 (NOD1) and NOD2. Chlamydophila pneumoniae and Listeria monocytogenes elicited NOD1-dependent interleukin (IL)-8 production in endothelial cells.8,9 A selective NOD1 ligand, FK565, but not a selective NOD2 ligand, muramyl dipeptide (MDP), induced nitric oxide synthase-II protein/activity and vascular hyporeactivity ex vivo and shock in vivo.10
Because innate immunity has been suggested to be involved in the pathogenesis or pathophysiology of cardiovascular diseases in adults,7 as well as vasculitis in Kawasaki disease (KD) in children,11 we have investigated the effects of stimulants for innate immune receptors, especially TLRs and NODs, on human artery endothelial cells in vitro and murine arteries in vivo. We found the stimulatory effects of pure NOD1 and NOD2 ligands on coronary artery endothelial cells in vitro and the induction of coronary arteritis by oral or parenteral administration of a pure selective NOD1 ligand with or without a microbial component in mice in vivo. This evidence indicates a possible linkage between an innate immune receptor, NOD1, and cardiovascular disorders.
Methods
Histological Evaluation
All organs were isolated using a Leica M500 ophthalmology microscope. Cryostat sections were used for the correct detection of 3 aortic valve cusps in these studies.
Severity of coronary arteritis was assessed by the cross-section with 3 aortic valve cusps as described,12 defined as follows: − indicates no inflammatory infiltration in the whole layer (from intima to adventitia) of nearest coronary arteries from aorta or in the aorta; + indicates that less than one third of the circumference showed inflammatory infiltration in the whole layer; 2+, between one third and two thirds; 3+, more than two thirds.
For a detailed description of methods, please see the supplemental materials, available online at http://atvb.ahajournals.org.
Results
NOD Ligands Enhance Intercellular Adhesion Molecule-1 Expression and Cytokine Production by Human Coronary Artery Endothelial Cells In Vitro
To investigate the direct effects of innate immune stimulants on the endothelial cells, human coronary artery endothelial cells (HCAEC) were challenged in vitro with microbial cell wall components that stimulate TLRs and NLRs. After preliminary time course studies (data not shown), we analyzed the effects of each reagent on intercellular adhesion molecule-1 (ICAM-1, CD54) expression and cytokine secretion by HCAEC on day 3. Significant ICAM-1 expression and IL-8 secretion were induced by a pure synthetic Nod1 ligand, γ-d-glutamyl-meso-diaminopimelic acid (iE-DAP); a pure synthetic Nod2 ligand, MDP; a pure synthetic TLR4 ligand, lipopolysaccharide (LPS) lipid A; and a TLR2 ligand, peptidoglycan from Escherichia coli K12 in HCAEC (Figure 1). Because synergistic effects of NLR and TLR ligands were observed in human cells,13 we analyzed the effects of various components in combination on HCAEC. Enhanced ICAM-1 expression and cytokine production were observed by combined stimulation with pure synthetic iE-DAP plus MDP or iE-DAP plus lipid A in HCAEC. No release of IL-1β, IL-10, tumor necrosis factor-α, or IL-12p70 was observed by any combination in HCAEC (data not shown). These results clearly demonstrate that pure synthetic Nod1 and Nod2 ligands and TLR ligands activate human artery endothelial cells in vitro.
Effects of innate immune stimulants on HCAEC. HCAEC (1×104 cells) were incubated with NOD1, NOD2, TLR, and other stimulants in various combinations. ICAM-1 expression (A) and IL-8 (B)/IL-6 (C) production in the culture supernatants were investigated in triplicate on day 3. The concentrations of stimulants are as follows: iE-DAP, MDP, and peptidoglycan from E. coli K12 (PGN K12), 1 (gray bars) or 10 (black bars) μg/mL; lipid A 10 (gray bars) or 100 (black bars) ng/mL. Data are presented as mean±SD. *P<0.01 compared with medium, †P<0.01 compared with either iE-DAP or MDP, ‡P<0.01 compared with either iE-DAP or lipid A (Dunnett test).
To rule out possible secondary effects of NOD stimulation on the day 3 experiment, we performed experiments on day 1 as well. Similar additive effects were observed between NOD1 and TLR4 ligands on day 1 (Supplemental Figure IA). In addition, NOD1 small interfering RNA completely inhibited additive effects of NOD1 and TLR4 ligands on day 1 experiments (Supplemental Figure IB). Thus, the additive effect between NOD1 and TLR4 ligands appeared to be not secondary but primary.
Nod1 Ligands Induce Site-Specific Inflammation In Vivo in Mice
Based on these in vitro results, we examined the in vivo effects of a pure synthetic Nod1 ligand (FK565), MDP, LPS, peptidoglycan from E. coli K12, and another bacterial component or bacteria (zymosan, OK432) on the artery endothelial cells in mice. As a Nod1 ligand, FK565 was mainly used for in vivo studies instead of iE-DAP or FK156, because FK565 is generally most effective among Nod1 ligands and showed stronger effects on HCAEC (Supplemental Figure II). ICAM-1 expression and cytokine production by HCAEC were enhanced by the combined addition of LPS (Figure 1), and priming with LPS upregulated the expression levels of TLR gene, resulting in the enhancement of innate immune response to peptidoglycan in mice.14 Therefore, BALB/c mice were intraperitoneally primed with or without LPS, and 24 hours later, each reagent was injected 4 times at an interval of 1 week (Supplemental Table IA). Subcutaneous injection of MDP, peptidoglycan from E. coli K12, zymosan, or OK432 with LPS priming or LPS priming alone did not induce any cellular infiltration in the arteries (Supplemental Table IA and Figure 2Aa). On the other hand, when mice were subcutaneously injected by FK565 (500 μg) with LPS priming, diffuse cellular infiltration was observed in the aortic root, including aortic valves and the origin of coronary arteries in all mice (Figure 2Ab). The histopathologic features of this coronary arteritis model were characterized by panarteritis with dense inflammatory infiltrate consisting mainly of neutrophils and macrophages (Supplemental Figure III) and not associated with fibrinoid necrosis, similar to those in the acute phase of KD,15 which is an acute febrile illness of childhood characterized by the occurrence of vasculitis, especially coronary arteritis and valvulitis. This model did not show coronary aneurysm, but the rupture of elastic fiber in coronary artery was observed, just as in KD. Formation of neither thrombus nor granuloma was recognized in aortic and coronary lesions of all experimental mice. MDP, by itself, did not induce coronary arteritis at 500 μg, but the enhancement of the effect of Nod1 agonist by MDP became apparent when MDP was added to suboptimal doses of FK565 (200 and 100 μg). Vascular inflammation was not observed in pulmonary, celiac, renal, or other arteries, whereas mild cellular infiltration was observed in other parts of aorta and in common carotid and subclavian arteries (data not shown). Subcutaneous injection of iE-DAP or FK156 induced a slight inflammatory reaction (data not shown), whereas that of FK565 induced highly reproducible and remarkable inflammation in the aortic root, indicating that the effects of Nod1 agonists vary greatly in vivo, depending on the chemical structure of Nod1 agonist used. FK565 administration induced arteritis of the similar severities and frequencies in other strains, such as C57B/6, DBA/2, CD1, CBA/J, and CH3 (data not shown). Severe combined immunodeficiency (SCID) mice developed weaker but significant arteritis, suggesting a partial involvement of acquired immunity in the inflammation induced by a pure Nod1 ligand, whereas no arteritis was observed in Nod1-knockout mice (Supplemental Table IB and Figure 2B).
Histopathologic changes after administration of Nod1 ligand in BALB/c, SCID, and Nod1 knockout mice. A, a, Cross sections with 3 aortic valve cusps of a BALB/c mouse treated with 100 μg of LPS IP 4 times (Supplemental Table IA). Severity of coronary arteritis: grade 0. Left, hematoxylin and eosin stain; right, elastica Van Gieson stain. b, Grade 3+ coronary arteritis/valvulitis of the mouse subcutaneously challenged by FK565 (500 μg) with LPS priming weekly 4 times. Upper panels, hematoxylin-eosin stain (left) and elastica Van Gieson stain (right); magnification, ×40. Lower panels, coronary artery (left) and aortic valve (right) in hematoxylin-eosin stain; magnification, ×200. c, Grade 3+ coronary arteritis and valvulitis of FK565–orally administrated mouse without LPS priming (left, 100 μg once a day, 6 days/week, 4 weeks; right: tap water ad libitum, 120 μg/day, 4 weeks, hematoxylin-eosin stain; magnification, ×40). Coronary arteritis and aortitis including valvulitis were histopathologically characterized by panarteritis with dense inflammatory infiltrate. Neither aneurysmal dilatation nor thrombus was associated. Control mice did not show any vasculitis by either tap water ad libitum or oral administration of water alone. Coronary artery and aortic valves are indicated by arrows and arrow heads, respectively. B, SCID mice (a) and Nod1−/− mice (b) were treated with FK565 with LPS priming, as shown in Supplemental Table IB. Grade 3+ coronary arteritis/valvulitis and no coronary arteritis/valvulitis are shown in SCID mice and Nod1−/− mice, respectively. Shown are hematoxylin-eosin stain (left) and elastica Van Gieson stain (right). Coronary artery is indicated by arrows. C, Oral administration of FK565 (100 μg for 6 consecutive days) after LPS priming induces no inflammation in the gut mucosa. a, Stomach (magnification: large panel, ×200; small panel, ×40). b, Small intestine (magnification: large panel, ×200; small panel, ×100). c, Colon (magnification: large panel, ×200; small panel, ×100).
As FK565 is highly stable and effective by parenteral and oral routes,16 FK565 was orally administered (Supplemental Table IC). All BALB/c mice showed coronary arteritis/valvulitis after oral administration of 6 days (100 μg/day) per week of FK565 with LPS priming, and the severity of coronary arteritis/valvulitis increased with the duration of the administration. Even in the absence of LPS priming, 4 of 5 mice developed severe coronary arteritis/valvulitis after 4 weeks (Figure 2Ac). When mice were given FK565-containing tap water (estimated daily doses of 120 μg/day [n=3] and 180 μg/day [n=3]) ad libitum for 4 weeks in the absence of LPS priming, all 6 developed severe (grade 3+) coronary arteritis/valvulitis (Figure 2Ac). On the other hand, oral administration of FK565 induced no inflammation in gut mucosa (Figure 2C) or in many arteries and organs (data not shown).
Site-Specific Vascular Inflammation In Vivo Is Associated With High Expression of Chemokine/Cytokine and Metallopeptidase Genes in Each Tissue
First, Nod1 expression levels in normal mice were quantified by real-time reverse transcription–polymerase chain reaction with glyceraldehyde-3-phospate dehydrogenase (GAPDH) and cadherin 517 (CDH5, a constitutive endothelial marker) genes as internal controls. Relative Nod1/GAPDH and Nod1/CDH5 ratios in normal aortic root were not higher than those in other arteries (Figure 3A). Consistent with the above results, Nod1-positive cells were detected with no great differences among normal vascular cells of coronary artery, aortic valve, and pulmonary artery by immunohistochemical staining (Figure 3B).
Nod1 expression in various tissues and organs of normal mice. A, Nod1 expression levels in various tissues and organs of normal mice were determined by quantitative real-time reverse transcription–polymerase chain reaction with GAPDH (top) and CDH5 (bottom) as internal controls. a. indicates artery; asc. ascending; abd., abdominal. Similar expression levels of Nod1 were observed in the vascular system when CDH5 was used as an internal control. B, Immunohistochemical stainings with Nod1-specific (left) and CD31-specific (right) antibodies in normal mice. a and b, coronary artery; c and d, aortic valve; e and f, pulmonary artery. Nod1-positive cells were immunohistochemically detected diffusely in normal endothelial cells and focally in smooth muscle cells of coronary and pulmonary arteries, and valvular fibroblasts. Scale bars=20 μm.
To explore the molecular mechanisms of site-specific inflammation induced by Nod1 ligand, microarray analysis was performed in the most inflamed tissue (aortic root), uninflamed tissue (pulmonary artery), and much less inflamed tissue (ascending to abdominal aorta) from the vascular system and in the immune system (spleen). As shown in Figure 4A, tissue-specific gene expression patterns were observed by the in vivo stimulation with LPS, FK565, or LPS plus FK565. Among 10 chemokine/cytokine genes highly expressed in FK565- and LPS plus FK565–stimulated aortic root, only 2, 4, and 0 were continuously elevated more than 5-fold in microarray data from FK565-stimulated pulmonary artery, aorta and spleen, respectively (Supplemental Table II). Several Mmp genes were also highly expressed in FK565-stimulated aortic root, indicating that the degree of inflammation was associated with persistent high expression of chemokine/cytokine and Mmp genes in each tissue in vivo. Marked inflammatory responses in aortic root by Nod1 ligand and TLR4 agonist were associated with the synergistic induction of chemokine (Ccl2, Cxcl13, Ccl8, Ccl7, Cxcl2), cytokine (Il6), Mmp (Mmp3, Mmp19, Mmp8), and Cam (Icam1, Selp, Jam2) mRNA levels.
The in vivo gene expression patterns of vascular and immune tissues of mice treated with LPS, FK565, or LPS plus FK565 (A) and the production of chemokine/cytokine by vascular tissues ex vivo (B) and endothelial cells in vitro (C) treated with innate immune ligands. A, The gene expression patterns of aortic root (AR), pulmonary artery (PA), ascending to abdominal aorta and spleen (SP) of mice stimulated by LPS priming (1), FK565 PO (2), or LPS priming+FK565 PO (3) on days 2, 4, and 7 are shown. a, 44 170 genes; b, 13 546 genes of which expression levels were more than 2-fold enhanced in AR after oral administration of FK565 with or without LPS priming compared with those without administration. Blue-to-red scale indicates expression levels from low (less than half) to high (more than 16-fold) compared with those with no stimulation (yellow indicates no change after stimulation). B, Supernatants collected from each tissue after 24 hours of culture with each reagent were assayed for CCL2 and IL-6 (n=6). The reagents were as follows: darker blue, none; green, 10 μg/mL iE-DAP; purple, 10 μg/mL FK565; medium blue, 10 μg/mL MDP; red, 100 ng/mL lipid A; yellow, none in NOD1−/− mice; lighter blue, 10 μg/mL FK565 in Nod1−/− mice. *P<0.01 vs none (Dunnett test); †P<0.01 vs abdominal (abd.) aorta; ‡P<0.01 vs aortic arch and abdominal aorta; §P<0.01 vs PA, aortic arch, and abdominal aorta (Tukey-Kramer honestly significant differences test). Both pulmonary artery and aortic root produced similar levels of CCL5 in response to FK565 stimulation (no stimulation, 0.01±0.01 ng/mg tissue protein in pulmonary artery and aortic root; FK565 stimulation, 0.57±0.40 ng/mg tissue protein in pulmonary artery or 0.57±0.25 ng/mg tissue protein in aortic root), proving that the arteries were properly prepared and viable. C, HCAEC and human pulmonary artery endothelial cells (HPA EC) (4×104 cells) were cultured in the presence or absence of a reagent for 24 hours, and supernatants were assayed in triplicate for each CCL2/IL-8 level. IL-8 is shown instead of IL-6 because there were no significant differences in the IL-6 production between HCAEC and HPA EC stimulated with Nod1 or Nod2 ligand. The concentrations of stimulants are as follows: darker blue, none; very light blue, 1 μg/mL iE-DAP; green, 10 μg/mL iE-DAP; lighter purple, 1 μg/mL FK565; darker purple, 10 μg/mL FK565; medium blue, 10 μg/mL MDP; red, 100 ng/mL lipid A. *P<0.01 vs none (Dunnett test), †P<0.01 vs HPA EC (Student t test).
Vascular Tissue/Cell-Specific Responses to Nod1 Ligand Ex Vivo or In Vitro
Comparison of the gene expression levels of 3 murine vascular tissues (aortic root, pulmonary artery, and arch portion of aorta) ex vivo, as well as of 2 human endothelial cells in vitro, in the presence or absence of FK565 showed vascular tissue/cell-specific responses to FK565 (Supplemental Figure IV, Supplemental Table III).
To further investigate the mechanisms of site-specific inflammation induced by Nod1 ligand, the production of chemokine (C-C motif) ligand 2 (CCL2) (monocyte chemoattractant protein-1) and IL-6, of which genes were highly expressed in in vivo FK565-stimulated aortic root (Supplemental Table II), was studied with ex vivo organ culture of aortic root, pulmonary artery, aortic arch, and abdominal aorta from the vascular system in the presence or absence of lipid A, Nod1 ligands (iE-DAP, FK565), or MDP. The production of CCL2 and IL-6 was significantly higher in aortic root than in pulmonary artery, aortic arch, and abdominal aorta by stimulation with FK565 in normal mice, whereas no production of CCL2/IL-6 in response to FK565 was observed in any vascular tissue from Nod1−/− mice (Figure 4B). In addition, the production of CCL2 and IL-8 by FK565 was also higher in HCAEC than in human pulmonary artery endothelial cells (Figure 4C), suggesting that site-specific vascular inflammation is ascribed to the intrinsic nature of vascular cells, including endothelial cells.18
Discussion
The present study has demonstrated that pure selective Nod1 ligands (iE-DAP, a dipeptide with a molecular mass of 319.3 Da; FK565, an acyltripeptide with a molecular mass of 502.6 Da; and FK156, a synthetic tetrapeptide originally isolated from culture filtrates of Streptomyces strains, with a molecular mass of 519.5 Da) and a Nod2 ligand (MDP) induced ICAM-1 expression and CCL2/IL-8 production in HCAEC, suggesting a possible role of these bacterial components in the pathogenesis of vasculitis.
In addition, coronary arteritis was induced in vivo in mice by selective Nod1 ligands, FK565, FK156, and iE-DAP. Nod2 ligand showed a significant effect on the development of coronary arteritis when the FK565 dose was suboptimal. The induction of coronary arteritis by Nod1 ligand was enhanced by various microbial components, such as TLR4 ligand (LPS). These findings can be explained by synergistic effects of Nod1 ligands with TLR agonists to produce inflammatory cytokines.19
The pathological findings of Nod1 ligand-induced coronary arteritis in mice were consistent with those of human coronary artery lesions in acute-phase KD, which showed edema and a dominant infiltration of neutrophils with some macrophages and lymphocytes at early stages (until 9 days after KD onset).20 No animal models of coronary arteritis have been reported with a pure or synthetic reagent.
On the basis of the fact that a Nod1 agonist, FK565, is very stable against temperature and acid,16 coronary arteritis was successfully induced by oral administration of FK565. This is the first coronary arteritis animal model induced by oral administration of a pure synthetic Nod1 ligand. Absorption site of FK565 is not clear, but it is possible that gut mucosa is a major site because there were no great differences in the efficiencies of the induction of coronary arteritis between direct oral administration of FK565 solution and ad libitum drinking of FK565-containing tap water.
Nod1 stimulants include meso-DAP and meso-l-anthionine, amino acids specific to bacterial peptidoglycans, iE-DAP, l-Ala-γ-d-Glu-meso-DAP (TriDAP), FK156, FK565, GlcNAc-(β1 to 4)-(anhydro)MurNAc-l-Ala-γ-d-Glu-meso-DAP, bacterial extracts (Bacillus species, Bacillus anthracis spores, Legionella pneumophila, Salmonella typhimurium, Mycobacterium tuberculosis), and live microbes (S. flexneri, Helicobacter pylori, enteroinvasive E. coli, Pseudomonas species, Chlamydia species, L. monocytogenes).4,21–24 Nod1 agonists are considered to be derived from peptidoglycans of most Gram-negative and some Gram-positive bacteria25 and Chlamydia,4,21,22 although major natural Nod1 stimulants produced by bacteria remain unknown.26 Biologically active peptidoglycan fragments are released during growth by Gram-negative bacteria (E. coli breaks down nearly 50% of its peptidoglycan every generation).27 As peptidoglycan is constantly turned more than27 and partly translocated across the gut mucosa into the circulation,23 NOD1 agonists in water-soluble or water-insoluble (lipophilic) forms,26 together with various microbial components, may be released from normal or pathological microbiota, which contains 1014 microbes, which are estimated to weigh 1 kg in an adult human, in the gastrointestinal tract, airways, genitourinary tract, ducts of exocrine glands, and skin.28–30
Site-specific vascular inflammation was not related to Nod1 expression levels but appeared to be due to a site-specific production of chemokine/cytokine by respective vascular structures, because ex vivo organ culture in the presence of FK565 showed a site-dominant production of CCL2 and IL-6, as shown in Figure 4. It is likely that higher expression levels of chemokine and Mmp genes in in vivo FK565-treated aortic root than in ex vivo FK565-treated one by microarray analysis suggested an amplification of inflammation by the migration of inflammatory cells in a site-specific manner.
The site-specific nature of arterial inflammation in response to Nod1 ligand might be explained by a difference in the expression levels of certain molecules involved in Nod1 signaling pathway, such as receptor interacting protein-2, mitogen-activated protein kinases, and nuclear factor-κB, and their inhibitors or activators.1,3,4 Among them, A20 (tumor necrosis factor-α-induced protein 3) is a candidate because it is a negative regulator of TLR and NLR signaling via nuclear factor-κB31 and is significantly related to intestinal innate immunity, including LPS tolerance.32,33 Considering that oral administration of Nod1 ligand does not induce inflammation in gut mucosa or in many arteries, such as pulmonary artery, it is possible that a certain Nod1-specific regulatory mechanism, such as A20, is responsible for the inhibition of Nod1 signaling.31,34 Further study is going on to identify which molecule in Nod1 signaling pathway is responsible for the site-specific effect of Nod1 ligand by the extensive comparison of inflammatory and noninflammatory tissues and the use of knockout or transgenic mice.
The present study has been the first to demonstrate an unexpected role of Nod1 in the development of site-specific vascular inflammation, especially coronary arteritis and valvulitis. These findings might lead to clarification of the pathogenesis and pathophysiology of coronary artery and valvular lesions in KD in children and coronary artery disease in adults.
Sources of Funding
This work was supported by grants from the Japan Society for the Promotion of Science and from the Health and Labour Sciences Research Grants from the Ministry of Health, Labour and Welfare, Japan.
Disclosures
None.
Acknowledgments
The authors are thankful to H. Fujii and C. Arimatsu for technical assistance, J. Kishimoto for statistical analyses, and T. Sasazuki for critical discussions of the manuscript.
- Received September 14, 2010.
- Accepted February 1, 2011.
- © 2011 American Heart Association, Inc.
References
This Issue
Jump to
Article Tools
Share this Article
- Nod1 Ligands Induce Site-Specific Vascular InflammationPermalink:
