Critical Role of the Disintegrin Metalloprotease ADAM-like Decysin-1 [ADAMDEC1] for Intestinal Immunity and Inflammation

Background and Aims: ADAM [A Disintegrin And Metalloproteinase] is a family of peptidase proteins which have diverse roles in tissue homeostasis and immunity. Here, we study ADAM-like DECysin-1 [ADAMDEC1] a unique member of the ADAM family. ADAMDEC1 expression is restricted to the macrophage/dendritic cell populations of the gastrointestinal tract and secondary lymphoid tissue. The biological function of ADAMDEC1 is unknown but it has been hypothesised to play a role in immunity. The identification of reduced ADAMDEC1 expression in Crohn’s disease patients has provided evidence of a potential role in bowel inflammation. Methods: Adamdec1-/- mice were exposed to dextran sodium sulphate or infected orally with Citrobacter rodentium or Salmonella typhimurium. The clinical response was monitored. Results: The loss of Adamdec1 rendered mice more susceptible to the induction of bacterial and chemical induced colitis, as evidenced by increased neutrophil infiltration, greater IL-6 and IL-1β secretion, more weight loss and increased mortality. In the absence of Adamdec1, greater numbers of Citrobacter rodentium were found in the spleen, suggestive of a breakdown in mucosal immunity which resulted in bacteraemia. Conclusion: In summary, ADAMDEC1 protects the bowel from chemical and bacterial insults, failure of which may predispose to Crohn’s disease.


Introduction
ADAM proteins are related to matrix metalloproteases and snake venom proteins. 1 In recent years, ADAMs have become increasingly recognised as important in inflammation and tissue repair, and a growing interest has developed in the potential role of metalloproteases in gut homeostasis and bowel inflammation. 2 ADAMDEC1, first identified in 1997, is the sole member of a subsidiary class of the ADAM family. 1 It has been linked to a number of inflammatory diseases including atherosclerosis, 3 pulmonary sarcoidosis, 4 osteoarthritis, 5,6 Crohn's disease, 7,8 and to gastrointestinal [GI] malignancies-gastric adenocarcinoma 9, 10 and colorectal cancer. 11,12 The physiological role of ADAMDEC1 remains to be determined.
ADAMDEC1 is a unique member of the ADAM family: it contains pro-and catalytic domains but has a truncated disintegrin domain and lacks the typical transmembrane domain and cytoplasmic tail. 13 It is believed to be a soluble, secreted protein and has recently been identified in osteoarthritis synovial fluid 5 and in the supernatant from un-stimulated ADAMDEC1-transfected HEK293 cells. 14 It is predicted that the disintegrin site is non-functional as the 'disintegrin loop', reported to be essential for integrin binding, 13 is missing. The other distinguishing feature of ADAMDEC1 is that it is the only mammalian ADAM protease in which a histidine [H] is replaced by an aspartic acid [D] residue within the zinc-binding sequence of the metalloprotease domain, HEXXHXXGXXD, [H362D]. The metalloprotease site of ADAMDEC1 has been shown to be proteolytically active in vitro but the natural ligands remain unclear. 14 Recently it has been demonstrated that the unique nature of the active site in ADAMDEC1 allows it to escape inhibition by the tissue inhibitors of metalloproteases 1-3 [TIMP1-3]. 15 ADAMDEC1 seems to have evolved to function independently from the normal intrinsic inhibitory mechanisms that regulate other metalloproteases.
The tissue distribution of ADAMDEC1 in the non-inflamed, steady state is almost exclusively in the GI tract and to a lesser extent in lymphoid tissue and spleen in humans. 16 At a cellular level, ADAMDEC1 expression has been identified in macrophages isolated from non-inflamed human intestine 16 and in mature CD40 activated CD11c + dendritic cells from the thymus and tonsils. 17,18 Although undetectable in monocytes, in vitro it is upregulated when these cells mature to macrophages and more rapidly if this occurs in the presence of lipopolysaccharide [LPS] or 1α, 25-dihydroxy vitamin D 3. 16 This responsiveness to LPS and the association with a number of inflammatory diseases have resulted in speculation that ADAMDEC1 has an active role in the immune system, but no conclusive evidence has been presented to date.
We have used three different bowel inflammation models, dextran sodium sulphate [DSS], Citrobacter rodentium [C. rodentium], and Salmonella typhimurium [S. typhimurium] in Adamdec1-deficient mice to determine whether the lack of this molecule alters the susceptibility or immune response to these insults. We have demonstrated that in the absence of Adamdec1, mice are more susceptible to infection and colitis. These findings suggest a physiological role for ADAMDEC1 in protecting the GI tract against infection and inflammation.

Adamdec1 +/+ and Adamdec1 -/mice
Animal studies were performed in accordance with the UK Animals [Scientific Procedures] Act 1986 and European Directive 2010/63/ EU on the protection of animals used for scientific purposes. Adamdec1 +/mice were generated by targeted mutagenesis of the Adamdec1 gene 1227 on chromosome 8 and insertion of a neomycin-resistant cassette into exon 11. The line was reconstituted from frozen embryos from the Deltagen repository. Embryonic stem cells were re-derived from 129/OlaHsd mice. The chimeric mice were back-crossed onto C57BL/6 mice [Charles Rivers] for a minimum of 6 generations. Adamdec1 +/+ , Adamdec1 -/-, and C57BL/6 wild-type mice were bred and maintained in specific pathogen-free [

Anti-mouse ADAMDEC1 antibody
The catalytic domain of murine Adamdec1 was cloned into a GB1-pBR22b expression construct and the protein was expressed in E coli. This GB1-ADAMDEC1 protein, with the solubility enhancer GB1 at the N-terminus and a His tag at the C-terminus, was confirmed by mass spectrometry [data not shown]. The protein was injected into rabbits to produce the anti-serum. C. rodentium protein lysates were blocked with bovine serum albumin [Sigma] followed by serum from Adamdec1 +/+ and Adamdec1 -/mice. Anti-mouse IgG-HRP was applied for 30 min to detect anti-C. rodentium IgG.

Histology and immunohistochemistry
Bound antibody was detected using ECL Plus [Amersham], exposed to Hyperfilm ECL [Amersham].

Intestinal permeability
Mice were administered 600 mg/kg of fluorescein isothiocyanate conjugated dextran [FITC-Dextran], MW 4 000 [Sigma Aldrich, FD4], dissolved in 200 µl PBS for gavage following a 3 h fast. After 4 h, serum samples were collected and serially diluted in PBS. The concentration of FITC in serum was determined by spectrophotofluorometry with an excitation of 485 nm and an emission wavelength of 520 nm using as standard serially diluted FITC-dextran. Serum from mice not administered FITC-dextran was used to determine the background.

Tissue expression of ADAMDEC1 is highly conserved across animal species
In the steady state in humans, expression of ADAMDEC1 is almost exclusively restricted to the macrophages of the GI tract. 16

Adamdec1, Role in Intestinal Immunity and Inflammation
The gene atlas database, BioGPS [http://biogps.org/#goto = welcome], revealed a conserved expression pattern for Adamdec1 across numerous mammalian species including pigs and gorillas [data not shown]. A low level of expression has also been reported in activated dendritic cells, from secondary lymphoid organs, in humans and mice. 13,16 Our results were in accordance with those previous descriptions. Using a panel of mouse tissues, Adamdec1 gene expression was shown to be predominantly in the small intestine, caecum and large intestine [ Figure 1a]. A similar expression pattern was also seen in humans [ Supplementary Figure 1a Western blots on colonic lysates from wild-type mice identified two distinct bands that were absent from Adamdec1 -/mice [ Figure 1c]. The higher molecular weight band [~ 42 kDa] migrated in an identical manner to a recombinant full-length mouse ADAMDEC1-His tagged protein. The lower molecular weight band was roughly the same size, ~ 35 kDa, as previously reported for active ADAMDEC1 protein. 13 In situ hybridisation [ Figure 1d] and immunohistochemistry [ Figure

ADAMDEC1 expression does not require the presence of gut microflora
Due to the restricted tissue expression of ADAMDEC1 to regions of the bowel with a high concentration of microorganisms, and the increased expression observed in monocyte-derived macrophages [MDM] after exposure to LPS, it has been proposed that the gut microflora may be responsible for stimulating the induction of this gene. 16 To test this theory, we measured the expression of Adamdec1 in germfree (GF) mice. mRNA expression levels were equivalent in adult germ-free and wild-type animals [ Figure 2a, c], indicating that it was not dependent on the presence of gut microflora.
Further evidence to support the presence of Adamdec1 in the germfree gut was found in two online datasets. An online embryonic mouse gene expression atlas resource [http://www.emouseatlas.org/emap/ home.html] demonstrated that Adamdec1 was expressed in the developing intestine at embryonic E14.5, before bacteria colonise the gut [ Figure 2d-f]. Extrapolation of Adamdec1 gene expression from transcriptomic profiles of mesenchymal and epithelial cells isolated from the small intestine of E18.5 mice confirmed the presence of Adamdec1 in the prenatal sterile gut. Significantly higher levels of Adamdec1 were seen in embryonic intestinal mesenchymal cells compared with epithelial cells, similar to humans and adult mice [ Figure 2g]. These findings provide supportive evidence that the presence of Adamdec1 in the gut is not solely dependent on a microbial stimulus.  19 Their online software program identified two single nucleotide polymorphisms, close to the Adamdec1 gene, affecting the coding sequences of Tnfsf10b and Pnma2 genes, which had a > 90% chance of being retained in the Adamdec1 -/mice. Neither of these genes are, however, expressed in the colon of wild-type mice either in the naïve state or during a dextran sodium sulphate [DSS]-induced colitis [data not shown]. It is therefore unlikely that these passenger mutations, if present, would have any significant bearing on our studies of gut inflammation in Adamdec1 -/and wild-type mice.

Adamdec1 is upregulated in the intestine during DSS-induced colitis and loss of expression results in an increased systemic response
Adamdec1 -/and wild-type mice were exposed to the colitogenic agent DSS [2% in the drinking water] for 7 days. In wild-type mice, Adamdec1 was significantly upregulated in the colon during exposure to DSS, and on withdrawal of DSS Adamdec1 levels decreased towards baseline, mirroring the inflammatory response [ Figure 3a]. Following DSS challenge, Adamdec1 -/mice demonstrated an earlier and more pronounced systemic response than wild-type counterparts, with significantly greater weight loss [ Figure 3b] and a higher mortality rate [ Figure 3c]. Mice lacking Adamdec1 expression lost on average 20% of their original body weight and 80% died by Day 9 as compared with the wild-type mice which lost on average only 10% of their body weight and 20% died. These results clearly show that Adamdec1 deficiency renders mice more susceptible to DSS-induced colitis. Mice heterozygous for Adamdec1 demonstrated a response between that of the homozygous knockout and that of the wild-type mice [ Supplementary  Figure 3a, b, available as Supplementary data at ECCO-JCC online]. Exposure to DSS for 7 days resulted in increased serum levels of IL-6, IL-1β, TNF, KC, IFNγ, and IL-10 in wild-type and Adamdec1 -/mice compared with naïve animals [ Figure 3d]. Serum levels of IL-6 and IL-1β were significantly elevated in Adamdec1 -/mice on Days 10 and 14 above those of the wild-type mice. In contrast, the levels of IL-10 were reduced in the knockout animals compared with the wild type. The circulatory levels of TNF, IFNγ, and KC were similar in both genotypes.
3.5. Loss of Adamdec1 expression results in an increased sensitivity to DSS-induced colitis, with an elevation in neutrophil recruitment to the colon and increased expression of colonic IL-22 and IL-17 The increased sensitivity of mice lacking Adamdec1 to DSS resulted in more pronounced phenotypic changes to the colonic tissue [ Figure 4 a-c]. There was a significant shortening of the colon in wild-type and Adamdec1 -/mice 9 days post initiation of DSS, but this was more marked in the knockout animals [ Figure 4a, b]. Histological examination of the colon revealed that the colitis was more severe in the knockout mice than the wild type on Days 5 and 9 [ Figure 4c]. Mice that survived the acute DSS-induced colitis demonstrated restoration of normal colonic architecture by Day 21, irrespective of genotype [ Figure 4c]. These results suggest that Adamdec1 plays an active part in the initial inflammatory response to DSS, but has a negligible role in the resolution phase after removal of the insult.  3.6. Adamdec1 -/mice are more susceptible to C. rodentium-induced colitis To investigate the potential role of Adamdec1 in bacterial-induced colitis we used the Gram-negative enteric bacterium, C. rodentium, which is capable of inducing a self-limiting colitis in mice. 20 Following inoculation with ~ 10 9 organisms, Adamdec1 -/mice demonstrated significantly greater weight loss [ Figure 5a] and mortality [ Figure 5b] than the wild type. Wild-type mice exposed to C. rodentium did not demonstrate a significant change in body weight over the duration of the assay, and no animals succumbed to the infection. By contrast, animals lacking Adamdec1 lost approximately 15% of their starting weight and 70% died over the 13-day period. Heterozygous [Adamdec1 +/-] mice responded in a similar way to the wild type [data not shown]. The administration of a 10-times lower inoculation dose of C. rodentium [10 8 ] induced a significant loss of weight in Adamdec1-deficient animals, whereas the wild-type mice continued to gain weight over the same period [ Supplementary Figure 3c]. There was no mortality in either the wild-type or knockout animals at this lower inoculation dose [ Supplementary Figure 3d]. The numbers of C. rodentium in the faeces were similar in both genotypes [ Figure 5c]. These results demonstrate an increased sensitivity to C. rodentium infection in animals lacking Adamdec1 expression, despite adequate intraluminal clearance.

Adamdec1 -/mice are more susceptible to colonic inflammation and systemic infection upon C. rodentium infection
Following C. rodentium inoculation, serum levels of IL-6, IL-1β, TNF, KC, IFNγ, and IL-10 were measured in Adamdec1 -/and wildtype mice. The serum levels of both IL-1β and IL-6 were significantly higher in Adamdec1-deficient as compared with wild-type mice by Day 13 [ Figure 5d]. IL-10, TNF, IL12p40, and KC serum levels were similar in both genotypes [ Figure 5d and data not shown]. C. rodentium infection induces a colitis of the caecum and colon in mice. In Adamdec1 -/mice, a significant increase in the total weight of the caecum was observed at 13 days post inoculation, suggesting an increased inflammatory infiltration. In comparison, wildtype mice did not demonstrate a significant change in caecal mass [ Figure 6a].
Increased sensitivity to C. rodentium in mice lacking Adamdec1 may result in an increase in systemic infection. In order to test this, spleens were isolated, weighed, and screened for the presence of live C. rodentium. The splenic weight did not alter during C. rodentium infection in wild-type animals, whereas a significant increase in splenic mass was observed in Adamdec1 -/mice by Day 13 [ Figure 5b]. The increase in mass coincided with higher numbers of live C. rodentium in the spleens of Adamdec1 -/-, compared with wildtype, mice [ Figure 5c, d]. These findings are consistent with impaired containment of the bacteria locally in the bowel and increased bacterial translocation into the circulation in mice lacking Adamdec1. It has been reported that clearance of C. rodentium from the gut is dependent on B cells and IgG secretion. 21 Adamdec1-deficient mice are however capable of mounting an antibody response to C. rodentium infection. Antibodies against C. rodentium protein lysate were detectable by Day 13 in the serum from Adamdec1 +/+ and Adamdec1 -/mice [Supplementary figure 5, available as Supplementary data at ECCO-JCC online]. These results demonstrate that Adamdec1 deficiency does not influence the development of a B cell-mediated adaptive immune response to C. rodentium infection.
3.8. Adamdec1 -/mice are more susceptible to S. typhimurium infection S. typhimurium infection in mice results in a transient systemic infection and fever, controlled initially through the activation of the innate immune system, followed by generation of an adaptive response. 22 The major site of infection is in the small intestine where access to the host occurs through M cells within the epithelium. Pretreatment of mice with antibiotic allows colonisation of the intestine and development of colitis. To assess the role of Adamdec1 in protecting the mouse from S. typhimurium infection, antibiotic pretreated knockout and wild-type mice were orally gavaged with 10 8 live organisms and monitored for 48 h [ Figure 7]. Adamdec1 -/mice lost a greater proportion of their body weight after S. typhimurium inoculation compared with the wild type [ Figure 7a] and, at 48 h post infection, 55% of the knockout compared with only 11% of the wild-type mice had succumbed to the infection [ Figure 7b]. These results suggest that Adamdec1 provides some protection to the host during the early phase of S. typhimurium infection.

Discussion
ADAMDEC1 is a unique member of the classical ADAM family, 13 a group of proteins recognised as critical players in immunity and tissue remodelling. 23 In the steady state, we have shown that ADAMDEC1 is abundantly and almost exclusively expressed in the GI tract, particularly in the small bowel. A highly conserved tissue expression pattern coupled with an increased susceptibility to GI infection upon loss of this protein demonstrates that ADAMDEC1 is an important component of the GI immune system.
In the intestinal wall we have demonstrated that ADAMDEC1 is expressed predominantly in mononuclear cells within the lamina propria in mice and humans. ADAMDEC1 has previously been isolated from human colonic macrophages; and extrapolation of ADAMDEC1 expression profiles, from online datasets of isolated murine colonic lamina propria cells, supports the finding of ADAMDEC1 in resident intestinal macrophage populations, in the steady state [GSE27859; GSE42101; GDS2982]. 16,24-26, Classically these CD45 + CD11b + F480 + CD11c +/-Ly6c hi CD103 -CX3CR1 hi intestinal lamina propria cells are reported as functionally phagocytic, bactericidal, non-migratory, tolerant, and derived from peripheral blood monocytes. [27][28][29] In addition, peripheral blood monocytes cultured under sterile conditions in vitro upregulate ADAMDEC1 expression as they differentiate into macrophages. 8,16 ADAMDEC1 is not expressed in resident macrophages from extra-intestinal tissues. A number of studies and online datasets demonstrate that ADAMDEC1 is not detectable in mature macrophage populations resident in the peritoneum, lung, liver, and adipose tissues and is only moderately expressed in mature splenic macrophages [GSE56711; GDS2982]. 24,30 However, during inflammation ADAMDEC1 has been reported to be upregulated at sites where it is not constitutively expressed, such as lung, joints, and major blood vessels in pulmonary sarcoidosis, 4 osteoarthritis, 5 and atherosclerotic plaque instability, 3 respectively. This association with human inflammatory diseases has led to speculation that ADAMDEC1 may play a role in the human immune system and acute inflammatory response. 13  At a cellular level, we and other investigators have demonstrated a rapid and robust induction of ADAMDEC1 in response to bacterial antigens, in particular LPS, 16 which would support a role for ADAMDEC1 as a bacterial response gene. It has therefore been proposed that ADAMDEC1 expression is a consequence of background priming by LPS in the gut lamina propria. 13 Our results do not support this supposition; we have presented evidence that ADAMDEC1 is also expressed in the GI tract of germ-free mice and in the un-colonised sterile prenatal embryological bowel. In addition, during MDM differentiation a significant elevation in ADAMDEC1 expression is observed in the absence of any bacterial ligands. These findings suggest that exposure to bacterial antigens is not a prerequisite for ADAMDEC1 expression in MDM or gut.
An alternative explanation for the tissue and cellular expression of ADAMDEC1 may be found in the origin of resident tissue macrophages. Historically, all tissue-resident macrophages were believed to be part of a linear mononuclear phagocytic system and originate from circulating blood monocytes, similar to inflammatory macrophages. 31 There is, however, a growing body of evidence that resident macrophages in the mouse are laid down early in development, arise from embryonic progenitors 32 in the yolk sac and or foetal liver, and undergo clonal expansion in situ with little or no contribution from circulating monocytes in adult life. The intestine is an exception to this rule and it has recently been shown that the resident macrophage population within the adult bowel is maintained through continual recruitment of circulating monocytes which originate from the bone marrow. 33 These findings, coupled with in vitro data demonstrating that ADAMDEC1 expression is induced during peripheral blood monocyte to macrophage differentiation, suggest that these intestinal macrophages may differ from extra-intestinal yolk sac-and foetal liver-derived macrophages. 16 It is probable therefore that the reason ADAMDEC1 is selectively expressed in the bowel is because it is the predominant tissue in which resident macrophages are derived from the peripheral blood monocyte pool in the naïve non-inflamed state. The reported expression of ADAMDEC1 in extra-intestinal diseased tissue could result from induction of ADAMDEC1 in MDM recruited to sites of inflammation.
In contrast to extra-intestinal tissues, ADAMDEC1 has been identified as significantly under-expressed in a number of diseases that affect the gut, such as GI cancers and Crohn's disease. [7][8][9]11,12 This phenomenon is unlikely to be secondary to bowel inflammation per se, as we have clearly shown that Adamdec1 is significantly upregulated, at a tissue level, in the bowel of mice following exposure to a colitogenic agent and induction of an acute colitis. Furthermore, ADAMDEC1 has been reported as under-expressed in MDM and terminal ileal [TI] tissue from patients with quiescent as well as active Crohn's disease. 7,8 We predicted that loss of ADAMDEC1 may increase the host's susceptibility to bacterial infection and intestinal inflammation resulting in the development of Crohn's disease. This theory was substantiated through the use of a number of GI infection and inflammatory models in Adamdec1-deficient mice. These studies provided evidence supporting a role for ADAMDEC1 in the immune response of the gut. In pathogen-free conditions, the absence of Adamdec1 expression has no noticeable effect on murine gut development, but subjecting Adamdec1-deficient mice to colitogenic agents revealed an elevated local and systemic inflammatory response, increased bacterial translocation and systemic infection, and greater mortality. It is plausible that ADAMDEC1 may also play an important role in human intestinal immunity and protection against the development of infection and chronic inflammation. These findings provide the first piece of evidence linking ADAMDEC1 to a regulatory role in bowel immunity.
The exact mechanism by which ADAMDEC1 exerts this protective effect against colitis is yet to be determined. ADAMDEC1 evolved from a superfamily of zinc-dependent proteolytic enzymes which are well recognised to activate and degrade a variety of substrates including chemokines, cytokines, growth factors, and extracellular matrix proteins. 2,23,34 By their substrate interactions it has become increasingly evident that these metalloproteases influence the function and migration of inflammatory cells, maintain tissue homeostasis, aid wound healing, and are pivotal in the regulation of the innate and adaptive immune response. The proteolytic targets of ADAMDEC1, however, remain elusive. A number of metalloproteases have been shown to cleave, activate, and in some cases degrade IL-1β. [35][36][37] Adamdec1 -/mice display an exaggerated level of serum IL-1β and the downstream cytokine IL-6 during both chemical and bacterial induced colitis, which would suggest that Adamdec1 may play a role in secretion and/or degradation of IL-1β. Both IL-1β and IL-6 are crucial for neutrophil recruitment to sites of infection and play a central role in preferentially inducing Th17 differentiation, independent of TGF-β. 38,39 IL-6 has also been reported to contribute to innate immune-mediated chronic intestinal inflammation by promoting production of IL-22 and IL-17 by innate lymphoid cells [ILCs]. 40 Consistent with this finding, expressions of IL-22 and IL-17 were significantly increased in Adamdec1 -/mice compared with wild-type animals, following exposure to DSS and induction of an experimental colitis, and no genotype difference was seen in TGF-β levels. Further study is required to ascertain the mechanism by which Adamdec1 may target IL-1β and IL-6 and delineate the T cell and ILC response in Adamdec1 -/mice.
A recent study has identified three other potential substrates [α2macroglobulin, carboxymethylated transferrin, and casein] as well as demonstrating the lack of responsiveness to the classical ADAMs inhibitors TIMP1-3. 14,15 It is unclear if the substrates identified so far are the true biological targets of ADAMDEC1, but it is plausible that ADAMDEC1 has evolved to escape inhibition by endogenous metalloprotease inhibitors. Further work is needed to identify the biologically relevant ligands for ADAMDEC1 and identify the mechanisms responsible for its expression and regulation within the gut and inflamed tissue.