Article Text
Abstract
Background In Crohn's disease (CD) the deficiency of mannan-binding lectin (MBL) is associated with an increased prevalence of anti-Saccharomyces cerevisiae antibodies (ASCA) and with complicated phenotypes of the disease. However, the role of MBL in intestinal inflammation is currently unclear. A study was undertaken to analyse local MBL expression in human intestine and the consequences of MBL deficiency in experimental colitis and yeast infection.
Methods ASCA were measured by ELISA. MBL was assessed by ELISA and quantitative PCR. Wild type and MBL-deficient mice were administered dextran sulfate sodium (DSS) in the presence or absence of viable Candida albicans or adhesive invasive Escherichia coli (AIEC). Mice were infected with C albicans to assess generation of anti-yeast mannan antibodies.
Results MBL expression was virtually undetectable in the intestinal mucosa of both healthy controls and patients with CD, irrespective of macroscopic inflammation, indicating that systemic MBL must be responsible for the reduced risk of complicated disease in MBL-competent patients with CD. MBL-deficient mice showed enhanced DSS colitis upon oral challenge with C albicans or AIEC. C albicans could be recovered from the kidneys of colitic/C albicans-fed MBL-deficient, but not wild type mice. Infection with C albicans induced high titres of anti-C albicans mannan IgM and IgG in MBL-deficient mice but only a modest and transient IgM response with no class switch to IgG in wild type mice. Cross-reactive ASCA IgM continuously increased in MBL-deficient mice but rapidly declined after transient induction in wild type mice. In MBL-deficient mice, increased C albicans dissemination correlated with reduced early retention in the circulation.
Conclusions These results suggest that systemic MBL helps to prevent excessive inflammation upon access of normally mild pathogens across the damaged intestinal epithelium. Lack of this innate defence promotes antibody responses with cross-reactive potential against common mannan epitopes. These interpretations are compatible with the increased prevalence of ASCA and complicated disease phenotypes in MBL-deficient patients with CD.
- Crohn's disease
- experimental colitis
- Candida albicans
- immune response
- mucosal defence
- gene expression
Statistics from Altmetric.com
Significance of this study
What is already known about this subject?
Anti-Saccharomyces cerevisiae mannan antibodies (ASCA) are frequently found in patients with Crohn's disease (CD) but rarely in unrelated healthy individuals, and are mainly directed against a mannotetraose motif common in many yeasts and some bacteria.
In patients with CD, mannan-binding lectin (MBL) deficiency predisposes to ASCA positivity as well as yeast mannan-specific T cell reactivity and complicated disease phenotypes.
In patients with CD ASCA cross-react with cell wall mannan from other yeast strains and, to some extent, with mycobacterial lipoarabinomannan, and may represent an unusual immune response against normally tolerated ubiquitous antigens present in the intestinal flora.
The opportunistic pathogenic yeast Candida albicans can, to some extent, induce ASCA in rabbits but is a poor inducer of ASCA in mice.
What are the new findings?
Local MBL expression in the intestine is negligible, indicating that the reduced risk of complicated disease in MBL-competent compared with MBL-deficient patients with CD is mediated by circulating MBL.
Simultaneous oral challenge with the opportunistic commensal pathogens C albicans or adherent invasive E coli, both recognised by MBL, lead to accelerated/enhanced dextran sulfate sodium (DSS) colitis in MBL-deficient mice but not in wild type mice.
Viable C albicans can be recovered from the kidneys of colitic MBL-deficient but not wild type mice treated with DSS + C albicans.
After repeated infection with C albicans, MBL-deficient mice show continuously increasing titres of cross-reacting ASCA and, in contrast to wild type mice, a potent and persistent anti-C albicans mannan antibody response with efficient class switch to IgG.
How might it impact on clinical practice in the foreseeable future?
Other innate immune deficiencies as postulated for CD may have similar consequences regarding handling of pathogens and/or commensal microorganisms. Further investigations may eventually allow for individually tailored therapeutic programmes, specifically addressing individual, aberrant host–environment, host–pathogen or host–commensal interactions.
Background and aims
Crohn's disease (CD) and ulcerative colitis (UC) are of largely unknown aetiology and genetic, environmental and immunological factors play important roles in disease development. Once established, CD is driven by antigens of the intestinal flora, reflecting a loss of tolerance against commensal microorganisms.1–3 A number of serological markers have been identified in subgroups of patients with CD, several of which are specific for microbial antigens such as Escherichia coli outer membrane porin C (anti-OmpC2) or Pseudomonas fluorescence I2 protein.4 Most intriguing, >50% of patients with CD express antibodies against the yeast Saccharomyces cerevisiae cell wall mannan (ASCA).5 6 Little is known about the origin of ASCA, but they clearly reflect an abnormal immune sensitisation in patients with CD.7 Rather than originating from S cerevisiae, the triggering antigen for ASCA development may originate from opportunistic or pathogenic microorganisms containing similar cell wall mannan motifs such as C albicans or mycobacteria.8–10 The dominant ASCA epitope consists of terminal α-1,3-linked mannotriose or mannotetraose side chains.11 12 Such terminal α-1,3-linked mannoses are postulated to be the major antigenic determinant in the yeast cell wall.13
Yeast cell wall mannan is recognised by mannan-binding lectin (MBL). MBL represents an important first line of innate immune defence14–16 and, together with the MBL-associated serine proteases, initiates the lectin pathway of complement.17 Apart from this, MBL directly opsonises bound particles/microorganisms for phagocytosis.18–20 Deficiency of MBL is fairly common and has been linked to susceptibility to certain infections,21–23 as confirmed by animal studies using MBL-deficient mice.19 24 25 It can be hypothesised that mannan-containing antigens may be rapidly eliminated by this innate immune mechanism and that the absence of MBL would delay clearance of such antigens, allowing the host to mount an adaptive immune response.26 27 In agreement with this, we have previously shown that MBL-deficient patients with CD as well as their healthy relatives more frequently express ASCA.28 29
More recently we showed that MBL deficiency not only predisposes patients with CD to develop ASCA but it also increases the risk of complicated disease phenotypes.30 We therefore aimed to compare wild type and MBL-deficient mice with regard to the severity of intestinal inflammation on dextran sulfate sodium (DSS)-induced colitis with concomitant exposure to opportunistic pathogens that can be recognised by MBL such as C albicans. Since MBL also binds to lipopolysaccharide (LPS)31 and adhesive and invasive E coli (AIEC) strains have frequently been isolated from the bowel mucosa of patients with CD,32 33 we also determined whether colonisation with the prototype CD-derived AIEC LF8234 may have a differential effect during experimental colitis in MBL-deficient mice compared with wild type mice.
Standaert-Vitse et al have shown that infection of rabbits with C albicans can induce antibodies that cross-react with S cerevisiae mannan.8 In addition, we were able to generate ASCA IgM, but not IgG, in mice first immunised and then challenged orally with C albicans during chronic DSS colitis.9 These ASCA IgM were, however, of a transient nature only. Since MBL deficiency predisposes to the generation of (stable35) ASCA in CD, it can be asked whether C albicans-infected MBL-deficient mice might mount a more persistent antibody response that cross-reacts with S cerevisiae mannan, compared with wild type mice.
Materials and methods
Patients and controls
The study participants comprised 164 patients with CD of mean±SD age 40±17 years (range 18–82), mean±SD CDAI 141±51 (range 13–236) and mean±SD disease duration 11±13 years (range 1–39) grouped according to the Montreal classification into pure inflammatory (non-stricturing, non-penetrating, n=65); stricturing (n=57) and penetrating (n=42); 32 patients with ulcerative colitis (UC) of mean±SD age 40±12 years (range 21–70), mean±SD Mayo score 7±3 (range 3–10) and mean±SD disease duration 10±11 years (range 2–33); and 51 controls of mean±SD age 37±8 years (range 22–51).
Sample preparations
Mucosal biopsies were collected during endoscopic examination and stored in RNAlater (Qiagen, Hombrechtikon, Switzerland) until RNA isolation according to the manufacturer's instructions (Qiagen), including on-column DNA digestion. Snap frozen liver pieces were transferred to RLT lysis buffer (Qiagen) and further processed as the biopsies. RNA was reverse-transcribed using ImProm-II reverse transcriptase and random primers (Promega, Madison, Wisconsin, USA). For RNA isolation from cell suspensions, cell pellets were resuspended and lysed in RLT buffer and homogenised using QIAshredder columns (Qiagen). Homogenised cell lysates were mixed 1:1 with 70% ethanol and further processed as the biopsies.
Real-time PCR
cDNA samples were amplified using the TaqMan Universal PCR master mix and an intron spanning, minor groove binding TaqMan assay by design specific for human MBL2 (forward, 5′-CCAGGACCAAAGGGCCAAAA-3′; reverse, 5′-TGAGGCAGCCAGGCTACTAT-3′; reporter sequence, 5′-CACCATCCGGACTTTT-3′) (Applied Biosystems, Rotkreuz, Switzerland). Human TATA box binding protein (TBP) and cytokeratin 8 (KRT8)-specific TaqMan assays served as internal controls. Amplification was performed in the TaqMan 7900HT (Applied Biosystems).
ELISA
Serum concentrations of MBL were determined after 1/100 dilution using a commercial ELISA kit (Antibodyshop, Gentofte, Denmark) according to the manufacturer's instructions. ASCA and anti-C albicans antibody (ACAA) ELISAs were performed in Nunc-Immuno Maxisorp 96-well plates (Nunc, Wiesbaden, Germany) as described by Schaffer et al9 and in the online supplement.
Mice
C57BL/6J (B6) wild type mice were originally purchased from Harlan Europe (Harlan, Horst, The Netherlands). MBL-A and MBL-C double knock out (MBL-deficient or MBL−/−) mice on a B6 background19 36 were kindly provided by A Ezekowitz and J Jensenius. Mice were reared and bred in a specific pathogen-free (SPF) environment and experiments were approved by the local authorities of the Canton of Bern.
Immunisation/infection with viable C albicans
C albicans used in this study was kindly provided by K Mühlethaler, Institute for Infectious Diseases, University of Bern and was originally isolated from an ASCA-positive non-IBD patient with acute candidiasis. C albicans was grown in YPD (yeast peptone dextrose) broth at 37°C and harvested during exponential growth. Mice were intravenously infected with 104 C albicans blastoconidiae. Booster injections with 104 C albicans blastoconidiae were performed and blood samples drawn for serum preparations at the time points indicated in the corresponding figure.
Adherent invasive E coli strain LF82
LF82 was originally isolated from the ileal mucosa of a patient with CD34. It was grown in M17 broth (Merck, Zug, Switzerland) and harvested during exponential growth. The optical density (OD) was determined at 600 nm and the number of bacteria estimated with the following equation: bacteria/ml=OD600×3.6×108×dilution.
Detection of antibodies against MBL-bound LF82-derived antigens
ELISA plates were coated with 1 μg/ml recombinant human MBL (rhMBL, kindly provided by Enzon Pharmaceuticals, USA). After washing, wells were incubated with phosphate buffered saline supplemented with Ca2++ and Mg2++ (PBS-CM) containing 1% bovine serum albumin (BSA, background control), 1 μg/ml S cerevisiae mannan or 10 μg/ml whole LF82 lysate for 2 h at room temperature. The plates were washed and probed with serum samples from patients and controls diluted 1/500 in PBS-CM+0.1% BSA for 2 h at room temperature. Wells coated with sequentially diluted IgG of known concentration (Octagam, Octapharma, Vienna, Austria) were used to generate a standard curve.
DSS-induced colitis
Continuous low-dose DSS colitis was induced with 2.5% DSS (ICN Biomedicals Inc, Costa Mesa, California, USA) in the drinking water for 11 days. The 2.5% DSS solution was supplemented with 2×107 C albicans blastoconidiae or 108 E coli LF82/ml. Body weight was monitored every other day. At necroscopy, stool samples and colon and caecum tissue were collected. H&E-stained sections were scored independently by two experienced laboratory members (FS and BF) in a blinded fashion. The kidneys of mice fed with C albicans during DSS treatment were assessed for colony forming units (CFU) of C albicans, as described below. Faecal interleukin-1β (IL-1β) was determined as described by Konrad et al37 and in the online supplement.
Intravenous injections of C albicans and assessment of blood titres and renal invasion
106 C albicans in 100 μl 0.9% NaCl were either injected alone or together with 15 μg recombinant human (rh) MBL intravenously into the lateral tail vein. Blood samples were collected from tail tips into tubes containing 50 units sodium heparin (Liquemin, Drossa Pharm, Basel, Switzerland) and plated directly onto Sabouraud agar plates (Oxoid, Pratteln, Switzerland). The mice were killed and the kidneys homogenised in 0.9% NaCl. CFU of C albicans were assessed at appropriate dilutions on CandiSelect-4 plates (Bio-Rad, Reinach, Switzerland).
Statistics
Raw data were imported into a statistical package program (STATA Version 9.0, Texas, USA). The tests used are given in the corresponding figure legends. A p value <0.05 was considered statistically significant.
Results
MBL deficiency is predominantly found in ASCA-positive patients with CD
One hundred and sixty-four patients with CD, 32 with UC and 51 healthy controls were tested for MBL oligomer serum concentrations by ELISA. There was no difference between the mean±SD concentrations in patients with CD, those with UC and controls (2077±1547, 1809±1422 and 1958±1590 ng/ml, respectively). The proportions of MBL-deficient individuals were similar in patients with inflammatory bowel disease and controls (table 1).
ASCA were found significantly more frequently in the MBL-deficient subgroup of patients with CD than in the MBL-positive subgroup (χ2 test, p=0.029).
Low/undetectable MBL production in the gastrointestinal tract
Since extrahepatic MBL expression has been described in the small intestine,38 we compared intestinal production of MBL in patients with CD and controls. In the colon, MBL mRNA was never detected. None of the tested controls (n=12) or patients with UC (n=8) had cycle threshold values <40 for MBL mRNA in biopsies of the terminal ileum. In 7 of 28 macroscopically normal biopsies of CD ileum, traces of MBL mRNA were observed (cycle threshold values 36.9–40), although at 10 000–100 000-fold lower levels than in liver tissue (figure 1A and C). Only 2 of 17 crude ileal lavage samples contained minute amounts of measurable MBL oligomer (10 ng/ml and 2 ng/ml lavage). Enterocytes from ileal biopsies did not contain stored MBL protein, as determined by ELISA with detached and homogenised epithelial cells (n=3). RNA was also isolated from small bowel full-thickness specimens to include deeper layers of the mucosa. Only one out of four samples contained detectable MBL mRNA at about 10 000-fold lower levels than in liver tissue. MBL mRNA was found in three out of five duodenal samples at about 10 000-fold lower concentrations than in liver tissue (figure 1B). There was no difference in MBL mRNA expression between macroscopically normal and inflamed biopsies from the terminal ileum (figure 1C). Collectively, our analyses found extremely small amounts of MBL mRNA in only a fraction of the samples analysed. These results are summarised in table 2 where the minimal cycle threshold values obtained with each tissue are indicated.
MBL captures human ASCA-specific antigens of the AIEC strain LF82
We tested whether the AIEC strain LF82, originally isolated from the ileum of a patient with CD,34 39 contained antigens that bind to MBL and are recognised by serum antibodies of patients with CD. We found that ASCA-positive but not ASCA-negative serum samples from patients with CD contained up to 1.5 μg/ml IgG antibodies that bind to MBL-bound compounds from LF82. Notably, affinity-purified ASCA showed the same binding pattern as the unfractionated ASCA-positive sera, indicating cross-reactivity of ASCA between yeast mannan and AIEC LF82 antigens (figure 2). More detailed analyses revealed a partial overlap of ASCA with antibodies against MBL-bound purified LF82 LPS and type 1 pili. No LF82 LPS and type 1 pili antibody responses were detected in ASCA-negative control or patient serum samples (see figure 1 in online supplement).
MBL−/− mice develop more severe DSS-induced colitis when orally challenged with C albicans or LF82
MBL−/− mice had similar DSS-induced weight loss to age- and sex-matched wild type mice (figure 3A). Faecal IL-1β as well as disease scores of the colon and caecum were also comparable (figure 3B and C, respectively). However, MBL−/− but not wild type mice showed significantly accelerated weight loss, increased faecal IL-1β and higher histopathology scores when C albicans or LF82 were continuously administered during the DSS treatment (figure 3A–C). In MBL−/− mice, DSS+LF82 significantly increased caecal histopathology compared with DSS alone (figure 3C), accompanied by bleeding into the caecal lumen (not shown). Finally, viable C albicans could be recovered from the kidneys of DSS+C albicans-treated MBL−/−, but not wild type mice (figure 3D).
Sustained expression of experimentally-induced ASCA in MBL−/− mice
The increased proportion of ASCA-positive individuals in the MBL-deficient subgroup of patients with CD suggests that the absence of this innate defence mechanism promotes generation of an adaptive immune response against mannan, similar to that recently shown for systemic anti-commensal antibody responses in the absence of TLR4/MyD88 signalling in mice.40 To test this assumption we repeatedly infected wild type and MBL−/− mice intravenously with C albicans. We have previously shown that infection with S cerevisiae did not lead to notable ASCA production in wild type mice9 or in MBL−/− mice (not shown). However, upon infection with the opportunistic pathogenic yeast C albicans, both wild type and MBL−/− mice developed IgM antibodies against C albicans mannan (ACAA) which cross-react with S cerevisiae mannan. While in wild type mice this cross-reactive ASCA response was rapidly induced but very short-lived, in MBL−/− mice the ASCA titre slowly increased over the entire course of the experiment (figure 4). Intriguingly, the specific ACAA IgM response was markedly stronger and an efficient class switch to IgG was observed in MBL−/− mice but not in wild type mice.
Increased dissemination of C albicans into kidneys in MBL−/− mice
106 viable C albicans blastoconidiae were injected intravenously and C albicans CFU were determined in the blood after 15 min, 4 h and 32 h (figure 5A) and in the kidneys after 24 h, 32 h and 48 h (figure 5B). In wild type mice, blood counts of C albicans fell by a factor of 10 within 4 h after infection and thereafter continued to fall, although less rapidly, for the next 28 h. In contrast, MBL−/− mice showed an accelerated fall in blood counts of C albicans within 4 h after infection followed by a rapid increase thereafter. When C albicans was injected together with 15 μg rhMBL, the initial fall in viable C albicans blood counts was similar to that observed in wild type mice while the 32 h phenotype was only partially rescued (figure 5A). To assess C albicans dissemination after infection, we determined C albicans CFU from kidney homogenates. Infected MBL−/− mice but not wild type mice or MBL−/− mice co-injected with rhMBL showed a dramatic increase in C albicans CFU in the kidneys between 24 and 48 h after infection (p<0.05 at 32 h and 48 h, figure 5B).
Patients with CD with low serum MBL levels more frequently have a severe disease phenotype
We have previously shown in a Swiss nationwide cohort of patients with CD that low/deficient serum MBL levels are associated with complicated disease phenotypes.30 We therefore grouped our local cohort of patients with CD according to the same criteria and were able to confirm the results obtained with the larger nationwide cohort that low/deficient serum MBL levels were significantly associated with stenoses (p=0.006) or fistulae (p=0.04), which are considered complicated disease phenotypes. On the other hand, low/deficient serum MBL levels were negatively associated with the inflammatory UC-like disease phenotype (p<0.001) which is considered mild disease.
Discussion
We have shown in an experimental model that MBL deficiency results in mannan-containing pathogen-mediated acceleration of experimental colitis and predisposes to antibody formation and class switch against unusual mannan epitopes during C albicans infection, most likely owing to decreased innate resolution of the infection.
Our analyses of human intestinal tissue samples suggest that MBL is not locally expressed or produced at the mucosal barrier in large quantities. However, whole blood or plasma accumulating at sites of mucosal damage during excessive inflammation would efficiently deliver relevant concentrations of MBL to control invading microorganisms. In the event of mucosal damage, commensal microorganisms are able to access the systemic circulation and the presence or absence of MBL may influence innate handling of invading gut commensals. Indeed, we found that MBL-deficient mice showed accelerated DSS-induced colitis when the opportunistic pathogenic yeast C albicans or the AIEC LF82 were present in the bowel lumen, while the pre-existing SPF commensal flora did not have this effect. This enhanced colitis was not due to increased intestinal permeability since leakage of serum albumin into the bowel lumen under normal conditions and after DSS ± pathogen feeding was comparable in MBL−/− and wild type mice (data not shown). Binding of MBL to bacterial mannose-rich LPS, such as from certain salmonella strains, has been demonstrated.31 41 42 Our experiments showed that whole LF82 lysate contains antigens that readily bind to MBL and were at least linked to antigens recognised by ASCA in patients with CD. The partial overlap of antibodies against purified LF82 LPS and type 1 pili with ASCA indicates the highly polyclonal nature of these antibodies, and their absence in control and ASCA-negative sera tested so far is intriguing (see figure 1 in online supplement). Collectively, MBL and ASCA have overlapping specificity, not only with regard to yeast mannan but also to antigens of an AIEC originally isolated from the ileum of a patient with CD. A relationship between MBL and handling of LF82 is further accentuated by our observation that MBL−/− but not wild type mice showed excessive bleeding into their caecum and an increased histopathology score of the caecum when simultaneously fed with viable LF82 during DSS treatment. In contrast to our observations, a recent study by Jawhara et al found slightly enhanced DSS colitis upon oral C albicans administration in normal C57BL/6 mice.43 Several factors could be responsible for these contradictory results. First, the composition of the commensal flora most likely differs between the two animal facilities. Second, our C albicans strain is a unique isolate of a local patient with candidiasis. Third, the DSS regimen and C albicans administration protocols differ. In contrast to Jawhara et al, neither anti-C albicans nor anti-S cerevisiae mannan antibodies could be detected at the end of our DSS experiments. We assume that 11 days was too short and/or the severity of the colitis was too low. However, even after three cycles of DSS alternated with DSS-free tap water and concomitant oral application of C albicans, we did not find notable titres of anti-yeast mannan antibodies in wild type mice9 or in MBL-deficient mice (unpublished observations).
The association of MBL deficiency with an increased prevalence of ASCA in patients with CD may be explained by delayed innate clearance of yeast infections, allowing sufficient antigen persistence to mount an adaptive immune response. On the other hand, a large subgroup of ASCA-positive patients with CD expressed normal levels of MBL. The innate recognition of mannan has to be considered to be a redundant mechanism and disturbances of other components such as the TLR4 or the mannose receptor signalling pathway, as well as downstream events of the lectin pathway of complement, may also play a role. Anti-C albicans mannan antibodies not recognising the ASCA epitopes of S cerevisiae are frequently found in the healthy population.9 44 In our SPF wild type mice we failed to efficiently induce anti-C albicans mannan responses, indicating that long-term colonisation possibly with repeated local infections may be required. However, in MBL−/− mice, systemic C albicans infection alone induced high titres of anti-C albicans mannan IgM and efficient class switch to IgG. Intriguingly, while C albicans-infected wild type mice transiently developed notable amounts of IgM recognising S cerevisiae mannan (ASCA), MBL−/− mice only developed such cross-reacting antibodies very slowly which, however, continuously increased over time. These results suggest that MBL deficiency in fact predisposes to the development of a persistent adaptive immune response against mannan antigens, eventually leading to a stable ASCA-positive phenotype. Thus, together with the observed association of MBL deficiency with increased ASCA prevalence in CD, a considerable body of evidence now indicates that impaired innate handling of mannan-containing microorganisms in the absence of MBL supports the development of an abnormal adaptive immune response against these antigens. In contrast to C albicans, repeated intravenous infection with live LF82 only showed a transient and modest expression of anti-LF82 antibodies and no class switch to IgG in wild type and MBL−/− mice. A similar kinetic was observed for IgM cross-reacting with S cerevisiae mannan during the 10 weeks of repeated intravenous infections in both mice (see figure 2 in online supplement). Thus, although LF82 and yeast share antigens recognised by MBL, at least in mice the lack of MBL does not appear to predispose to persistent antibody responses against this E coli.
Complement-deficient mice are highly susceptible to lethal candidiasis.45 Less strikingly, this is also true when only the lectin pathway of complement is defective due to ablation of MBL.45 46 A common feature of complement- or MBL-deficient mice is an increased systemic dissemination of C albicans, most prominently observed in the kidneys. By binding to circulating C albicans, MBL supports opsonisation and subsequent uptake by phagocytes as has been shown by Brouwer et al,47 supporting a non-redundant role for MBL in yeast phagocytosis. However, using undiluted native sera, MBL did not further enhance opsonophagocytosis of S cerevisiae by neutrophils and monocyte-derived macrophages in our experiments (data not shown). Alternatively, rapid complexing of C albicans with circulating MBL might prevent efficient transmigration across the vascular endothelium, particularly in the kidneys. Indeed, we observed a more rapid decline of C albicans CFU in the blood of intravenous infected MBL−/− mice than in wild type mice, accompanied by increased dissemination into the kidneys, thus supporting a role for MBL in restricting C albicans to the blood circulation. The observed subsequent relapse with increasing C albicans CFU in the blood of MBL−/− but not wild type mice may then be the result of inefficient opsonophagocytosis and clearance. In contrast to C albicans, high-dose intravenous infections with LF82 (108 viable bacteria per mouse) did not lead to markedly different blood and kidney counts between wild type and MBL-deficient mice within 4 h after infection, and co-injection of rhMBL had no notable effect (data not shown). Therefore, at least in mice, MBL may represent a redundant innate mechanism to control systemic high-dose infections with E coli.
After oral feeding with C albicans during DSS treatment, viable C albicans could be recovered from the kidneys of MBL−/− mice but not wild type mice. This indirectly shows that systemic MBL indeed contributes to prevent orally applied C albicans from entering the blood circulation. Similarly, MBL deficiency in Helicobacter pylori-infected patients is significantly associated with the risk of severe gastric mucosal atrophy,48 which indicates that MBL is important in limiting pathogen-induced mucosal damage. After oral feeding, the AIEC LF82 is efficiently transported to the mesenteric lymph nodes (S Müller, unpublished observations). During DSS colitis an interaction with MBL may, however, rather take place at the level of the inflamed mucosa, as proposed for C albicans. As a more direct proof of the efficacy of MBL in our colitis model, we decided to inject rhMBL every two days into MBL−/− mice during DSS + C albicans administration. However, this treatment did not show any effect (data not shown). Either the half-life of rhMBL in these mice of about 12 h was not sufficient or the human MBL lacks critical functional characteristics such as efficient complexing with MBL-associated serine proteases in mice in vivo. In an attempt to shorten the time window in which rhMBL is supposed to interact with C albicans or LF82, on day 3 of an acute DSS treatment MBL−/− mice were gavaged with C albicans or LF82 and simultaneously administered rhMBL intravenously. Indeed, rhMBL administration slightly delayed weight loss between days 3 and 5. However, after the next simultaneous gavage/injection, this difference was no longer observed (see figure 3 in online supplement). We propose that, unlike the two murine MBL proteins (MBL A and MBL C), rhMBL may be less capable of preventing enhanced inflammation in response to C albicans or LF82 once a certain number of these pathogens has disseminated into the inflamed colonic tissues.
In conclusion, these results strongly suggest that MBL helps to control aberrant inflammation where microorganisms cross the damaged intestinal epithelium. Lack of this first line of innate defence promotes the generation of a systemic adaptive immune response in MBL-deficient mice. This is further supported by recent reports showing increased immune responses in MBL−/− mice26 27 and our observations that MBL-deficient patients with CD are more likely to have severe disease phenotypes and to express ASCA (present work and Schoepfer et al30).
Acknowledgments
The authors thank A Ezekowitz, J Jensenius and S Thiel for generously providing MBL−/− mice along with helpful information; D Grimmecke for the isolation of the different lipopolysaccharide species along with structural information/explanation; C Conover and M Hilgart for generously providing recombinant human MBL; H Madsen for helpful discussion and for replication of our mbl2 mRNA results; A Macpherson for helpful discussion; and E Slack for critically reading the manuscript and helpful comments.
References
Supplementary materials
Web Only Data gut.2010.208348
Files in this Data Supplement:
Footnotes
Funding This work was supported by the Swiss National Science Foundation grant number SNSF 3200B0-107527/1 to FS.
Competing interests None.
Ethics approval Informed consent was obtained from all patients and all procedures were approved by the ethical committee of the local authorities of the Canton of Bern.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Digest