Targeted Intestinal Tight Junction Hyperpermeability Alters the Microbiome, Behavior, and Visceromotor Responses.

Markedly increases in intestinal permeability occur in inflammatory bowel disease, graft-versus-host disease, celiac disease, and multiple organ dysfunction. In these diseases, effectors of increased permeability include immune signaling,1Su L. et al.Gastroenterology. 2009; 136: 551-563Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar microbiome,2Edelblum K.L. et al.Cell Mol Gastroenterol Hepatol. 2017; 4: 285-297Abstract Full Text Full Text PDF PubMed Google Scholar and corticosteroids3Meddings J.B. et al.Gastroenterology. 2000; 119: 1019-1028Abstract Full Text Full Text PDF PubMed Scopus (217) Google Scholar that, in part, signal through epithelial myosin light chain kinase (MLCK). More modest permeability increases occur in other disorders, including irritable bowel syndrome (IBS), autism spectrum disorder, depression, and stress-related disorders. However, data directly linking barrier loss to phenotypes of these diseases are lacking. To define the impact of modestly increased intestinal permeability, we studied transgenic mice with intestinal epithelial-specific constitutively-active myosin light chain kinase (CAMLCK) expression. This MLCK-dependent tight junction regulation increased intestinal permeability (Supplementary Figure S1A and B).1Su L. et al.Gastroenterology. 2009; 136: 551-563Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar Nevertheless, postnatal growth (Supplementary Figure S1C), reproduction, intestinal transit (Supplementary Figure S1D), intestinal histology, epithelial proliferation (a sensitive indicator of epithelial damage), and epithelial turnover are unaffected in CAMLCK transgenic (CAMLCKTg) mice.1Su L. et al.Gastroenterology. 2009; 136: 551-563Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar In contrast, mucosal tumor necrosis factor-α, interferon-γ, interleukin (IL)-10, and IL-13 transcripts as well as numbers of lamina propria neutrophils, CD4+ T cells, and IgA+ plasma cells are modestly increased by CAMLCK expression.1Su L. et al.Gastroenterology. 2009; 136: 551-563Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar,2Edelblum K.L. et al.Cell Mol Gastroenterol Hepatol. 2017; 4: 285-297Abstract Full Text Full Text PDF PubMed Google Scholar Subclinical inflammation is, therefore, present and, by microbiome-dependent, IL-17–mediated processes, affords partial protection from acute pathogen invasion.2Edelblum K.L. et al.Cell Mol Gastroenterol Hepatol. 2017; 4: 285-297Abstract Full Text Full Text PDF PubMed Google Scholar Immune activation is nevertheless unlikely to amplify CAMLCK-driven permeability increases, as barrier function and ZO-1 anchoring are both acutely normalized by enzymatic MLCK inhibition.1Su L. et al.Gastroenterology. 2009; 136: 551-563Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar,4Yu D. et al.Proc Natl Acad Sci U S A. 2010; 107: 8237-8241Crossref PubMed Scopus (135) Google Scholar We initially analyzed the gut microbiome of 31 wild-type (WT) and CAMLCKTg pups born to 8 WT dams. The microbiomes segregated by pup genotype but not dam (Supplementary Figure S1E) and included increased Clostridium and decreased Bacteroidetes, Enterococcus spp, and Prevotella in CAMLCKTg mice (Supplementary Figure S1F). Increased intestinal permeability can therefore cause dysbiosis-like microbiome shifts. Interestingly, maternal separation, which increases intestinal permeability, causes similar alterations and can be partially corrected by MLCK inhibitor–induced barrier restoration.5Rincel M. et al.Psychopharmacology (Berl). 2019; 236: 1583-1596Crossref PubMed Scopus (16) Google Scholar Microbiome alterations overlapping with the above have been reported in IBS and autism spectrum disorder. We therefore asked if CAMLCKTg mice displayed anxiety-like behavior, as occurs in those disorders, using the open field test (Figure 1A). Both the percentage of distance traveled in the center and the fraction of time spent in the center of the open field were reduced in CAMLCKTg mice (Figure 1A); this did not reflect reduced locomotor activity, as total distance traveled in the entire area was similar in CAMLCKTg and WT mice (Figure 1A). These data are consistent with increased anxiety-like behavior in CAMLCKTg mice. Although the results cannot differentiate between direct effects of increased permeability and those requiring intermediate mediators, these data demonstrate that intestinal permeability increases can trigger behavioral changes. Stress and increased permeability have been associated with enhanced visceral sensitivity in humans and rodents. Surprisingly, CAMLCKTg mice displayed striking visceral analgesia to colorectal distension relative to WT littermates (Figure 1B). Sensitivity was restored by enzymatic MLCK inhibition, water avoidance stress, or naloxone-mediated opioid receptor antagonism (Figure 1B). Although this effect of increased permeability on visceral sensitivity was unexpected, it is remarkably similar to the naloxone-reversible visceral analgesia reported in chronically stressed female rats6Larauche M. et al.Neurogastroenterol Motil. 2012; 24 (1031-e547)Google Scholar and naloxone-sensitive inhibition of nociceptive neurons by supernatants of colitic human and murine tissues.7Guerrero-Alba R. et al.Gut. 2017; 66: 2121-2131Crossref PubMed Scopus (27) Google Scholar Studies of female IBS patients have linked increased permeability to altered functional and structural brain connectivity.8Witt S.T. et al.Neuroimage Clin. 2019; 21: 101602Crossref PubMed Scopus (21) Google Scholar Thus, although responses to colorectal distension can be mediated by spinal reflexes as well as sensory, limbic, and paralimbic regions of the brain,9Larauche M. et al.Neurogastroenterol Motil. 2019; 31e13489Crossref PubMed Scopus (11) Google Scholar we asked if neuronal activation was modified by CAMLCK-induced permeability increases. C-Fos immunolabeling, an indicator of neuronal activity, was significantly greater in the paraventricular nucleus of the thalamus, the paraventricular nucleus of the hypothalamus, and the hippocampus but not the medial prefrontal cortex, nucleus accumbens, or amygdala of CAMLCKTg, relative to WT, mice (Figure 2, Supplementary Figure S2). Increased intestinal permeability may therefore increase basal neuronal activity in areas of the brain that regulate responses to visceral pain or stress9Larauche M. et al.Neurogastroenterol Motil. 2019; 31e13489Crossref PubMed Scopus (11) Google Scholar but not those associated with conscious visceral sensation. These results demonstrate that increased intestinal permeability can impact (1) gut microbiome composition, (2) behavior, (3) visceral pain responses, and (4) neuronal activation within the brain. Critically, these changes are all results, rather than causes, of intestinal barrier loss, as the latter was induced by targeted CAMLCK expression. The sites of neuronal activation in CAMLCKTg mice support the hypothesis that increased intestinal permeability can activate the hypothalamic-pituitary-adrenal axis.10Ait-Belgnaoui A. et al.Psychoneuroendocrinology. 2012; 37: 1885-1895Crossref PubMed Scopus (407) Google Scholar Conversely, hypothalamic-pituitary-adrenal axis activation by exogenous stress can induce intestinal permeability increases.3Meddings J.B. et al.Gastroenterology. 2000; 119: 1019-1028Abstract Full Text Full Text PDF PubMed Scopus (217) Google Scholar Thus, as has been proposed in inflammatory bowel disease and graft-versus-host disease, a self-amplifying cycle may ultimately direct the diverse phenotypes induced by MLCK-dependent, intestinal permeability increases. Further study is needed to define the complex relationships between intestinal permeability, stress, behavioral alterations, visceromotor responses, microbiome composition, and other abnormalities. These data are the first to assess behavior in a model in which a targeted increase in intestinal tight junction permeability is the only direct perturbation. The results demonstrate, unequivocally, that modest tight junction permeability increases induced via a physiologically and pathophysiologically relevant mechanism are sufficient to trigger local and systemic microbial, behavioral, and neurosensory changes. This provides a new perspective with which to understand previously hypothesized cause-effect relationships that have been proposed on the basis of correlative data. CAMLCKTg mice1Su L. et al.Gastroenterology. 2009; 136: 551-563Abstract Full Text Full Text PDF PubMed Scopus (255) Google Scholar, 2Edelblum K.L. et al.Cell Mol Gastroenterol Hepatol. 2017; 4: 285-297Abstract Full Text Full Text PDF PubMed Google Scholar, 3Meddings J.B. et al.Gastroenterology. 2000; 119: 1019-1028Abstract Full Text Full Text PDF PubMed Scopus (217) Google Scholar, 4Yu D. et al.Proc Natl Acad Sci U S A. 2010; 107: 8237-8241Crossref PubMed Scopus (135) Google Scholar (Tg(Vil-FLAG-CAMLCK)#Jrt) were maintained as male heterozygotes on C57BL/6J background. These were mated with wild-type (WT) C57BL/6J female mice to produce WT and CAMLCKTg littermates. At weaning, female mice were separated and housed at constant temperature (22 ± 1°C) with a 12-hour light/dark cycle. Food (Teklad 2018; Envigo, Indianapolis, IN) and water were available ad libitum. All experiments were performed at 8 weeks of age. Procedures were approved by the Ethical Committee CEEA-86, under the number APAFiS#4145. Gut microbiota were analyzed in two cohorts (15 WT and 16 CAMLCKTg) from 8 different WT dams. At sacrifice, colonic contents were stored at –80°C. DNA was extracted using the ZR fecal DNA MiniPrep kit (Zymo Research, Irvine, CA) and adjusted to 1 ng/μL. Changes in relative abundance of 24 microbial 16S rRNA gene targets were obtained by quantitative reverse-transcription polymerase chain reaction (PCR) using an adapted Gut Low-Density Array platform.5Rincel M. et al.Psychopharmacology (Berl). 2019; 236: 1583-1596Crossref PubMed Scopus (16) Google Scholar, 6Larauche M. et al.Neurogastroenterol Motil. 2012; 24 (1031-e547)Google Scholar, 7Guerrero-Alba R. et al.Gut. 2017; 66: 2121-2131Crossref PubMed Scopus (27) Google Scholar A universal bacterial primer set was included as the reference gene. quantitative reverse-transcription PCR was performed in duplicate on a ViiA7 (Applied Biosystems, Foster City, CA). Fluorescence data were imported into LinRegPCR to perform baseline corrections, calculate mean PCR efficiency per amplicon group. and calculate initial quantities. Among the 24 targeted amplicon groups, 6 were not detected in any fecal samples and were removed from the analysis (Bacteroides vulgatus, Alistipes spp, Parabacteroidetes distasonis, Roseburia spp, Escherichia coli, and Akkermansia muciniphila). Normalized N0-values were log10-transformed and processed by mixOmics (v6.1.1; https://www.bioconductor.org/packages/release/bioc/html/mixOmics.html) with RStudio (v1.0.44; RStudio, Boston, MA) to build a partial least-squares discriminant analysis. This multivariate supervised approach projects samples (X) onto a low-dimensional space of latent variables to maximize separation between groups according (Y = genotype). Leave-one-out cross-validation was used to select the optimal number of latent variables for partial least-squares discriminant analysis models. Mice explored a 50 × 50 cm arena (illumination 300 lx) for 10 minutes. Exploration was automatically assessed using a video tracking system (Bioseb, Vitrolles, France). The percentage of distance traveled and time spent and in the center area (20 × 20 cm) and total distance traveled in the entire arena were assessed. Two 0.08-mm diameter electrodes were implanted in the abdominal external oblique muscle and a third in the abdominal skin. On postoperative days 3–6, colorectal distension (CRD) was performed using a balloon catheter (Fogarty 4F catheter [Edwards Lifesciences, Irvine, CA], 1.1 cm length, tip 3.5 cm from the anus)8Witt S.T. et al.Neuroimage Clin. 2019; 21: 101602Crossref PubMed Scopus (21) Google Scholar in 10-second periods with increasing volumes from 0.02 mL to 0.10 mL, with 5 minutes’ rest between distensions. Abdominal electromyography activity was registered after the amplification (10,000×) and analyzed (Powerlab Chart 5; ADInstruments, Sydney, Australia). Basal electromyographic activity was subtracted from electromyographic activity registered during distension. Some mice were treated with ML-7 (2 mg/kg intraperitoneal) or naloxone sulfate (2 mg/kg intraperitoneal) 1 hour before CRD. For others, water avoidance stress was induced on a floating platform (3 × 3 cm) in the middle of a water-filled tank (40 × 40 cm) for 1 hour daily over four days. Recovery (30 minutes) preceded CRD. Animals received 70 μL of 100-mg/mL TRITC-70kDa dextran in tap water by gavage and were sacrificed 1 hour later.9Larauche M. et al.Neurogastroenterol Motil. 2019; 31e13489Crossref PubMed Scopus (11) Google Scholar The stomach and small and large intestine were cut in 11 equal parts. Luminal contents of each segment were centrifuged and fluorescence determined. Transit was calculated as the geometric center of the values for each mouse. Jejunal sections were mounted in Ussing chambers (Physiologic Instruments, San Diego, CA) filled with Krebs buffer and continuously oxygenated (95% O2, 5% CO2). After 1 hour of equilibration, fluorescein (1 mg/mL) was added in the apical chamber and fluorescence intensity of the basolateral chamber was measured after 1 hour. Mice were fasted for 4 hours before gavage with 150 μL of 100-mg/mL FITC-4kDa dextran in tap water. Blood (200 μL) was collected after 4 hours and plasma fluorescence determined. Vibratome sections (40 μm) were stained using polyclonal rabbit anti-c-Fos (Santa Cruz Biotechnology, Dallas, TX) and secondary horseradish peroxidase–conjugated goat anti-rabbit antisera (Jackson ImmunoResearch, West Grove, PA). NDPI images (×20) were obtained (Nanozoomer; Hamamatsu Photonics, Hamamatsu, Japan) and converted into TIFF format using ImageJ (NDPI tools plugin; National Institutes of Health, Bethesda, MD; https://imagej.nih.gov/ij, version 1.52a). Regions of interest were manually circumscribed using region-of-interest tools and c-Fos–immunoreactive cells quantified automatically using the particle analysis function (size: 5–20 μm2; circularity: 0.5–1). For each animal, 3–6 sections of each brain area were assessed by a blinded observer. Statistical significance was determined by 2-tailed t test, 2-tailed Mann-Whitney U test, or 2-way analysis of variance and set at P < .05. For microbial analyses, univariate analysis was realized in parallel to compare each amplicon separately using unpaired t test followed by the Benjamini-Hochberg adjustment of P values for multiple comparisons.Supplemental Figure 2CAMLCKTg (blue circles, n = 5–6) and wild-type (WT) (red squares, n = 5–6) littermates. Representative images of c-Fos–immunolabeled brains from CAMLCKTg and WT mice. Scale bars = 200 μm. Values are mean ± SD. ∗P < .05, t test.View Large Image Figure ViewerDownload Hi-res image Download (PPT)