Three different groups of mice were investigated: wild-type mice without colitis (WT group, n = 3), wild-type mice with acute chemically-induced colitis (DSS group, n = 3), IL-10 knockout mice without colitis (IL-10 C- group, n = 2) and IL-10 knockout mice with manifest colitis (IL-10 C+ group, n = 3). All mice studied were of C57BL/6J background and obtained from Jackson Laboratories (Bar Harbor, ME). Mice were housed in micro-isolator racks and allowed access to food and water ad libitum. Colitis was induced in 5 to 8-wk-old wild type mice (WT) by adding 3% dextran sulfate sodium (DSS, 40 kDa MW; obtained from ICN, Costa Mesa, CA) to the drinking water for 5 d. Acute colitis developed in all treated animals by d 4 of DSS exposure. Symptoms included diarrhea, gross or occult blood in stool (measured by Hemoccult II Sensa cards; Beckman Coulter, Palo Alto, CA), and weight loss. DSS-treated mice were sacrificed on d 7 after first exposure to DSS. The spontaneous development of colitis in mice genetically deficient in IL-10 was monitored by daily weight measurements and weekly stool occult blood tests. Animals were sacrificed by CO₂ asphyxiation. A portion of the midjejunum, the terminal ileum and the whole colon were obtained. The colon was divided into 5 parts representing the cecum, proximal, mid, distal colon, and the rectum. Each colon part was analyzed separately. Intestinal segments were fixed in Carnoy’s solution[5] for 3 h, then processed into paraffin blocks using standard techniques, cut into 4-10 µm sections and placed on SuperFrost slides (R. Langenbrinck, Emmendingen, Germany) for pathologic examination and FISH studies.

The studies were conducted under protocols approved by the Institutional Animal Care and Use Committee of Duke University Medical Center.

Oligonucleotide probes were synthesized with a Cy3 or Cy5 reactive fluorescent dye at the 5’ end (MWG Biotech, Ebersberg, Germany). Forty domain, group and species-specific FISH probes (Table 1) were applied to murine colon sections. The Eub338 universal bacterial probe was used to detect virtually all relevant bacteria. The nonsense probe Non338 was used to test for nonspecific binding of oligonucleotide probes. Formamide concentration and hybridization temperature were used as described to achieve the optimal stringency[6-30]. Additional hybridizations using a permeation step with lysozyme were performed in parallel for detection of Gram-positive bacteria.

In situ quantification of fecal bacteria Bacteria were visualized by FISH and 4, 6-diamidino-2-phenylindole stain (DAPI) under a Nikon e600 fluorescence microscope (Nikon, Tokyo, Japan) and photo-documented with a Nikon DXM1200 camera and software (Nikon). The enumeration of bacteria was performed only when hybridization signals were clear and morphologically distinguishable as bacterial cells by triple identification with universal and group-specific FISH probes and DAPI stain, in absence of cross-hybridization or hybridization with the Non338 nonsense probe.

For each group-specific FISH probe, high power (×1000 magnification) photographs were made of three different microscopic fields within the intestinal lumen at the widest, narrowest, and median diameters of each colonic segment. The estimation of bacterial concentrations was made based on the assumption that a 10-µL sample with a cell concentration of 10⁷ cells per mL brings 40 cells per average microscopic field at magnification of 1000[4]. Two investigators independently counted bacteria within a 50 µm² area of the microscopic field (about 1 cm² of the 100% scale photographic image). The mean number of bacteria was recorded for each colonic segment. In cases where single bacterial cells were morphologically indistinguishable due to high bacterial concentrations and the fluorescence was confluent to a homogeneously fluorescent mass, no enumeration was attempted and the bacterial concentration was assigned a value of >10¹² per mL. If single bacterial cells were not distinguishable within the bacterial carpet but some empty gaps could be recognized, the bacterial concentration was assigned a value of >10¹¹ per mL.

Evaluation of spatial interrelationship of bacteria and assessment of cross-hybridization The spatial organization of bacteria was evaluated in three steps. Group or species-specific signals (green-orange fluorescence Cy3) were visualized simultaneously with a Eub338 probe universal for all relevant bacteria (dark red fluorescence Cy5). All probes positively hybridizing with more than 1% of bacteria were further combined with each other in pairs, in a two-color analysis (Cy3 vs Cy5). This allowed us to characterize the position of different bacterial groups relative to each other within the intestine and to assess potential cross-hybridization. Although probes chosen for this study were designed for definite bacterial groups and extensively tested (Table 1)[6-30], the specificity of each probe is relative. In these experiments, multiple observations were made to determine whether bacteria that hybridized with one probe gave a positive hybridization signal with a probe representing other bacterial groups, rather than to rely on results obtained with a single probe. When probes of unrelated bacterial groups hybridized with the same bacteria, the hybridization stringency was adjusted until a clear differentiation was possible. Otherwise the results achieved with cross-hybridizing probes were excluded.

Quantification of dynamic compartments The quantification of the dynamic structures, such as bacterial populations of crypts, bacteria organized in layers adjacent to the mucosa, and the width and spread of the mucus, was based on the mean values of at least ten photographs of different microscopic fields made within each bowel segment at a magnification of ×400 for layers and ×1000 for crypts.

Mean values and standard deviations were calculated for bacteria concentrations and measurements of layer (e.g., mucus or interlaced) thickness. Using t-tests, P<0.05 was considered statistically significant.

Healthy wild type mice The ileum was narrow. The few microorganisms found were heterogeneously composed, random, and without signs of adhesion or contact with the intestinal wall (Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 I = ileum of a wild type normal mouse narrow and free of bacteria in most parts. E = epithelium; L = lumen.

The cecum was wide and contained a highly concentrated mass of bacteria. The exact enumeration of bacteria was impossible since single bacteria could not be distinguished within the confluent fluorescent carpet (Figure 2A). Luminal bacteria were in direct contact with the wall of the cecum. Despite this contact, the bacteria lining the cecal mucosa were probably non-adherent. Shrinkage caused by fixation sometimes led to dissection of feces from the cecal wall. In all regions where this dissection took place, no bacteria adhered to the mucosal surface (Figure 2B). Abundant bacteria were observed within crypts (Figure 2B, arrowheads).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Concentrated bacterial mass in direct contact with the mucosal surface (A) and non-adherent bacteria (B) in cecum of a healthy wild type mouse. E = epithelium; F = feces. Arrow indicates the shrinkage of feces, arrowheads indicate bacteria in crypts.

The proximal colon was wide initially, and narrows distally. Fecal bacteria were present in high numbers and contacted the colonic wall along most of its length. Numerous bacteria were seen in nearly half of all intestinal crypts, even very deep within the wall. The number of crypts containing bacteria and the number of bacteria within these crypts were even higher in the proximal colon than in the cecum.

A thin but dense band of homogeneous bacteria was present in the distal part of the colon. Because this band of bacteria was interlaced between the epithelial wall and the unorganized fecal masses, it was termed the interlaced layer (Figure 3).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Interlaced layer in the distal portion of the proximal colon.

The mid-colon was narrow, and generally contained little feces. If present, fecal material was restricted to a fine tube-like structure located centrally within the intestinal lumen and separated from the epithelial surface by mucus. Bacteria were seen less often within crypts (Figure 4).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Middle colon in a normal mouse.

The colon lumen widened in the distal colon and was filled with masses of feces containing a high concentration of bacteria. A growing mucus gap devoid of bacteria completely separated the colonic wall from this fecal biomass. Nearly no bacteria was found in crypts and no interlaced layer was seen.

The lumen of rectum was wide. The mucus layer separating the bacteria from the mucosa continuously thickened (Figure 5). No bacteria were found within crypts in the rectum.

Open in New Tab   Full Size Figure   Download Figure

Figure 5 Rectal mucosa covered with thick mucus (M). E = epithelium, F = feces.

IL-10 knockout mice without colitis The bacterial concentrations in the ileum of IL-10 knockout mice without colitis were similar to findings in healthy WT mice. However, the concentrations and occurrence of bacteria in crypts, the thickness of the interlaced layer, and the overall concentrations of fecal bacteria were noticeably less compared to healthy WT mice (Table 2).

DSS-treated wild-type mice and IL-10 knockout mice with colitis Five striking changes were observed both in mice with colitis due to DSS exposure and in IL-10 knockout mice with spontaneous severe colitis (Table 2). First, there was a massive reduction of bacterial concentrations within the fecal masses in all colonic segments compared to healthy WT mice. Second, in inflamed colon, the middle colon lost its dividing function. The narrowing observed in normal WT mice was not observed in the colitis groups. The area where fecal masses had direct contact with the mucosal surface was reduced. The mucus gap separating the colonic wall from the fecal biomass began more proximally than in healthy WT mice. In mice with colitis, the mucus layer was evident in the distal cecum and became broad and complete in the proximal colon. Third, the number and occurrence of bacteria within crypts increased manifold compared to mice without colitis. Bacteria could be regularly seen even in crypts of the inflamed distal colon, while bacteria were seen only in the crypts of cecum and proximal colon in healthy WT mice. Fourth, the interlaced layer became extremely thick in the proximal and middle colon of mice with colitis, and reached thicknesses of >300 µm in some locations. The interlaced layer could be observed separating feces from mucosa, beginning in the distal cecum and continuing to the proximal portion of the distal colon. Fifth, in addition to these gradual changes, bacteria adhering to the mucosa or invading mucosal epithelial cells were seen in mice with colitis, but not in healthy WT mice.

Intestinal bacteria were organized in crypt, interlaced, fecal, adhesive, and invasive compartments. Crypt, interlaced, and fecal compartments were observed in all groups; however, adhesive and invasive bacteria were seen only in animals with colitis. The bacterial groups found in different compartments are summarized in Table 3. Generally all bacterial groups that were observed within the crypt compartment could also be found in the interlaced and fecal compartments. However, not all bacterial groups found in feces had access to the interlaced compartment. In addition, some bacterial groups seen in both fecal and interlaced compartments were never seen in the crypt compartment (Table 3).

The composition of the bacterial populations within single compartments (feces, interlaced, crypt) was the same in all investigated groups of animals. However, the spread, extent, and volume of these different compartments within different colonic segments differed markedly between groups with and without colitis, irrespective of the etiology of the inflammation.

Fecal compartment The composition of the bacterial community in the lumen of ileum could not be reliably evaluated in all groups of mice due to low bacterial concentrations and their irregular distribution within the lumen. Different bacterial groups found in the ileum appeared to be at random and varied from animal to animal in an unpredictable manner.

All bacterial groups that were positively hybridized with FISH probes were homogeneously mixed within the feces, without a gradient in distribution between regions adjacent to the mucosa and distant regions (Figure 6). The Arch915, Gam42a, Beta42a, Srb385, Bfra602, CF319a probes demonstrated a high grade of cross-hybridization and were excluded from the evaluation. The Erec, Lach, Alf1b, Phasco, Lab158 and Bac303 probes hybridized with more than 10% of the fecal population at least in one of the animals tested. The Rbro, Chis150, Clit135, Ehal, Ecyl, LGC, Cor653, Bdis659 probes hybridized with more than 1% of the fecal population. The Bif164, HGC, Rfla, Enc131, Veil, Ato291, UroB, UroA, Strc493, Ebac, Arc1430 probes hybridized with less than 1% of the population. Despite this low overall proportion, the absolute number of these bacterial groups was higher than 10⁷ cells/mL and therefore, easy to distinguish from the non-specific background. The Ec1531, Y16s-69, Sgd, Fprau, Sfb, Hpy-1, Efaec, Ser1410, Dss658 probes failed to give signals that were different from the background fluorescence seen with the nonsense probe.

Open in New Tab   Full Size Figure   Download Figure

Figure 6 Fecal bacteria hybridized with Bacteroides (Cy3 green-orange, 6A) and Erec (Cy5 red, 6B) probes, cecum of healthy WT mice.

Changes in bacterial concentrations within the fecal compartment from the cecum to the rectum The types of bacteria present in the fecal compartment were constant in all colonic segments and animal groups; however, the concentrations differed markedly. This was illustrated by the changes in mean concentrations of bacteria that hybridized with Erec and Bac303 probes, the two most abundant bacterial groups within the fecal compartment, during the transition from cecum to rectum (Table 4).

Bacterial groups found both in the crypt and fecal compartments (e.g., Erec, Alpf1b, Phasco, etc.; Table 3) were predominant in the cecum of healthy WT mice and their concentrations declined distally. The concentrations of bacterial groups found mainly in feces (e.g., Bacteroides, Clit135, Chis150, etc. Table 3) remained relatively stable throughout the colon. Thus their concentrations increased relative to the Erec-like groups.

Crypt bacterial communities Crypt bacteria were mainly found in the cecum and proximal colon of healthy WT mice (Table 2). Starting with the middle colon, bacteria could be only sporadically observed within the crypts. No bacteria were found within crypts of the distal colon and rectum. The occurrence and number of crypt bacteria were significantly reduced in IL-10 knockout mice without colitis. In contrast, in both DSS mice with colitis and IL-10 knockout mice with spontaneous colitis, the crypt population was amplified and crypt bacteria could be seen sporadically even in the rectum (Table 2). Despite this significant difference in the extent and the distribution of crypt communities, the composition of the crypt population was the same in all investigated animal groups. Groups hybridizing with the Alf1b, Erec (Lach), Phasco and Lab158 probes were detected within crypts in different combinations. In the cecum, crypt bacteria directly contacted fecal masses and these bacteria were the main constituents of the fecal compartment. Bacteria positively hybridizing with Bac303 (Bacteroides), Chis150 (Clostridium histolyticum), and Clit135 (including Clostridium difficile) probes were never identified within crypts of any of the groups investigated (Table 3), although they were homogeneously intermixed within the fecal compartment and directly contacted the mucosa and crypt mouths in the cecum. In the proximal colon, the groups of bacteria that were present within crypts formed the interlaced layer before they mixed with the fecal compartment.

Interlaced layer The interlaced layer was mainly composed of the same bacterial groups that were present within the crypts (Table 3). These bacteria were condensed in extremely dense mats adjacent to the mucosa, which were clearly demarcated from the rest of the feces. This layer was relatively thin in WT mice and in IL-10 knockout mice without colitis. However, the interlaced layer was markedly amplified in all mice with colitis.

The concentration of bacteria into the interlaced layer was not simply numerical. The interlaced layer prevented the mucosa from contact with bacterial groups hybridizing with Bac303 (Bacteroides) and Clit135 (Clostridium difficile) probes which were observed in the cecum and initial parts of the proximal colon in healthy WT mice (Figure 6). The blocking role of the interlaced layer was especially well seen when pairs of probes representing all (Eub338)/interlaced (Erec, Alf1b, Phasco, Lab158, etc.) and exclusively fecal (e.g., Bac303, Clit135) bacterial population were simultaneously used (Figures 7, 8).

Open in New Tab   Full Size Figure   Download Figure

Figure 7 Interlaced layer in the proximal colon of IL-10 mice with colitis visualized simultaneously by hybridization with Bac303 (Cy3 green-orange, 7A) and Eub338 probes (Cy5 red, 7B).

Open in New Tab   Full Size Figure   Download Figure

Figure 8 Interlaced layer in proximal colon of DSS mice visualized simultaneously by hybridization with Clit135 (Cy5 red, 8A) and Lab158 probes (Cy3 green-orange, 8B). I = interlaced layer, E = epithelium, F = feces.

Adhesive bacteria Bacteria had direct contact with the cecal wall in healthy WT mice. These bacteria were separated completely from the wall when the fecal masses shrank during fixation without that bacteria adhered to the colonic surface (Figure 2B). Bacterial-mucosal contact in the cecum was therefore not adhesive.

True adhesion was observed in all animals with colitis. This was characterized by bacteria, which lined the mucosal surface and were located beneath the mucus layer. This true adhesion was found in at least one location in all animals with colitis, but was observed mainly in the distal portion of the colon (Figure 9). Sixty percent of adhesive bacteria were Bacteroides.

Open in New Tab   Full Size Figure   Download Figure

Figure 9 Bacteria location below the intact mucus layer (arrowheads) and adherence to the colonic mucosa in DSS-exposed mice.

Invading bacteria Bacteria invading the mucosa were found exclusively in the rectum and distal colon of DSS and IL-10 knockout mice with colitis (Figure 10). The types of invading bacteria were heterogeneous but mainly represented bacteria of the Bacteroides and Erec groups. The proportion of Bacteroides to Erec was more than three to one.

Open in New Tab   Full Size Figure   Download Figure

Figure 10 Bacteria infiltrating the mucosa in distal colon of IL-10 mice with colitis (arrowheads).