Dynamics of the Mucosa-Associated Flora in Ulcerative Colitis Patients during Remission and Clinical Relapse

Abstract

The colonic mucosa-associated flora (MAF) in patients with active ulcerative colitis (UC) (n = 13) was investigated by examining 16S rRNA gene signatures during remission and relapse against levels for controls (n = 5). Baseline reduction, temporal instability, and decrease of bacterial richness toward relapse were observed for UC patients, whereas the MAF for controls was stable over time.

Microbial studies have shown a persistent reduction in biodiversity of the mucosa-associated flora (MAF) in patients with active inflammatory bowel disease compared with the MAF in normal and non-inflammatory bowel disease inflammatory controls (2). A loss of normal anaerobic components (Bacteroides, Eubacterium, and Lactobacillus spp.) has contributed to this difference (7, 11). A shift to an increased representation of gram-negative bacteria appears to accompany a reduced bacterial diversity in Crohn's disease patients during remission, suggesting that these abnormalities may be primary and not secondary to the inflammatory process (4). We have conducted a prospective study of the MAF in patients with ulcerative colitis (UC) during remission to investigate temporal changes in bacterial richness and diversity. Monthly biopsy specimens were obtained during a 1-year follow-up. Follow-up was terminated if patients developed a clinical relapse.

The 1-year prospective follow-up protocol was conducted at St. Mark's Hospital, Harrow, United Kingdom. The patients included had established UC confirmed by clinical, endoscopic, and histological features and had frequently relapsing disease (defined as at least two episodes requiring medical treatment in the past 12 months). Clinical characteristics are given in Table 1. Inclusion criteria allowed concurrent medication with oral 5-aminosalicylates (5-ASA) at a stable dose for at least 4 weeks and azathioprine or 6-mercaptopurine at a stable dose for at least 3 months prior to screening. Treatments with steroids, antibiotics, anti-tumor necrosis factor agents, and rectal 5-ASA were not permitted for a minimum of 4 weeks before or at any time during the study. The UC disease activity index (UCDAI) was used for assessment of disease activity (8). At enrollment, patients were in remission, defined as a UCDAI of 1 or 0 and an endoscopic subscore of 0 (normal mucosal appearance). Following screening, UC patients attended monthly study visits until relapse or for a maximum of 12 months. Relapse was defined as a UCDAI of greater than 3, including an endoscopic score of at least 1 (mucosal friability). Five healthy volunteers were included for benchmarking the assessments (Table 2). Specimens from these controls were collected at two time points (0 and 6 months) from the same location. No colonic cleansing with oral agents or enemas was used. All laboratory investigations were carried out at the Institute for Clinical Molecular Biology in Kiel, Germany. The research protocol was approved by the ethics committee of the Medical Faculty of Christian Albrechts University in Kiel. PCR, single-stranded conformation polymorphism analysis, cloning, and sequencing of bands, including sequence analysis, were performed as described earlier (5, 6, 7). For comparison of remission status and relapse status in UC, a total of 43 bands were included in the statistical analysis. Standard statistical tests were conducted using SPSS (10.0; Chicago, IL). To investigate the regularities of the bacterial dynamics in individuals with UC, information on the intensity levels of 43 bacterial bands for the 13 patients were used. These multidimensional data were treated using methods of the analysis of neural networks (9) (see the supplemental material). These measurements are based on community fingerprint profiles of bacterial consortia of monthly biopsy specimens taken after initial induction of remission until the moment of relapse. The time to relapse, in months, is given in Table 1.

TABLE 1.

Patient characteristics

Patient no.	Age (yr)	Gendera	Disease duration (yr)	Disease extent	No. of relapses per yr	Medication(s)	EIMc	Smoking statusd	Time to relapse (mo)
1	66	M	3	Total	2	5-ASA, Azab		Non	2
2	26	F	4	Left sided	2	5-ASA, Aza		Non	2
3	49	M	6	Total	3	5-ASA		Non	3
4	46	M	18	Left sided	2	5-ASA		Non	10
5	59	F	3	Left sided	2	5-ASA	Arthritis	Non	3
6	51	M	16	Total	4	5-ASA	Arthritis	Ex	3
7	35	M	7	Total	2	5-ASA		Non	3
8	27	M	7	Total	2	5-ASA, Aza		Non	3
9	62	F	30	Total	2	5-ASA		Ex	3
10	53	F	5	Total	2	5-ASA		Ex	3
11	30	M	22	Total	5	5-ASA, Aza		Non	3
12	45	M	5	Total	3	5-ASA, Aza		Non	4
13	45	F	10	Left sided	2	5ASA		Non	>3e

^aM, male; F, female.

^bAza, azathioprine.

^cEIM, extraintestinal manifestations.

^dNon, nonsmoker; Ex, former smoker.

^eLost to follow-up.

TABLE 2.

Controls

Control IDa no.	Age (yr) at time of first sampling	Genderb	Description	Endoscopic and histological diagnosis	Localization
NC3	24	M	Normal healthy volunteer	Normal	Sigmoid colon
NC8	20	F	Normal healthy volunteer	Normal	Sigmoid colon
NC13	24	F	Normal healthy volunteer	Normal	Sigmoid colon
NC24	32	F	Normal healthy volunteer	Normal	Sigmoid colon
NC23	40	M	Normal healthy volunteer	Normal	Sigmoid colon

^aID, identification.

^bM, male; F, female.

A total of 43 different band classes were identified for UC patients and 61 band classes for controls, corresponding to a baseline reduction in bacterial richness by 15% in UC patients at study start in comparison to levels for normal controls. One band of each class was excised, cloned, and sequenced. Sequencing of the 61 band classes resulted in taxonomic results from 53 bands, while 8 gave no reproducible results. The overall richness of bacterial operational taxonomic units (OTUs) as revealed by BLAST analysis is shown in Table 3. A total of 91 different OTUs were identified from the 53 band classes, including 17 (18.7%) sequences assigned to uncultured bacteria. As indicated in Fig. 1A, a significant reduction of bands was seen for UC patients between remission and relapse in comparison with repeated sampling of controls at the two different time points (25.31 versus 18.38 for UC patients, 38.4 versus 36.4 for the control sample, P = 0.031). A similar change in diversity scores was seen (Fig. 1B). No specific OTUs that would predict the time point of relapse were found during remission (i.e., at study start). Densitometric intensities of matched bands for the UC patients were pairwise compared between enrollment and time of relapse by use of a Wilcoxon matched-paired signed-rank test. For five bands, a significant difference was seen (P = 0.011), due to a reduction of the bacterial OTUs listed in Table 4. As shown, the decrease in bacterial complexity at relapse that was observed was due to a loss of normal intestinal bacterial taxa, such as Bacteroides, Escherichia, Eubacterium, Lactobacillus, and Ruminococcus spp. In comparison with the repeat biopsy specimens from normal controls, bacterial communities in specimens from patients with UC during remission showed variation in richness and diversity over time. Unfortunately, the relapse biopsy specimen endpoint was reached at month 2 for four patients and at month 3 for six patients (one patient lost to follow-up at month 3). One patient stayed in remission until month 4.

TABLE 3.

Taxonomic correlates of overall bacterial richness (alphabetical order)

Bacterial species (according to BLAST analysis)a
Bacteroides acidifaciens
Bacteroides caccae
Bacteroides distasonis
Bacteroides fragilis
Bacteroides merdae
Bacteroides sp. (different strains)
Bacteroides stercoris
Bacteroides thetaiotaomicron
Bacteroides uniformis
Bacteroides vulgatus
Bifidobacterium adolescentis
Bifidobacterium breve
Bifidobacterium lo ngum subsp. infantis
Bifidobacterium longum
Bilophila wadsworthia
Burkholderia sp.
Burkholderiales bacterium
Citrobacter freundii
Citrobacter sp.
Citrobacter youngae
Clostridium amygdalinum
Clostridium indolis
Clostridium leptum
Clostridium methoxybenzovorans
Clostridium ramosum
Clostridium sp. (different strains)
Desulfovibrio piger
Dialister invisus
Escherichia coli
Escherichia fergusonii
Eubacterium eligens
Eubacterium rectale
Faecalibacterium prausnitzii
Finegoldia magna
Gemella sanguinis
Hafnia alvei
Lactobacillus gasseri
Lactobacillus johnsonii
Lactobacillus sp.
Morganella morganii
Obesumbacterium proteus
Peptoniphilus ivorii
Peptoniphilus sp.
Peptostreptococcus anaerobius
Peptostreptococcus micros
Peptostreptococcus sp.
Photorhabdus luminescens
Prevotella bivia
Prevotella bryantii
Prevotella melaninogenica
Prevotella sp. oral clone
Prevotella zoogleoformans
Rahnella aquatilis
Ruminococcus gnavus
Ruminococcus lactaris
Ruminococcus luti
Ruminococcus obeum
Serratia fonticola
Serratia marcescens
Streptococcus anginosus
Streptococcus constellatus
Streptococcus mitis
Sutterella wadsworthensis
Veillonella atypica
Veillonella ratti
Veillonella sp.
Yersinia frederiksenii
Uncultured bacteria
Bacterium 84634
Bacterium mpn-isolate group 1
Bacterium mpn-isolate group 20
Butyrate-producing bacterium L2-21
Butyrate-producing bacterium M21/2
Butyrate-producing bacterium SM4/1
Butyrate-producing bacterium SR1/5
Butyrate-producing bacterium T2-132
Uncultured bacterium
Uncultured bacterium adhufec151
Uncultured bacterium adhufec405
Uncultured eubacterium
Uncultured eubacterium E1-K8
Uncultured gammaproteobacterium
Uncultured human fecal bacterium
Uncultured rumen bacterium 2C28d-8
Unidentified rumen bacterium RFN27

^aNomenclature according to reference 3.

FIG. 1.

Comparison of indicators (±standard deviations) for bacterial richness (number of bands in community fingerprinting gels [10]) (A) and bacterial diversity (weighted diversity scores calculated according to the method of Shannon and Weaver, as described earlier [7]) (B) at the points of first and last (i.e., the second for the controls) sampling, carried out using a nonparametric Wilcoxon test. As indicated, richness and diversity decrease in association with relapse in UC patients, whereas the same parameters were relatively stable for controls. The test indicators for richness in UC patients reached statistical significance compared to levels for controls (P = 0.031).

TABLE 4.

Intestinal bacteria downregulated at relapse in UC patients

Bacterial species (according to BLAST analysis)a
Bacteroides distasonis
Bacteroides merdae
Bacteroides sp.
Butyrate-producing bacterium
Escherichia coli
Escherichia fergusonii
Eubacterium rectale
Lactobacillus gasseri
Lactobacillus johnsonii
Lactobacillus sp.
Morganella morganii
Peptostreptococcus anaerobius
Ruminococcus obeum
Streptococcus gordonii
Streptococcus mitis
Uncultured bacterium
Uncultured eubacterium
Uncultured gammaproteobacterium

^aNomenclature according to reference 3.

The loss of commensal organisms could profoundly modify gut mucosal homeostasis through a loss of essential micronutrients and redox potential (e.g., short-chain fatty acids) (1). A further decrease of microbial diversity therefore not only could be a marker of relapse in UC but might be a causative factor driving the inflammatory change. Gut microbiota in normal individuals are remarkably stable over time (12, 13). Although the study was hampered by the early relapse that most patients developed, it appears that UC is associated with a high degree of instability in richness and diversity of the MAF. The inability to find specific changes in the microbial composition to predict relapse or to detect a temporal interaction between different species of the MAF may result from a lack of power due to the limited observation time with the small number of patients. Larger consecutive series, including long-term remission studies, are necessary to investigate this question. There is a significant decrease in diversity and richness of the MAF of patients with UC in remission compared to that of controls, and a further decrease in diversity was observed at relapse. In particular, the decrease in bacterial complexity at relapse that was observed was due to a loss of normal intestinal bacterial taxa, such as Bacteroides, Escherichia, Eubacterium, Lactobacillus, and Ruminococcus spp. The composition of MAF in UC patients appears to be remarkably unstable. Further studies will investigate treatment outcome and long-term courses of remission to examine a putative link between MAF and the pathophysiology of active disease for UC.