Duodenal biopsies were collected from children newly diagnosed with CD on a normal gluten-containing diet and from HC. The HC group consisted of children who were investigated for functional intestinal disorders of non-CD origin. All HC subjects had both negative celiac serology and showed normal small intestinal mucosa (Marsh 0).14 CD patients had both positive celiac serology markers (anti-tissue transglutaminase antibodies IgA, deamidated gliadin antibodies IgG, and antiendomysium antibodies IgA) and histological lesions in duodenal biopsy.

This study was conducted according to the guidelines established in the Helsinki Declaration according to EEC Good Clinical Practice Guidelines (document 111/3976/88, July 1990), and under the guidelines of current Spanish law which regulates clinical research in humans (Royal Decree 561/1993). The study protocol was approved by the Committee on Ethical Practice from CSIC and the Hospital.

Written informed consent was obtained from the parents of the children included in the study. No case included in the study received antibiotic treatment within 2 months before the completion of the endoscopy. Four duodenal mucosa biopsies were taken from each patient. Two samples from each case were sent for histological examination using the Marsh score and two other samples were immediately frozen at −20°C and kept until processing for bacteriological study. These were collected with 1ml of PBS and centrifuged at 1000×g for 5min.

Total DNA was extracted from each sample as described by Martínez et al.,15 with some modifications, as reported by Tapia-Paniagua et al.16 Twenty microlitres of DNA were treated with 2μl of sodium acetate 3M and 46μl of isopropanol. Samples were centrifuged 12,000×g for 6min. Pellet was cleaned with ethanol 70% and centrifuged 12,000×g for 6min. Pellet was dried and resuspended in water.

In order to compare DGGE patterns of the duodenal microbiota, DNA was amplified using the 16S rDNA bacterial domain-specific primers 968-GC-F and 1401-R. These primers were used to amplify V6-V8 regions of 16S rDNA. PCR mixtures and conditions to perform PCR were those previously described.12 Specific amplicons were also used for the genus Lactobacillus (Table 1). PCR mixtures (50μl) contained 1.25U Taq polymerase (Life Technologies Gaithersburg, MD, USA), 20mM Tris–HCl (pH 8.5), 50mM KCl, 3mM MgCl2, 200μM of each deoxynucleoside triphosphate, 5pmol of the primers, 1μl of DNA template, and UV-sterilized water. The PCR was performed in a T1 thermocycler (Whatman Biometra, Göttingen, Germany) using 1 cycle of 94°C for 2min, 35 cycles of 95°C for 30s, 56°C for 40s, and 72°C for 1min, followed by 1 cycle of 72°C for 5min. Aliquots (5μl) were analyzed by electrophoresis on 1.5% (w/v) agarose gels containing ethidium bromide to check for product size and quantity.

The amplicons obtained were separated by DGGE according to the specifications of Muyzer et al.17 using a Dcode TM system (Bio-Rad Laboratories, Hercules, CA). Electrophoresis was performed in 8% polyacrylamide gels (37.5:1 acrylamide–bisacrylamide; dimensions, 200mm×200mm×1mm) using a 30–55% denaturing gradient for separation of PCR products. The gels contained a 30–55% gradient of urea and formamide increasing in the direction of the electrophoresis. A 100% denaturing solution contained 7M urea and 40% (vol/vol) deionized formamide. PCR samples were applied to gels in aliquots of 13μl per lane. The gels were electrophoresed for 16h at 85V in 0.5× TAE (20mM Tris acetate [pH 7.4], 10mM sodium acetate, 0.5mM Na2-EDTA) buffer at a constant temperature of 60°C and subsequently stained with AgNO3.18

A DGGE analysis was performed and the banding patterns were analyzed using FPQuest Software version 4.0 (Applied Maths BVBA, Sint-Martens-Latem, Belgium). A matrix of similarities for the densitometric curves of the band patterns was calculated using the band-based Pearson coefficient, and clustering of DGGE patterns was achieved by construction of dendrograms using the Unweighted Pair Groups Method using Arithmetic Averages (UPGMA). The values of Pearson coefficients obtained for each treatment were compared using multiple range tests for similarity by diet. In order to determine the structural diversity of the microbial community corresponding to the DGGE banding patterns several parameters were calculated: (1) the specific richness (r) was calculated based on the total number of bands; (2) the Shannon index (H′) was calculated following the function: H′=−ΣPilogPi, where Pi is defined as (ni/N), ni is the peak surface of each band, and N is the sum of all the peak surfaces of all bands; (3) range-weighted richness (Rr) calculated as the total number of bands multiplied by the percentage of denaturing gradient needed to describe the total diversity of the sample analyzed, according to the following formula: Rr=(N2×Dg), where N represents the total number of bands in the pattern, and Dg the denaturing gradient comprised between the first and the last band of the pattern.19

In order to determine the principal bacteria, predominant bands in the DGGE gels were retrieved for sequencing with sterile pipette tips, placed in 100μl of double distilled water (ddH2O) and incubated at 4°C overnight. Volumes of 5μl were used as template in PCR amplification reaction performed as described above. The product was re-run on DGGE to confirm its position and further subjected to cycle sequencing with primers 968-without the GC clamp (5′-AACGCGAAGAACCTTAC-3′) and 1401-R and (5′-ACG′GCTACCTTGTTACGACTT-3′). The PCR was performed in a T1 thermocycler (Whatman Biometra, Göttingen, Germany) using one cycle of 94°C for 2min, 28 cycles of 95°C for 30s, 56°C for 40s and 72°C for 1min, followed by one cycle of 72°C for 5min. Products were purified using the High Pure Spin Kit PCR purification kit (Roche). The sequencing of the amplicons was performed on the ABI PRISM 377 sequencer (Perkin-Elmer). The sequence was read from both directions with primers RV-M and M13-47, respectively. The resulting sequences (∼500bp) were compared with the sequences from the National Center for Biotechnology Information (NCBI) or Greengenes DNA sequence database using the BLAST21 sequence algorithm. Database sequences showing the highest identity were used to infer identity. Although the high resolution of DGGE does not exclude the possibility that two different 16S rDNA sequences might migrate to exactly the same position, all sequences that migrated to the same position were sequenced. Chimeric sequences were identified by using the CHECK CHIMERA program of the Ribosomal Database Project.22

Principal component analysis (PCA) is a mathematical procedure that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables. It has been used to summarize RFLP profiles.23 In the present study it has been applied to correlate different such as ecological parameters (H′, R and Rr), histological parameters and the proportion of the DNA of each microbial species detected in the intestinal microbiota assayed with the intestinal microbiota of these patients. PCA has been performed using the LXSTAT 2012 software (Addinsoft, Spain) calculated with the help of the MS excel software XLSTAT 2012.1.01; Addinsoft The influence of the variables in the PCA was determined by the distance of each variable from the center of the axis d ≥√2/n, where n is the number of variables. For determining the quantity of DNA, the intensities obtained from each of the DGGE bands were used.

The data expressed as mean and standard deviation were analyzed with SPSS (Statistical Package for Social Science for Windows, version 15.0, SPSS Inc., Chicago, IL, USA). The ecological indices were analyzed with the Kruskal–Wallis test using the Statgraphics Plus 5.0 software (Statgraphics Corporation, Rockville, MD, USA). Differences were considered statistically significant when p≤0.05.

The values of richness, diversity, and range-weighted richness did not show significant differences among HC subjects and CD patients (Table 3), whereas the three values were significantly lower for CD compared to HC when the Lactobacillus genus was analyzed (Table 3).

Furthermore, a clustering analysis was applied to the duodenal microbiota DGGE patterns. This resulted in high intragroup similarity percentages (98%) in patients that presented histology associated with Marsh 3c type. The other patients showed a similarity from 80% until 30% and the group was not based on other characteristics (Fig. 1).

Figure 1.

Cluster analysis of DGGE patterns of bacterial group of the duodenal microbiota CD patients and HC.

(0.12MB).

A second dendrogram was done only with the amplicons of Lactobacillus genus showing two separate clusters with a 40% of similarity, approximately but the groups were not based on any analyzed characteristic. (Data not shown.)

Bands showing higher intensities in DGGE gels of samples of biopsies were sequenced and compared with BLAST references based on the phylogenetic relationship of the ∼500bp partial 16S rDNA sequence (Table 4). The bands were mainly related to species such as Streptococcus, Bacteroides and Escherichia coli in CD patients, while in HC the predominant bands were Bifidobacterium, Acinetobacter, and Lactobacillus. Other groups like Streptococcus (17–13%) or Bacteroides (16–11%) were minor.

The PCA shows that two factors (F1 and F2) are able to explain 83.44% of the variability found between the microbiota and the other variables studied in patients and controls. It is noteworthy that the variables related to the richness, diversity and habitability of the genus Lactobacillus change from reverse to the other variables used by the F1 addressed. Also the F1 separates both axes, HC from CD patients, so that F1 is a factor related to the presence or absence of disease (Fig. 2).

Figure 2.

PCA applied to microbiota composition of biopsies and parameters analyzed in CD patients and in healthy controls (HC). Correlation of assayed variables and principal components explaining 83.44% of variability. Distribution of subjects of CD and HC along axis F1 and F2.

(0.22MB).