Antonie van Leeuwenhoek

, Volume 94, Issue 3, pp 423–437

Investigation of chicken intestinal bacterial communities by 16S rRNA targeted fluorescence in situ hybridization

Authors

  • K. N. Olsen
    • Department of Veterinary Pathobiology, Faculty of Life SciencesUniversity of Copenhagen
    • DTU, National Food Institute
  • M. Henriksen
    • Department of Veterinary Pathobiology, Faculty of Life SciencesUniversity of Copenhagen
  • M. Bisgaard
    • Department of Veterinary Pathobiology, Faculty of Life SciencesUniversity of Copenhagen
  • O. L. Nielsen
    • Department of Veterinary Pathobiology, Faculty of Life SciencesUniversity of Copenhagen
    • Department of Veterinary Pathobiology, Faculty of Life SciencesUniversity of Copenhagen
Original Paper

DOI: 10.1007/s10482-008-9260-0

Cite this article as:
Olsen, K.N., Henriksen, M., Bisgaard, M. et al. Antonie van Leeuwenhoek (2008) 94: 423. doi:10.1007/s10482-008-9260-0

Abstract

The aim of the investigation was to quantify selected dominant bacterial groups in the chicken intestinal tract. Conventional production was used as model and the effect of the supplement with Salinomycin was evaluated. Hybridization conditions were optimized for published probes with respect to a panel of reference bacteria. In chicken intestinal samples bacteria were quantified by fluorescence in situ hybridization with 16S rRNA oligonucleotides directed towards bacteria related to Lactobacillus, Bacillus, Enterococcus-Streptococcus-Lactococcus, Enterobacteriaceae, Bacteroides, Clostridium and the domain Bacteria in lumen of ileum and cecum as well as on the intestinal wall including mucus of four individuals. Salinomycin in feed reduced counts of the Lactobacillus-, Enterobacteriaceae- and Clostridium-like bacteria in lumen of ileum compared to the conventional control. Increased or decreased bacterial counts were registered by Salinomycin in the ceca compared to the control. Relatively higher counts of Bacteroides- and Clostridium-like bacteria were found on the intestinal wall including mucus compared to lumen. The increase in numbers of some bacterial groups as well as the expected reduction by Salinomycin and the observed difference in the relative distribution of bacteria between lumen and intestinal wall are new observations that will need further investigation.

Keywords

SalinomycinIleumCecumFISH

Introduction

Bacteria in the gastro-intestinal tract (GIT) survive on the dietary compounds which are resistant to digestive fluids or which are absorbed so slowly by their host that they enables the bacteria to successfully compete for them (Lan et al. 2005). Intestinal bacteria probably also benefit from host material such as mucus and dead intestinal cells as well as from nutrient cycling between bacteria (Donoghue et al. 2006; Lan et al. 2005). The GIT microbiota seems highly dependent on the chicken diet, however, the immune system and external factors such as the administration of antibiotic growth promoters (AGP) and coccidiostats probably also determines growth and diversity of the populations (Lan et al. 2005).

As a result of the persistent occurrence of resistant bacteria in the environment and the fear of cross-resistance to human therapeutic drugs, the use of AGPs was completely banned in the European Union by January 2006 (Dibner and Richards 2005). Coccidiostats are widely used as feed additives, primarily to prevent coccidiosis, a parasitic disease of the intestinal tract. In addition to their anti-parasitic action, ionophores have a bacteriostatic effect on Gram-positive bacteria, such as Clostridium perfringens (Bjerrum et al. 2006). However, there have only been limited investigations of the effect of AGPs and ionophore type coccidiostats on the relative proportions of the bacterial groups. The use of coccidiostats in animal feed is scheduled to be banned from chicken feed in EU by around 2013. Since AGPs are not used and current coccidiostats are scheduled to be phased out, alternatives are being investigated including tools to document their effect.

Limited detailed information is available regarding bacterial numbers in the chicken intestine despite intense research in this field. Total numbers of bacteria determined three days post hatching reached 109 and 1011 g−1 of ileum and cecal content, respectively and remained relatively constant for the next 30 days (Apajalahti et al. 2004; Gong et al. 2002a, b). Distinct bacterial communities have been found in different sections of the digestive tract such as the crop, gizzard, duodenum, jejunum, ileum, ceca and colon (Gong et al. 2002b; van der Wielen et al. 2002). Members of Lactobacillus, Clostridiaceae and Streptococcus have often been found in ileum and ceca, and Enterococcus have been found only in ileum and members of Fusobacterium, Bacteroides, Ruminococcus, Bifidobacterium, Sporomusa, and Enterobacteriaceae have been reported to dominate only in the ceca (Apajalahti et al. 2001; Bjerrum et al. 2006; Gabriel et al. 2006; Gong et al. 2002a, b; Lan et al. 2002, 2005; Lu et al. 2003; Zhu et al. 2002).

It is recognized that culture-based investigations may be biased by the unculturability of some bacteria (Lan et al. 2005) and the use of culture-independent methods is becoming the choice for further detection of bacteria in chicken GIT (Apajalahti et al. 2004). For example, cloning and sequencing of 16S rRNA genes has turned out to be an excellent technique for the characterization of both culturable and non-culturable bacteria in the chicken intestine (Bjerrum et al. 2006; Gong et al. 2002a; Lan et al. 2002; Lu et al. 2003; Zhu et al. 2002). Unfortunately, only qualitative comparisons have been possible by use of this technique and results might have been biased towards certain groups of bacteria (Gill et al. 2006). A link between the qualitative information obtained from such cloned sequence libraries to quantitative measures can be obtained with fluorescence in situ hybridization (FISH) (DeLong et al. 1989; Langendijk et al. 1995). Furthermore, FISH allows comparison to information gained by various PCR-related 16S rRNA profiling-techniques (SSCP-PCR, T-RFLP, RT-PCR, DGGE) (Gong et al. 2002b; Lan et al. 2004; van der Wielen et al. 2002; Wise and Siragusa 2005; Zhu et al. 2002). While FISH has been used extensively for the analysis of human fecal flora (Barc et al. 2004; Franks et al. 1998; Langendijk et al. 1995; Rigottier-Gois et al. 2003; Schwiertz et al. 2000; Zoetendal et al. 2002), only few studies have dealt with the quantification of bacteria in chicken GIT (Moreno et al. 2001; Zhu and Joerger 2003). In the present study, we evaluated existing FISH protocols to investigate contribution of dominant bacterial groups to the chicken GIT microflora in relation to production system and the feed additive Salinomycin.

Material and methods

FISH analysis of cultured bacterial reference strains

Bacterial strains used for optimization of the hybridization conditions were obtained from the sources listed in Table 1 and cultured as recommended in the catalogues of the culture collection. All solutions used for washing of cells by centrifugation, fixation and hybridization (including intestinal samples in the following) were filtered (pore size, 0.2 μm) prior to use. Cells were grown for 16 h and harvested by centrifugation at 14,000g for 3 min, washed with phosphate-buffered saline (PBS) and resuspended in 3 volumes of ethanol and PBS (1:1 vol/vol). After fixation for 16 h at 4°C, the cells were washed once with Millipore quality water (MQW) and stored in ethanol and PBS (1:1 v/v) at −20°C until FISH analysis was performed. Three μl of fixed cells were spotted onto 8-well Teflon-coated slides (NovaKemiAB, Enskede, Sweden) and allowed to dry. The cells were then dehydrated by sequential washes in 50, 80 and 100% ethanol (3 min each). Hybridization was based on 10 μl well−1 of a preheated hybridization buffer with final conc. of 0.9 M NaCl, 20 mM TrisHCl adjusted to pH 7.5 with 0.1% w/v SDS and containing 5 ng Cy3-labelled probe μl−1 buffer that was allowed to hybridize for 2 h in a moisture chamber. The probes were synthesized and labelled with Cy3 at the 5′-end (TAGC, Copenhagen) (Table 2). The samples were rinsed with 2 ml hybridization buffer followed by incubation in 35 ml of hybridization buffer for 30 min. Finally, the samples were rinsed with water, air dried and mounted in Slowfade (Molecular Probes, Eugene, U.S.A.) for fluorescence microscopy.
Table 1

Reference strains and results of the optimized FISH analysis

Species

Oligonucleotide probes

Straina

NONEUB338

LGC354B

Bacto1080

Chis150

LGC354C

Strc493

LGC354A

ProbeD

EUB338

Bacillus cereus

ATCC11778

b

+b

+

Bacillus licheniformis

CCUG7422T

+

+

Bacillus megaterium

CCUG1817T

+

+

Bacillus subtilis

ATCC6633

+

+

Listeria monocytogenes

EGC

+

+

Campylobacter jejuni

NCTC11168T

+

+

+

Bacteroides fragilis

CCUG4856T

+

+

Bacteroides vulgatus

DSM1447T

+

+

Clostridium aldrichii IIIc

DSM6159T

+

+

Clostridium cocleatum XVIII

DSM1551T

+

+

Clostridium coccoides XIVa

DSM935T

+

+

Clostridium sartagoforme I

DSM1292T

+

+

Clostridium sordelli XI

DSM2141T

+

+

Clostridium spiroforme XVIII

DSM1552T

+

+

Clostridium sporogenes I

DSM795T

+

+

Clostridium tertium I

DSM2485T

+

+

Clostridium leptum IV

DSM753T

+

Clostridium malenominatum

DSM1127T

+

Enterococcus faecium

CCUG542T

+

+

Enterococcus faecalis

CCUG19916T

+

+

Enterococcus hirae

CCUG19917T

+

+

Enterococcus cecorum

CCUG27299T

+

+

Streptococcus alactolyticus

CCUG27297T

+

+

+

Streptococcus gallolyticus (bovis)

CCUG17828T

+

+

+

Streptococcus intermedius

CCUG17827T

+

+

+

Streptococcus pluranimalium

CCUG43803T

+

+

Streptococcus pleomorphus

CCUG11733T

+

Escherichia coli

ATCC25922

+

+

Salmonella enterica

LT2T

+

Lactobacillus acidophilus

CCUG5917T

+

+

Lactobacillus delbrueckii

CCUG34222T

+

+

Lactobacillus fermentum

CCUG30138T

+

+

Lactobacillus gasseri

CCUG31451T

+

+

Lactobacillus plantarum

CCUG30503T

+

+

Lactobacillus ruminis

CCUG39465T

+

+

Lactobacillus salivarius

CCUG31453T

+

+

Lactobacillus aviarius

CCUG32230T

+

aStrains were obtained from DSM, Deutsche Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany; CCUG, the Culture Collection, University of Goteborg; ATCC, the American. Type Culture Collection, Rockville and NCTC, National Collection of Type Cultures, Health Protection Agency, Centre for Infections, London and Department of Veterinary Pathobiology, Copenhagen

b“+” signal allowing detection, “−” no signal

cGroups from Collins et al. (1994)

Table 2

FISH oligonucleotide probes

Probe

Sequence (5′–3′)

16S rRNA gene region

Groups targeted

References

LGC354A

TGGAAGATTCCCTACTGC

354–371

Mainly members of Lactobacillales and Bacillales such as Lactobacillus

Meier et al. (1999)

LGC354B

CGGAAGATTCCCTACTGC

354–371

Mainly members of Bacillales and Lactobacillales such as Bacillus

Meier et al. (1999)

LGC354C

CCGAAGATTCCCTACTGC

354–371

Mainly members of Lactobacillales such as Enterococcus, Lactococcus and Streptococcus

Meier et al. (1999)

Bacto1080

GCACTTAAGCCGACACCT

1080–1098

Bacteroides, Porphyromonas, Prevotella

Dore et al. (1998)

Strc493

GTTAGCCGTCCCTTTCTGG

439–457

Streptococcus

Franks et al. (1998)

Chis150

TTATGCGGTATTAATCTYCCTTT

150–173

Clostridiaceae

Franks et al. (1998)

Probe D

TGCTCTCGCGAGGTCGCTTCTCTT

1251–1274

Enterobacteriaceae

Ootsubo et al. (2002)

EUB338

GCTGCCTCCCGTAGGAGT

338–355

Bacteria

Stahl et al. (1988)

Non-EUB338

CGACGGAGGGCATCCTCA

338–355

Negative control

Wallner et al. (1993)

In order to ensure maximal stability and permeability of bacteria, the optimal fixation procedure was tested with cultures of Escherichia coli and Salmonella enterica representing bacteria with a Gram-negative cell wall and cultures of Bacillus cereus, B. subtilis, Enterococcus faecium, Ent. faecalis, Ent. hirae and Listeria monocytogenes representing Gram-positive bacteria. The effect of fixation of bacteria with 3 volumes of 4% paraformaldehyde (PFA) or ethanol and PBS (1:1 v/v) at 4 or 20°C were further investigated. To investigate the necessity of employing lysozyme for the hybridization of Gram-positive bacteria, cultures of Gram-negative (S. enterica and E. coli) or Gram-positive (L. monocytogenes and Streptococcus spp.) bacteria were treated with lysozyme or left untreated and subsequently hybridized with EUB338.

Sampling and FISH of chicken intestinal content and sections

Danish broilers of the type Ross 308 were obtained at the age of 28 days from a conventional production system in the Autumn of 2006. They had been fed a conventional wheat and soy based diet without or with 70–75 mg Salinomycin kg−1 feed. From each of the two treatments with or without Salinomycin two male and two female birds were compared individually. After killing the chickens by cervical dislocation, the birds were positioned on their back and opened with sterile scissors. The gastro-intestinal tract was removed aseptically and the site of sampling identified. The specific part of ileum or cecum was removed antiseptically and the content emptied into sterile containers by gentle squeezing. The ileum was sampled between Meckel’s diverticulum and the ileo-caecal junction. Samples were transferred to sterile containers and processed within a maximum of 10 min.

Half a gram of gut-material was suspended in 3 ml of ethanol:PBS (1:1 v/v) and incubated for 16 h at 4°C. After fixation, the sample was washed once with MQW by centrifugation (7,000g, 10 min) and resuspended in 3 ml ethanol:PBS (1:1 v/v). For homogenization of each sample, five borosilicate solid-glass beads (3 mm) (Sigma) were added and the sample vortexed for 5 min. The sample was then distributed in aliquots corresponding to 0.1 g of the original weight of the sample and stored at −20°C until FISH analysis. Samples were hybridized within one month from sampling. The protocol used was modified from previous protocols published by Langendijk et al. (1995), Franks et al. (1998) and Christensen et al. (1999). Briefly, hybridization of intestinal content was performed by washing a sub-sample corresponding to 1.5 mg fresh weight by centrifugation with MQW (14,000g, 5 min) and resuspending the sample in 90 μl pre-heated hybridization buffer with 10 ng μl−1 Cy3-labelled probe. The cells were hybridized for 24 h in the darkness at the appropriate temperature with rotation. Each sample was then centrifuged (14,000g, 10 min) and washed with 1 ml pre-heated hybridization buffer at the appropriate temperature with rotation for 20 min. The sample was finally centrifuged (14,000g, 10 min) and resuspended in 5 ml MQW before filtration onto 0.22 μm black polycarbonate filters (Osmonics Inc., Minnetonka, MN, USA). Finally, each filter was washed twice with 5 ml MQW, and processed immediately for epifluorescence microscopy or stored at −80°C until analyzed. One sub-sample from each of the conventional birds was hybridized with each probe (Table 2).

To analyze bacteria attached to the epithel-mucus of ileum and cecum, the intestines were opened with sterile scissors, rinsed free for intestinal content with PBS, fixed for 24 h in 10% neutral buffered formalin, stored in 70% ethanol until embedded in paraffin wax, sectioned at 3–4 μm and mounted on Super Frost/Plus slides (Menzel-Gläser, Braunschweig, Germany) as described by Boye et al. (1998). The procedure secured the conservation of the intestinal mucus layer. Paraffin wax was removed by xylen and the tissue sections were rinsed twice with 99% ethanol for 3 min and allowed to air-dry. One hundred μl hybridization buffer with 2 ng μl−1 of the respective probes (Table 2) was applied and hybridization performed for 24 h. All samples were hybridized within one month after sampling and slides stored at −80°C until fluorescence microscopy was carried out.

Epifluorescence microscopy and cell counts

The filters and slides were examined with an Axioplan II epifluorescence microscope (Zeiss, Oberkochen, Germany) equipped for epifluorescence with a 100-W mercury lamp and filterset 15 (Zeiss) for visualization of Cy3 and a Zeiss AxioCam digital camera. Bacterial shaped cells were scored in three size classes according to Christensen et al. (1999). At least 1:5,000 of the filter area was inspected by moving the specimen stage at random and inspecting small fields of a cross-field graticule counting at least 50 cells per filter. Bacteria in the villi and mucus layers were counted the same way as those on filters, by randomly selecting microscopic fields and counting the bacteria shaped cells.

Bioinformatical comparisons and statistical treatment of data

For theoretical evaluation of 16S rRNA oligonucleotide probes in relation to sequence targets, BLAST was used (Altschul et al. 1997). Determination of pair-wise similarity was performed by the program WATER included in EMBOSS (Rice et al. 2000). Statistical comparison of counts between treatments groups (Salinomycin) was performed by the Mann–Whitney U-test at 95% level.

Results

FISH analysis of cultured bacterial reference strains

It was important to optimize the fixation procedure in order to ensure maximal stability and permeability of bacteria both with a Gram-positive and Gram-negative cell wall structure. The result of the fixation procedure on cultured bacteria was evaluated by subsequent hybridization with the Cy3-labelled probe EUB338. Best results were obtained with a mixture of ethanol and PBS (1:1 v/v) at 4°C for 16 h. Fixation in ethanol and PBS at 20°C resulted in increased clumping of cells and was therefore not considered useful.

The optimal hybridization conditions were found by performing hybridization at 45°C with all probes except for EUB338 and Non-EUB338 that were hybridized at 37°C. Formamide was not used with Bacto1080 and Non-EUB338 probes whereas 10% formamide was used in the hybridization buffer with LGC354C, 15% with Chis150 and EUB338 and 25% formamide with LGC354A, LGC354B, Strc493 and ProbeD probes. The hybridization conditions of probe EUB338 were in accordance with the original publication (Amann et al. 1990), however, 5–10°C higher hybridization temperatures were required with probes Strc493 and Chis150 compared to originally reported by Franks et al. (1998) in order to obtain signal with the tested bacterial reference strains. The hybridization temperature was set 5–10°C lower with LGC354A, LGC354B, LGC354C, ProbeD and Bacto1080 compared to original publications (Meier et al. 1999; Ootsubo et al. 2002; Zhu and Joerger 2003). These comparisons are based on a decrease in Tm approx. 0.7°C by 1% formamide (McConaughy et al. 1969).

Probe LGC354A was able to detect seven out of eight species of Lactobacillus tested. The lack of detection of Lact. aviaries cannot be explained since the probe was found to have perfect match to the 16S rRNA sequence published (acc. no. M58808). Unfortunately, a signal was also obtained with Campylobacter jejuni. Database search showed two mismatches of probes LGC354A and LGC354B to the campylobacter, however, this was not sufficient to avoid detection of C. jejuni even after optimizing hybridization conditions. With respect to bacteria expected to be present in the chicken GIT, probe LGC354A was predicted to detect members within the genera Lactobacillus, Leuconostoc, Pediococcus, belonging to order Lactobacillales, some Bacillus, and Weissella belonging to order Bacillales and probably others with only single mismatches (Meier et al. 1999). In the chicken GIT, probe LGC354A most likely targets members of Lactobacillus and some related groups in the following referred as Lactobacillus-like.

Signals were obtained with the LGC354B probe from the reference strains of Bacillus species tested and from List. monocytogenes. As for the LGC354A probe a signal was also obtained with C. jejuni (Table 1). In relation to bacteria expected to be present in the chicken GIT, LGC354B was predicted to detect members within the genera Aerococcus, Carnobacterium, Globicatella and some Lactobacillus belonging to order Lactobacillales, most Bacillus, Gemella, Listeria,Paenibacillus, and Staphylococcus belonging to order Bacillales and probably others with only single mismatches (Meier et al. 1999). In the chicken GIT, probe LGC354B most likely mainly targets members of Bacillus and related groups in the following referred as Bacillus-like.

The LGC354C probe provided signals with all three species of Enterococcus tested. Signal was also obtained from this probe by hybridization to four out of five Streptococcus species. The lack of signal from Strep. pleomorphus was probably related to two mismatches to the probe. In relation to bacteria expected to be present in the chicken GIT, LGC354C was predicted to detect members of the genera belonging to order Lactobacillales such as Enterococcus, Lactococcus, Streptococcus, and Vagococcus (Meier et al. 1999). In the chicken GIT, probe LGC354C most likely targets Enterococcus, Streptococcus, Lactococcus and related groups in the following referred as Enterococcus, Streptococcus and Lactococcus-like.

The probe Strc493 hybridized only to Strep. alactolyticus, Strep. gallolyticus and Strep. intermedius in correspondence with Franks et al. (1998). They also found positive detection of other species of Streptococcus and of Lactoc. lactis and Leuc. lactis. Since the group of bacteria detected by Strc493 was nested within the group detected by LGC354C we only used LGC354C.

The expected signal was obtained with the Bacto1080 probe from the two species of Bacteroides tested (Table 1). In addition, the probe was predicted to detect other Bacteroides species, Porphyromonas and Prevotella (Dore et al. 1998).

For the Clostridia, signal was obtained with probe Chis150 for 11 type strains, however, not for Cl. leptum and Cl. malenominatum. Comparison between the probe sequence and 16S rRNA sequences of the two species (acc. no. AF262239 and M59099) showed three and two mismatches, respectively, which might explain the lack of detection. The probe Chis150 was described by Franks et al. (1998) and was predicted to target Clostridium clusters I and II. Our testing of reference strains further showed targeting of members of clusters III, XI, XIVa and XVIII also but not of cluster IV. ProbeD was specific for E. coli and is expected to detect most Enterobacteria (Ootsubo et al. 2002), although S. enterica was not covered.

Effect of fixation and lysozyme treatment of chicken intestinal material

A slightly stronger hybridization intensity was observed when the sample had been fixed with ethanol and PBS (1:1 v/v) and microscope counting revealed double the number of bacteria following fixation in ethanol and PBS compared to PFA (data not shown). Therefore, we proceeded with fixation of intestinal samples with an equal mixture of ethanol and PBS at 4°C for 16 h. We also investigated the necessity of employing lysozyme for the hybridization. The treatment with 0.2 μg ml−1 lysozyme at 0°C for 10 or 25 min lysed E. coli and S. enterica, in accordance with expectations whereas no morphological difference was observed regarding the Gram-positive bacteria. These results applied to fixation in ethanol and PBS. Apart from this, subsequent hybridization with EUB338 and counting of cells showed no advantage when the Gram-positive cells were treated with lysozyme compared to untreated cells. Quite opposite, treatment with lysozyme resulted in a higher fluorescent background during subsequent hybridization. Therefore, a pre-treatment with lysozyme was not employed in the hybridization protocol in this study.

Analysis of broilers

For the domain Bacteria (EUB338 probe), remarkably similar numbers of around 4.5 × 108 cells g−1 of ileum content were determined for controls without Salinomycin and Salinomycin treated broilers (Table 3). Mean numbers of bacteria in the lumen of ileum varied from 5.60 × 106 cells g−1 luminal content for Clostridium-like bacteria (Chis150 probe) up to 6.61 × 108 cells g−1 for Enterobacteriaceae-like bacteria (ProbeD probe) (Table 3). Salinomycin in the feed reduced counts of the Bacillus-like (LGC354A), Enterobacteriaceae-like (ProbeD) and Clostridium-like groups (Chis150) in the lumen of ileum by one order of magnitude compared to counts in the ileum of birds reared without the coccidiostat (Table 3).
Table 3

Bacterial numbers of conventional broilers in lumen of ileum according to 16S rRNA group specific FISH oligonucleotide probes

Sample

Oligonucleotide probe and bacterial group detected (×107 cells g−1)

EUB338

LGC354A

LGC354B

LGC354C

ProbeD

Bacto1080

Chis150

Bacteria

Lactobacillus

Bacillus

Enterococcus, Lactococcus, Streptococcus

Enterobacteriaceae

Bacteroides, Porphyromonas, Prevotella

Clostridiaceae

With Salinomycin

42.1a (38.0)

[100]

5.21 (2.74)

[12.4]

8.26 (2.97)

[19.6]

65.4 (37.9)

[155]

2.59 Ab (0.547)

[6.15]

12.2 (12.1)

[29.0]

0.56 A (0.309)

[1.33]

Without Salinomycin

44.9 (32.5)

[100]

46.3 (24.5)

[103]

21.4 (10.2)

[47.7]

33.7 (7.99)

[75.1]

29.4 B (10.2)

[65.5]

7.76 (4.42)

[17.3]

6.62 B (3.34)

[14.7]

aMean of four individual birds with SD in parenthesis. % of counts for each probe compared to EUB338 have been shown in brackets

bSignificant differences (U-test) in counts with the same oligonucleotide probe are indicated by different letters

In the cecum samples (Table 4), counts from 3.74 × 107 cells g−1 were determined for Clostridium-like bacteria (Chis150 probe) in the non-treated control up to 34.8 × 109 cells g−1 for the domain Bacteria (EUB338 probe) for the treated group. Counts of Bacillus-like bacteria (probe LGC354B) were significantly reduced by application of Salinomycin to feed, whereas the Streptococci,Enterococcus and Lactococcus-like group (LGC354C) and Clostridium-like bacteria (Chis150) were increased by the application of Salinomycin to the feed.
Table 4

Bacterial numbers of conventional broilers in lumen of cecum according to 16S rRNA group specific FISH oligonucleotide probes

Sample

Oligonucleotide probe and bacterial group detected (×109 cells g−1)

EUB338

LGC354A

LGC354B

LGC354C

ProbeD

Bacto1080

Chis150

Bacteria

Lactobacillus

Bacillus

Enterococcus, Lactococcus, Streptococcus

Enterobacteriaceae

Bacteroides, Porphyromonas, Prevotella

Clostridiaceae

With Salinomycin

34.8a (7.00)

[100]

5.18 (1.71)

[14.9]

3.83 Ab (1.08)

[11.0]

2.11 A (0.972)

[6.06]

0.283 (0.161)

[0.813]

0.262 (0.230)

[0.753]

0.137 A (0.0271)

[0.394]

Without Salinomycin

32.1 (7.01)

[100]

23.0 (16.2)

[71.7]

24.3 B (17.7)

[75.7]

0.212 B (0.0197)

[0.660]

0.0854 (0.101)

[0.266]

0.210 (0.0859)

[0.654]

0.0374 B (0.00958)

[0.117]

aMean of four individual birds with SD in parenthesis. % of counts for each probe compared to EUB338 have been shown in brackets

bSignificant differences (U-test) in counts with the same oligonucleotide probe are indicated by different letters

Counts of bacteria in tissue sections (Table 5) were estimated on a volume basis, based on information of tissue thickness and size of tissue section viewed under the microscope assuming a weight to volume ratio of one. Numbers varied from 6.20 × 106 cells g−1 tissue with the LGC354B probe targeting Bacillus-like bacteria for ileum in non-treated control up to 28.0 × 107 for the Bacto1080 probe targeting Bacteroides-like bacteria for cecum in the control group. Although there was a trend toward higher counts in the control group compared to the group fed Salinomycin in the diet, this difference was only significant with the Bacteroides- (Bacto1080) and Enterobacteriaceae-like bacteria (ProbeD) for cecum samples.
Table 5

Bacterial numbers of conventional broilers of ileum and cecum tissue sections according to 16S rRNA group specific FISH oligonucleotide probes

Region

Sample

Oligonucleotide probe and bacterial group detected (×107 cells g−1 tissue)

EUB338

LGC354A

LGC354B

LGC354C

ProbeD

Bacto1080

Chis150

Bacteria

Lactobacillus

Bacillus

Enterococcus, Lactococcus, Streptococcus

Enterobacteriaceae

Bacteroides, Porphyromonas, Prevotella

Clostridiaceae

Ileum

With Salinomycin

26.1a (5.99)

[100]

6.95 (3.24)

[26.6]

1.37 (0.795)

[5.25]

4.37 (1.62)

[16.7]

4.9 (7.97)

[18.8]

5.83 (1.94)

[22.3]

2.16 (0.270)

[8.28]

Ileum

Without Salinomycin

20.0 (11.3)

[100]

3.77 (1.57)

[18.9]

0.62 (0.588)

[3.10]

3.40 (0.956)

[17.0]

0.93 (0.827)

[4.65]

5.58 (2.07)

[27.9]

1.07 (0.574)

[5.35]

Cecum

With Salinomycin

40.2 (16.4)

[100]

9.94 (8.22)

[24.7]

0.938 (0.525)

[2.33]

4.13 (3.15)

[10.3]

1.40 Ab (0.84)

[3.48]

7.78 A (3.56)

[19.4]

2.80 (3.37)

[6.70]

Cecum

Without Salinomycin

51.2 (14.7)

[100]

15.8 (12.5)

[30.9]

2.96 (5.25)

[5.78]

4.34 (1.25)

[8.48]

3.37 B (1.43)

[6.58]

28.0 B (8.22)

[54.7]

6.12 (3.50)

[12.0]

aMean of four individual birds with SD in parenthesis. % of counts for each probe compared to EUB338 have been shown in brackets

bSignificant differences (U-test) in counts with the same oligonucleotide probe for either ileum or cecum are indicated by different letters

The intra assay coefficient of variation (cv) between triplicate FISH determinations of the same luminal sample was 33% as a mean. This variation seems to be related to the small sample size analyzed on each filter. However, the variation was less than the 50% cv found between the four individuals (calculated from Table 3). Counting of the thin sliced tissue sections resulted in a cv of 58% between four different animals (calculated from Table 5). Higher cv when counting the tissue section directly compared to counting on the filter is expected due to more uneven distribution of cells as observed on Fig. 1 since many bacteria were located in micro-colonies.
https://static-content.springer.com/image/art%3A10.1007%2Fs10482-008-9260-0/MediaObjects/10482_2008_9260_Fig1_HTML.jpg
Fig. 1

FISH of tissue sections of ileum (a, c, e, g) or cecum (b, d, f, h) of 28-day-old conventional broilers given feed without Salinomycin (a, b, e, f) or with Salinomycin (c, d, g, h) and hybridized with the general Bacterial probe EUB338 (ad), Bacto1080 (e, g) or LGC354A (f, h)

Some of the different bacterial morphologies observed are shown in Fig. 1 for ileum (Fig. 1a, c, e, g) and for the cecum (Fig. 1b, d, f, h), respectively. As mentioned, it was characteristic that the bacteria often were present as microcolonies (Fig. 1b–e, h) and that most bacteria in the ceca were present in the mucus layer rather than in the crypts (Fig. 1h).

Based on light microscopic observations of hematoxylin-eosin stained tissue sections, the height of the villi-plus-mucus layer for the cecum was significantly reduced from 105 to 47 μm in conventional broilers fed Salinomycin in the diet compared to the control. The height of villi-plus-mucus layer in ileum was unaffected by treatment with a mean height of 128 μm.

Knowing the height of the combined villi and mucus layer and the intestinal diameter of around 4 mm measured post mortem, an estimate about the relative proportion of volume of the epithel-mucus region in which the bacteria might be present compared to lumen was calculated considering a given cross section of the intestine. Adjusting bacterial counts for this difference resulted in the relative differences between bacteria in the epithel-mucus and lumen as shown on Fig. 2. For most groups and especially for the cecum, the proportion of bacteria in the villi plus mucus layer was below 1% compared to the lumen. The most remarkable difference was seen on ileum samples for the groups of Lactobacillus-like (LGC354A), Enterobacteriaceae-like (ProbeD), and Clostridia-like (Chis150) with much higher relative proportions in the villi-plus-mucus layer compared to lumen especially for the Salinomycin treatment.
https://static-content.springer.com/image/art%3A10.1007%2Fs10482-008-9260-0/MediaObjects/10482_2008_9260_Fig2_HTML.gif
Fig. 2

Relative proportions of bacteria attached to the epithel-mucus layer of the intestinal wall compared to lumen for a given cross-section of the ileum or cecum

Discussion

The protocol used for FISH analysis of luminal content of chicken GIT was based on hybridization of a dilute homogenate. Centrifugation allowed exchange of hybridization buffers and removal of excess probe while retaining all particles of bacterial size. Subsequent filtration onto 0.22 μm pore-size black pre-stained polycarbonate membranes retained bacterial sized particles and reduced background including excess probe. Compared to protocols of Langendijk et al. (1995) and Franks et al. (1998), the improvements included optimized procedures for fixation and permeabilization of bacteria in the GIT of chicken and quantification of a range of relevant bacterial groups. The protocol was also applicable to fecal samples with the same parameters as used for intestinal content (data not shown). Intestinal samples were preferred for analysis over fecal ones in the current study since fecal samples might be biased by floor litter contamination and fecal samples do not allow separation of ileum from cecal material. The major benefit of the protocol was quantitative recovery of bacteria from the original sample and conversion of bacterial numbers to original wet weight of intestinal matter. In principle these numbers might be used to estimate the metabolic load of the bacterial microbiota on the chicken when combined with estimates of bacterial growth rate and dynamics of movement of luminal content in the intestine.

Despite use of the FISH technique for nearly two decades, development and validation of probes are still needed. One problem in this study was the detection of C. jejuni by LGC354A and LGC354B probes expected to cover bacteria related to of Lactobacillus and Bacillus only. Alternatives to these probes targeting very important bacterial groups in the chicken GIT were not found in the literature. Although C. jejuni was predicted to be included with counts obtained by LGC354A and LGC354B probes, cellular morphology similar to C. jejuni was never observed with GIT samples investigated and for this reason, numbers of C. jejuni were not included with groups counted by these probes. Probes were further tested to document targeting of the expected groups. Chis150 was predicted to target Clostridium-like clusters I and II (Franks et al. 1998). Testing of a range of type strains of other species showed that at least groups III, XI, XIVa and XVIII were covered also. These groups include the potential pathogenic species, Cl. perfringens or species of Clostridium isolated from sporadic cases of disease only (Barnes 2003). Clostridium cluster IV found dominant in other investigations (Lan et al. 2002), was probably not covered since between two mismatches and one gap up to 4 mismatches and 8 gaps were found to differ between probe Chis150 and the species of this cluster listed by Collins et al. (1994). The species Cl. leptum was found dominant in cecal samples by Zhu and Joerger (2003) and Bjerrum et al. (2006). Unfortunately probe Chis150 was not documented to target this group. Signal was obtained with probe Chis150 for 11 type strains, however, not for Cl. leptum and Cl. malenominatum. Comparison between the probe sequence and 16S rRNA sequences of the two species (acc. no. AF262239 and M59099) showed three and two mismatches, respectively, which might explain the lack of detection. Two mismatches were probably also the reason for not detecting Strep. pleomorphus with the LCG354C probe. However, this is contrary to the mentioned detection of C. jejuni by both LGC354A and LGC354B probes despite of two mismatches. ProbeD was specific for E. coli and is expected to detect most Enterobacteriaceae (Ootsubo et al. 2002). A perfect match to S. enterica was obtained by database search although this bacterium could not be detected in the culture experiments (see Table 1). Further explanation cannot be give at present. The main limitation in the current investigation was the lack of well tested FISH probes with really well defined targets within the chicken GIT. Further work is needed to optimize the specificity of these probes by incorporating knowledge from culture independent sequence libraries of the chicken GIT since the chicken GIT is poorly investigated in regard to bacterial diversity realized.

Our study showed that careful evaluation need to be performed when FISH is used for diagnostic purpose of GIT bacteria since unexpected targets might be present in this environment. Although probes have been evaluated against a range of potential positive and negative controls, bacteria not yet characterized might be present in the chicken GIT.

Even though our probes were not expected to cover all bacterial groups of the chicken GIT, the accumulated sum of counts for these group some time added up higher than counts obtained with the EUB338 probe expected to target all members of the domain Bacteria. Previous investigations have documented that the relative intensity between 16S rRNA targeted oligonucleotide probes differs (see for example Yilmaz et al. 2006) and this is the most probable explanation for the lack of cumulative counts for the different bacterial groups.

Quantification of bacterial groups by FISH documented differences within the conventional production system related to the application of the coccidiostat Salinomycin. Numbers determined for chicken ileum and cecum of around 4.5 × 108 and 3.4 × 1010, respectively with the EUB338 probe targeting the Bacteria domain are in agreement with published total microscopic counts of 108–109 cells g−1 ileum content and 1011 cells g−1 of cecal content (Apajalahti et al. 1998, 2004; Gong et al. 2002a, b; van der Wielen et al. 2002).

There have only been limited investigations on the effect of AGPs and ionophore type coccidiostats on the relative proportions in the chicken GIT groups. The decreased counts of Clostridium-like bacteria obtained by probe Chis150 of lumen of ileum with Salinomycin compared to the control might reflect a potential lower level of Cl. perfringens also since the probe is expected to cover the species. This is in accordance with results obtained by cultural methods (Bjerrum et al. 2006). Lower prevalence of Cl. perfringens may decrease the incidence of intestinal disorders and this might be a side effect of the Salinomycin (Van Immerseel et al. 2004). For the conventional broilers, numbers within Bacillus-like and Enterobacteriaceae-like groups were also reduced in the lumen of the ileum following the Salinomycin treatment. For cecum, both reductions as well as increases in bacterial counts as a consequence of Salinomycin treatment were obtained.

Our results are in general accordance with previous publications reporting members of Lactobacillus, Clostridiaceae, Enterococcus and Streptococcus to be dominant in the ileum, however, we also found high frequency of members of the Enterobacteriaceae-like as well as of Bacteroides-like bacterial groups. All six groups were also found present in high numbers in cecum samples from the conventional broilers in accordance with previous reports (Apajalahti et al. 2001; Bjerrum et al. 2006; Gabriel et al. 2006; Gong et al. 2002a, b, Lan et al. 2002, 2005; Lu et al. 2003, Zhu et al. 2002). We did not investigate members of Fusobacterium and Ruminococcus reported dominant from ceca (Bjerrum et al. 2006; Gong et al. 2002a, b) since FISH probes were not available, not functioning or not tested for these groups. Five other published probes were testes but not found to give reliable results against the set of reference bacteria tested (data not shown).

The reason for the investigation of bacteria attached to mucus and the intestinal epithelium was limited information regarding both origin and microbial ecology of these bacteria. Assuming a constant diameter of the intestine and lack of movement of intestinal content after death, comparison of cell numbers in the lumen with the villi-plus-mucus layer showed bacteria of Enterobacteriaceae-like (ProbeD), Bacteroides-like (Bacto1080) and Clostridium-like (Chis150) groups to be relatively more frequent in the villi-mucus layer compared to lumen, especially in the ceca (see Fig. 2). Furthermore, these bacterial groups seems to be relatively better protected in the intestinal mucus layer against the bactericidal effect of Salinomycin compared to luminal bacteria. Their overrepresentation compared to typical Gram-positive genera such as Lactobacillus, Bacillus, Enterococcus, Streptococcus and Lactococcus has implications both in the understanding of the intestinal ecology and the need to perform further investigations that should focus both on the microbial ecology of the intestinal epithel-mucus layer as well as the lumen. For further quantitative investigations, it will be important to include bacteria attached to the intestinal epithel-mucus of the ileum. Bacteria in the lumen are moving through the GIT and might include an unknown fraction of bacteria carried over from feed and distal GIT sections that might have survived the distal parts of the GIT including the gizzard. On the other hand bacteria associated villi and mucus are more sedentary (see Fig. 1) and are probably under stronger influence of host cell components. For example, the mucin-type glycoproteins of the mucus layer are believed to prevent the attachment of pathogenic bacteria by modulating their adherence to the intestinal epithelium (Uni 2006).

Acknowledgements

Jens Peter Christensen is acknowledged for instructions about dissection of birds. Ricarda Engberg, Department of Animal Nutrition and Physiology, Danish Institute of Agricultural Sciences, Research Centre Foulum, Denmark; Jette Søholm Petersen, Landscentret, Fjerkræ, Århus; Henning Fynboe Madsen, Horsens and Karen Margrethe Balle, Landscentret, Fjerkræ, Århus are acknowledged for their kind assistance to get access to the poultry and to the detailed data on production parameters. The work is part of the EU project, POULTRYFLORGUT for the thematic call FP6-2003-FOOD-2-A T1.2.

Copyright information

© Springer Science+Business Media B.V. 2008