Culturable Gut Microbiota Diversity in Zebrafish

Abstract

The zebrafish (Danio rerio) is an increasingly used laboratory animal model in basic biology and biomedicine, novel drug development, and toxicology. The wide use has increased the demand for optimized husbandry protocols to ensure animal health care and welfare. The knowledge about the correlation between culturable zebrafish intestinal microbiota and health in relation to environmental factors and management procedures is very limited. A semi-quantitative level of growth of individual types of bacteria was determined and associated with sampling points. A total of 72 TAB line zebrafish from four laboratories (Labs A–D) in the Zebrafish Network Norway were used. Diagnostic was based on traditional bacterial culture methods and biochemical characterization using commercial kits, followed by 16S rDNA gene sequencing from pure subcultures. Also selected Gram-negative isolates were analyzed for antibiotic susceptibility to 8 different antibiotics. A total of 13 morphologically different bacterial species were the most prevalent: Aeromonas hydrophila, Aeromonas sobria, Vibrio parahaemolyticus, Photobacterium damselae, Pseudomonas aeruginosa, Pseudomonas fluorescens, Pseudomonas luteola, Comamonas testosteroni, Ochrobactrum anthropi, Staphylococcus cohnii, Staphylococcus epidermidis, Staphylococcus capitis, and Staphylococcus warneri. Only Lab B had significantly higher levels of total bacterial growth (OR=2.03), whereas numbers from Lab C (OR=1.01) and Lab D (OR=1.12) were found to be similar to the baseline Lab A. Sexually immature individuals had a significantly higher level of harvested total bacterial growth than mature fish (OR=0.82), no statistically significant differences were found between male and female fish (OR=1.01), and the posterior intestinal segment demonstrated a higher degree of culturable bacteria than the anterior segment (OR=4.1). Multiple antibiotic (>3) resistance was observed in 17% of the strains. We propose that a rapid conventional diagnostic bacteriological assay on the culturable microbiota profiles can be designed and used as quality measure of the husbandry routines of a zebrafish facility to ensure a bacterial standard safeguarding the zebrafish health and welfare.

Introduction

Eleven decades ago, E. Metchnikoff¹ mentioned that ‘the activity of the intestinal microbiota is correlated with the longevity of host’. Since then, the intestinal microbiota of animals and humans has been studied continuously.^2,3 Until the 1970s, a concrete report on the existence of a stable indigenous microbiota was not available for many aquatic animals,⁴ and it was believed that fish microbiotas were simpler and less complex than in endothermic animals.^5,6 However, during the last 20 years, the gastrointestinal (GI) tracts of a wide range of fish species have been characterized with the use of both traditional bacteriological methods^7,8 and modern current culture-independent technologies.⁹ So far, the gut microbiota of mammalian or aquatic animals has not been fully described, but it is clear that microorganisms are present in site-specific communities in the intestinal lumen and are crucial for health-associated homeostasis,¹⁰ host metabolism,¹¹ intestinal development,¹² neurodevelopment,¹³ and general evolution of intestinal ecosystems.¹⁴

The zebrafish model has been introduced in a wide variety of research fields^15–17 including studies of the importance of the gut microbiota composition including the generation and use of gnotobiotic zebrafish.¹⁸ Comparative metagenomic profiling of zebrafish and mouse gut microbiotas revealed the presence of Proteobacteria, Firmicutes, Bacteroidetes, Verrucomicrobia, Actinobacteria, and Planctomycetes divisions in both species.¹⁹ However, gut microbiota in domesticated and recently caught wild zebrafish were found to be similar, with a shared core gut microbiota identified by 16S rRNA gene sequence-based comparisons.²⁰ Still the normal composition of the culturable zebrafish intestinal microbiota has remained largely uncharacterized. Therefore, we chose the widely used AB/TU TAB14 (TAB) wild-type strain of zebrafish (http://zebrafish.org/zirc), which was established as standard laboratory fish stock among four different facilities (named Lab A, B, C, and D) belonging to the Zebrafish Network Norway (ZNN; http://zebrafish.no) for comparison of the composition of the culturable proportion of the intestinal microbiota in different zebrafish facilities. We evaluated the effect of the status of sexual maturity and gender on the composition of the intestinal microbiota. Further, the impact of husbandry introduced factors such as bacterial contents of the feed, the re-circulated water, and the water in the fish tanks were analyzed. In addition, phenotypical antibiotic resistance profiles of selected intestinal isolates were determined and compared among facilities. Antibiotic resistance has been documented to occur in bacteria related to ornamental fish living in aquaria and has been suggested to be a result of the use of antibiotics in the management of the fish health.²¹ However, to our knowledge, the occurrence of antimicrobial resistance in the normal microbiota of established zebrafish lab strains like the TAB, has not previously been reported.

Materials and Methods

Study design

The study was performed as an Open and Single Central Trial with a Stratified, Sequential Parallel Group Design.^22,23 Four separate facilities belonging to the ZNN were used as stratification factor. The reference population of the study consisted of the TAB strain zebrafish which was divided into two main clusters with regards to sexual maturity (<4 months: immature/adolescent and >6 months: mature/adult) and gender (male/female). With the feeding regime of the facilities, the age of sexual maturation is starting at 4 months post fertilization. The variation in the ages of the mature zebrafish was less than 2 weeks. A total of 72 TAB zebrafish (Lab A–C=20, Lab D=12) were studied. For each of the 4 Labs, both male and female fish were stocked together in 4 identical tanks (2 of each of Tank_mature and Tank_immature). After morning feeding, fish were starved approx. 6 h prior to sampling. Facility records of water quality, environmental parameters, feed, and light regimes were provided by each laboratory. The year of facility establishment and the manufacturer of the rearing systems were included in the records.

Sampling procedure

An overview over the microbiota sampling protocol design is shown in Figure 1. Sterile latex gloves or latex gloves wiped with 70% ethanol were used throughout the dissection, and the ventral side of the fish was sterilized using 70% ethanol. The fish were killed by decapitation. The ventral belly surface of the freshly euthanized fish was opened with sterilized micro surgical blade and forceps under a light source. The spleen, gall bladder, liver, and fat deposits surrounding the GI tract were gently removed to expose the peritoneal cavity without disturbing the intestine. Then according to Wallace et al.²⁴ and Oehlers et al.,^24,25 the intestinal system was divided into an anterior (including mid) and a posterior segment. This division of sample origins was used throughout the study. At each sampling point the divided intestinal segments were transferred into separate microfuge tubes and squeezed using sterile 1 μL disposable plastic loops that also were used to streak the specimen on 5% bovine blood agar (Blood agar base no 2, Difco) and Brocalin agar (Merck, Darmstadt, Germany) bacteriological plates, with three serial dilutions (for each dilution line a new sterile disposable plastic loop was used). Dry feed samples were moisturized with sterile 0.9% NaCl, while artemia and water samples were directly used for spreading on to the plates. The plates were incubated at 28°C (±1) and followed by inspections every third day for up to 4 weeks.

FIG. 1.

Overview samples analyzed for microbiota bacteria profile. (A) Samples were taken from anterior (1–2–3) and posterior (4–5–6) part of GI tract (dissected zebrafish intestine picture is taken from Oehlers et al.²⁵). (B) Samples (S) taken from the rearing unit: (S tank pre UV), (S tank post UV), (S dry feed), (S artemia).

Bacterial isolation and identification

After incubation, each morphologically distinct colony (form, size, surface, color, texture, elevation, margin, odor, and hemolysis) was encoded. From each sampling plate (age, gender, and part of the intestine) two representative colonies of each colony type were randomly selected and subcultured on a separate blood agar. Pure colonies were checked for motility under a light microscope (Leitz, 301–314.001, Germany) at 1000X magnification using a glass slide with suspended and Gram stained exponentially grown bacteria in physiological salt water.²⁶

Gram-positive cocci were tested for coagulase activity. Gram-negative bacteria were analyzed with the IMViC reactions; indol production, methyl red test (fermentation of glucose), Voges-Proskauer reaction (diacetyl-production), and the ability to use citrate as the sole carbon source (Simmons citrate medium) were tested. Catalase and oxidase activity, H₂S production, urease and β-galactosidase (ONPG) production and ability to degrade glucose, trehalose, lactose, maltose, mannitol, and esculin were tested for most of the isolates and susceptibility to the vibriostatic agent (O129) were tested for isolates in the family Vibrionaceae (Table 1).^27–29 Also the Staph-Zym System (ROSCO, Taastrup, Denmark) was applied for staphylocci isolates. Additionally, susceptibilities to the antibiotics novobiocin (5 μg) and polymyxins (Neo-Sensitabs, ROSCO) were determined on Mueller-Hinton agar by standard methods. Procedures were performed as directed by the manufacturer.

Table 1.

Phenotypical and Biochemical Properties of Isolates

Gram stain and colony descriptions	Haemolysis	Motility	Catalase	Oxidase	Coagulase	Indole	Methyl-red	Voges-Proskauer	Citrate utilization	H₂S production	Urea hydrolysis	O/129 disk	ONPG	Glucose	Trehalose	Lactose	Maltose	Mannitol	Esculin	Organisms
GN straight rods with rounded ends approaching a spherical shape (Cream, round, raised, entire colonies 2–3 mm)	V^a	+	+	+	N	V	−	+	V	V	−	−	+	+	+	V	+	+	+	A. hydrophila (Api 20 NE code: 7777755, 99.9%)
GN bacilli or coccobacilli (Cream, round, raised, entire colonies 2–3 mm)	−	−	+	+	N	V	−	V	V	V	−	−	V	+	V	−	V	+	−	A. sobria (Api 20 NE code: 7176755, 99.2%)
GN short rods (Convex, smooth colonies with entire edges 2–3 mm colonies)	−	+	N	+	N	+	−	−	−	−	−	+	N	+	+	−	+	+	−	V. parahaemolyticus (Api 20 NE code: 7077644, 99.9%)
GN short rods (Shiny, gray-yellow, entire, convex colonies 1–2 mm)	V	V	+	+	N	−	−	−	−	N	+	+	N	+	−	−	+	−	−	P. damselae (Api 20NE code: 5300044, 99.9%)
GN straight or slightly curved rods, but not helical (Large, smooth, with flat edges flat and elevated center 1–3 mm colonies)	+	+	+	+	N	−	−	−	+	N	V	N	N	+	−	N	V	+	−	P. aeruginosa (Api 20 NE code: 1154575, 99.5%)
GN rods (Flat, greenish, 1–2 mm colonies)	V^b	−	+	+	N	−	−	−	+	−	V	N	−	V	V	−	V	V	−	P. fluorescens (Api 20 NE code: 0146555, 99.4%)
GN rods (Typically rough, wrinkled yellow to orange colonies, 1 mm colonies)	−	+	+	−	N	−	−	−	+	−	−	N	+	+	−	+	−	+	+	P. luteola (Api 20 NE code: 4567651, 99.9%)
GN rods (Tan to brown color, 1–3 mm colonies)	−	+	+	+	N	−	N	N	N	−	−	N	N	−	−	−	−	−	−	C. testosteroni (Api 20 NE code: 1000054, %98.2)
GN rods (Circular, low convex, smooth, shinning, 1 mm entire colonies)	−	+	+	+	N	−	N	N	N	N	V	N	−	−	+	−	−	−	−	O. anthropi (Api 20 NE code: 1241344, 99.9 %:)
GP cocci singly, in pairs (Circular, convex to slightly peaked, smooth, glossy, opaque 4–5 mm colonies)	−	−	+	−	−	N	N	N	N	N	+	N	−	+	+	V	+	V	+	S. cohnii (Staph-ZYM code:0061-1)
GP cocci in pairs and tetrads (Circular, smooth, opaque colonies, 1–3 mm)	−	N	+	−	−	N	N	N	N	N	+	N	−	+	−	+	+	−	−	S. epidermidis (Staph-ZYM code:0611-5)
GP cocci in pairs and tetrads (Circular, smooth, chalk white, glistening colonies, 2–3 mm)	−	N	+	−	−	N	N	N	N	N	−	N	−	+	−	−	−	+	−	S. capitis (Staph-ZYM code:0041-3)
GP cocci singly and in pairs (Circular, smooth, elevated center, gray 3–4 mm white colonies)	−	N	+	−	−	N	N	N	N	N	+	N	+	+	+	V	+	V	−	S. warneri (Staph-ZYM code:0500-3)
Negative Control (Sterilized 0.9% NaCl)	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−

+=90% or more strains are positive; −=90% or more strains are negative; GN, Gram negative; GP, Gram positive; N, not done; ONPG, ortho-nitrophenyl-3-D-galactopyranoside; V, isolate variability (% not shown); V^a, 63.8%; V^b, 16.6%; V^c, 11.1%.

The API 20 NE system was used for classification of Gram-negative bacilli and evaluated according to the instructions of the manufacturer (BioMérieux, Marcy l'Etoile, France). Strain identification at the species level was accepted for excellent species identification subgroup which gives minimum percent identification of >98.2% and maximum T value of 1.

All registered morphological and biochemical characteristics were compared with Bergey's Manual of Determinative Bacteriology, Ninth Edition (2000).

Semi-quantiative evaluation of bacterial density

The bacterial growth density of each bacterial species within the intestinal contents was estimated based on experience from bacterial diagnostic routine work to be scored semi-quantitatively by estimating the number of colonies detected on the plate as follows: no growth (0: Absent); <10 colonies (1: Very few); 10–50 colonies (2: Few); 50–100 colonies (3: Moderate); 100–200 colonies (4: Rich); >200 colonies (5: Very rich).³⁰

DNA isolation, 16S rDNA amplification

DNA was extracted from 40 different isolates of multi-drug resistant (>3 drugs) Gram-negative bacteria by transferring a loop-full of randomly picked colonies to the DNeasy Blood & Tissue Kit^® (Qiagen S.A., France), according to the manufacturer's guidelines. Following nucleic acid purification, the 5’ portion of the 16S rRNA gene (corresponding to Escherichia coli positions 10 to 806) was amplified using primers V1 [5’-AGA GTT TGA TCA TGG CTC AGA-3’] and V3 [5’-GGT TAC CTT GTT ACG ACT TC-3’]. Cycling conditions: (i) denaturation 3 min at 94°C; (ii) 30 cycles of 30 s at 94°C, 30 s at 56°C, and 2 min at 72°C; (iii) 10 min extension at 72°C. Two μL of DNA extract was used for amplification in a total volume of 25 μL containing 2.5 μL 10X PCR Buffer, 1 μL 50 mM MgCl_2, 1 μL 10 mM dNTP mix, 5 pmol each of forward and reverse primers, 1 U of Taq DNA polymerase (Fermentas, Vilnius, Lithuania) and 16.4 μL sterile nuclease-free PCR grade water. DNA amplifications were performed in a GeneAmp 9700 PCR system thermocycler (Applied Biosystems, Foster City, CA). Positive (Escherichia coli NVH 1067/03) and negative control (sterilized dH₂O) samples were included in all amplifications. The PCR products were analyzed on 1% agarose gels and stained with SYBR Safe^® DNA Gel Stain (Invitrogen). After electrophoresis at 100 V for 60 min, DNA bands were visualized by a Gel Doc^™ XR+ Imaging System (Bio-Rad, Hercules, CA).

Sequence analysis and construction of phylogenetic trees

Amplicons were purified with QIAquick PCR Purification Kit^® (Qiagen) and sequenced by GATC laboratories (Konstanz, Germany) with V1 and V3 primers. Afterwards, 16S rRNA gene sequences were compared with those available in the GenBank, EMBL, and DDBJ databases using a two-step procedure. A first search was performed with the FASTA algorithm of the Wisconsin GCG program package.³¹ All positions showing differences to the best-scoring reference sequence were visually inspected in the electropherogram, and the sequence was corrected manually if necessary. Thereafter, a second search was done using BLASTN. Undetermined nucleotides (designated N) in either the determined sequence or the reference sequence were counted as matches. The mean length of the obtained 16S rRNA sequences was 1150 bp containing 14±9 undetermined (N) positions (range, 5–19 N). A phylogenetic tree was constructed as described by Roblero et al.³² with 16S rRNA gene sequences of representative type strain sequences (Fig. 1).

Antibiotic susceptibility test

Two colonies of each of the identified Gram-negative isolates were randomly selected from each individual fish. Thus a total of 1008 Gram-negative isolates were tested for their phenotypical antibiotic susceptibility to 8 selected antimicrobial agents: Tetracycline (Tet 80 μg), sulphonamide (Sul 240 μg), trimethoprim (Trim 5.2 μg), trimethoprim-sulphonamide (Sul 240 μg+Trim 5.2 μg), streptomycin (ST 100 μg), gentamicin (Gen 240 μg), oxolinic acid (Oxoli 10 μg), and nalidixic acid (Nal 130 μg), using the disc-diffusion method (Neo-Sensitabs, Rosco, Taastrup, Denmark). Inhibition zones were measured and resistance categorization was carried out according to MIC break points of the Norwegian AFA group (2005, 2006), described in “User's guide Neo-sensitabs^®” (http://www.rosco.dk). Intermediate zones were recorded as resistant.

Data analysis

The levels of growth of the bacterial clusters (total and Gram±) were associated with different sampling points. Sexual maturity, gender, and intestinal segment factors were encoded and together with the corresponding levels of the results from cultivable bacterial growth (absent=0, very few=1, few=2, moderate=3, rich=4, and very rich=5) entered into a Microsoft Excel^® spreadsheet. From primary data analysis, a spider web was used to visualize how the facilities with correlation to the chosen factors varied on a ‘Median Counts of Selected Gut Bacteria Growth Degrees' (Fig. 2). Antibiotic sensitivity records for each isolate at the facility level were entered into Microsoft Excel^® spreadsheet. The average distribution of antibiotic resistance frequencies at the facility level was summarized in a histogram (Fig. 3). For further statistical analysis, all data were transferred to Stata (Stata SE/10 for Windows, Stata Corp., College Station, TX). A survey-data-analysis procedure was used for estimating the distribution of the bacterial growth from different sampling points and its univariate association with individual exposure variables, considering each individual animal as primary sampling unit. For each factor, an Ordered linear logistic regression model analysis was built and Odds ratio (OR) with 95% Confidential interval (CI) detected.

FIG. 2.

Phylogenetic neighbor-joining tree of the randomly selected multi-resistant (>3 drug) strains, based on nucleotide sequence of the 16S rDNA genes. Neisseria canis ATCC 14687 type strain was used as an out-group. Nodes with > 95% bootstrap support for all analyses (•) and > 75% bootstrap support (◊). Values below 50% are not shown. Brackets define the gene accession numbers. Lab A–D, randomly picked bacteria 1–2.

FIG. 3.

Growth Degree Median Counts of Selected Gut Bacteria among four facilities. (a) Facility variation. Lab A (blue diamonds) serves as a baseline. Lab B ( filled red squares), Lab C (triangles), and Lab D ( filled purple triangles). (b) Age-related variation. Comparison between sexually mature (red squares) and immature zebrafish (selected as a baseline; blue diamonds). (c) Intestinal segment position. Anterior (selected as a baseline; filled diamonds) and posterior intestinal segments ( filled squares). (d) Gender. Male zebrafish (selected as a baseline; filled diamonds) and female fish (empty squares). Axis scale: Absent=0, very few=1, few=2, moderate=3, rich=4, and very rich=5 (see Materials and Methods section Semi-quantiative evaluation of bacterial density).

Results

All examined fish appeared healthy with no gross deformities or lesions related to the gut at the macroscopic level after visual inspection. Furthermore, by naked eye inspection, there was no discernable difference in the amount of gut content within the guts of the individual fish.

GI microflora

From the intestinal samples taken, a total of 13 morphologically different bacterial species were isolated and characterized: Aeromonas hydrophila, Aeromonas sobria, Vibrio parahaemolyticus, Photobacterium damselae, Pseudomonas aeruginosa, Pseudomonas fluorescens, Pseudomonas luteola, Comamonas testosteroni, Ochrobactrum anthropi, Staphylococcus cohnii, Staphylococcus epidermidis, Staphylococcus capitis, and Staphylococcus warneri. The basic identification criteria employing cultural and biochemical test results are shown in Table 1. 16S rDNA sequences were obtained and analyzed by FASTA (see Methods section) to support the biochemical identifications and a more in depth classification of the drug resistant isolates to reveal potentially antibiotic selected subspecies or clones. A phylogenetic tree constructed with the resulting PCR amplified 16S rRNA gene sequences is presented in Figure 2.

Lab A was selected to represent the baseline, Lab B had slightly but significantly higher levels of total bacterial growth (OR=2.025; 95% CI=1.73–2.38), whereas Lab C (OR=1.008; 95% CI=0.86–1.18) and Lab D (OR=1.122; 95% CI=0.89–1.41) were found to have similar levels as for Lab A (Fig. 3a). The comparisons of adolescent and adult fish revealed that the younger individuals (up to 4 months of age) had a significantly higher level of detected culturable total bacterial growth than mature fish (OR=0.823; 95% CI=0.73–0.93; Fig. 3b). Of the two parts of the intestine, the posterior segment demonstrated a higher content of culturable bacteria than the anterior (OR=4.998; 95% CI=4.38–5.71; Fig. 3c). Comparison of gender revealed no statistically significant microbiota related differences between male and female fish (OR=1.005; 95% CI=0.89–1.13; Fig. 3d).

Fish rearing environment

The records provided from the Lab A–D facilities (Table 2) showed that selected parameters were within acceptable ranges for zebrafish. pH was kept 0.5 units lower in Lab D and the temperature in Lab B was 0.8°C higher, but all values were kept constant and are expected to have no important negative impact on fish health. All facilities stored their dry feed at 4°C except for Lab B in which it was kept at room temperature (20±2°C). Lab B was at the time of the samplings using an old UV lamp (close to exchange time-point) to sterilize the recycling water within the rearing system. Lab A and Lab C both had permanently employed fish nursing technicians, while Lab B and Lab D were running their facilities with the aid of different temporary employed technicians and students, a situation which may have an impact on keeping up with stable conditions of the fish rearing environment.

Table 2.

Records of Facilities

Parameters	Lab A	Lab B	Lab C	Lab D
Establishment year	2005 (1995)¹	2006	2006	1999
Rearing system	Marine Biotech²	Aquatic Habitats³	AquaSchwarz⁴	AquaSchwarz⁴
Fish capacity	10000	4644	25000	20000
Temperature (^oC)	28	28.8	28	28
pH	7.50	7.50	7.54	7.00
Conductivity μS	500	500	494	539
General water quality
NH₃ (mg/L)	0	<0.05	0.2	0.4
NO⁻₂ (mg/L)	0	0.1	0	0
NO⁻₃ (mg/L)	<25	10	6	0
Chlorine (mg/L)	0	NA⁵	NA⁵	0
DO₂ (mg/L)	>7	NA⁵	NA⁵	NA⁵
CO₂ (mg/L)	<15	NA⁵	NA⁵	NA⁵
Feed source
Artemia	SDS-brine shrimp^6,a (artemia)	Silver Star Artemia^7,a	Silver Star Artemia^7,a	Silver Star Artemia^7,a
Dry Feed	SDS 100–400^6,a	ST 1,2,3,4^7,b	ST-3^7,a	ST-5^7,a
Presence of sound isolation	Yes	None	None	None
Photo period D:N	14:10	14:10	14:10	14:10

Facility established in 1995, but upgraded 2005 to AAALAC accreditation standards; ²Marine Biotech, Florida; ³Aquatic Habitats, Florida; ⁴AquaSchwarz, Göttingen; Germany. ⁵NA: not available; ⁶Special diet services, Essex, UK; ⁷AquaSchwarz, Göttingen, Germany. ^aDry feed/artemia kept at +4°C; ^bDry feed kept outside at facility room temperature.

The culturable GI bacteria profile was determined at four check points in the rearing system (Fig. 1). Cultures from tank water (Fig. 1, ‘S tank pre UV’) consisted of a mixture of intestinal and feed related bacterial floras. In general, very mixed bacterial growth was observed from all facilities regardless of fish age. Very high levels of A. hydrophila and high levels P. fluorescens were detected (data not shown) from all facilities. The level of total bacterial growth observed from tank water samples of Lab B (OR=2.576; 95% CI=1.26–5.25) was significantly higher, as compared to the levels of Labs C (OR=0.631; 95% CI=0.26–1.48), D (OR=0.694; 95% CI=0.29–1.64) and the baseline Lab A. Similar significantly higher levels of Gram-positive bacterial growth were measured from pre UV tank water at Lab B (OR=9.060; 95% CI=1.96–41.84) and Lab D (OR=11.043; 95% CI=2.30–52.95). Comparison of the adolescent with adult fish revealed higher levels of total bacterial growth from the tanks with sexually immature fish (OR=0.430; 95% CI=0.25–0.71) relative to the sexually mature fish at all facilities.

Different levels of bacterial species were observed from facility-related samples. A. hydrophila and P. fluorescens were common for all 4 facilities and were detected at relatively high levels from pre-UV light water samples. In addition, a moderate level of P. aeruginosa and low numbers of Staphylococcus spp. were observed at Lab B, C and D. As expected, after the UV light units (Fig. 1, ‘S tank post UV’) a clear decrease in the level of culturable bacteria was observed. In Lab A, very low numbers of A. hydrophila and P. fluorescens were detected, while low numbers of A. hydrophila, very low numbers of A. sobria, and moderate numbers of P. fluorescens and P. aeruginosa were detected at Lab B. In addition, low numbers of Staphylococcus spp. were detected in Lab D.

The culturable flora of dry feed samples (Fig. 1, ‘S feed’) examined mainly consisted of P. luteola, O. anthropi, C. testeroni, and Staphylococcus spp., and unidentified fungi (identification of fungi was out of scope of the study and fungi were detected only from feed samples). For total numbers of culturable bacteria, Lab A had the lowest level with Lab B having the highest and the Labs C and D in-between. On the other hand, Lab D had higher numbers of Gram-positive bacteria. The cultures from the Artemia live feed samples (Fig. 1, ‘S artemia’) consisted of variable levels of A. hydrophila, A. sobria, P. damselae, P. aeruginosa, and P. fluorescens. All facilities had low numbers of P. damselae in the Artemia samples. The D Lab had moderate numbers of Staphylococcus spp. while other facilities had low numbers of Staphylococcus spp. Likewise low numbers of Aeromonas spp. and Pseudomonas spp. were common among the A, C, and D Labs, but were detected at moderate levels at B Lab.

Antibacterial resistance

Of the screened isolates, 40.7% were found to be resistant to more than one antibiotic and 17% showed multiple antibiotic (>3) resistance (Fig. 4). Among the facilities, Lab B showed significantly higher resistance against tetracycline (OR=1.829; 95% CI=1.14–2.92) and sulfonamide (OR=1.514; 95% CI=1.05–2.16). Statistically no significant differences were observed for the sulfonamide-trimethoprim combination for the complete selected bacterial population, whereas Pseudomonas spp. disclosed a higher resistance score in Lab B (OR=1.830; 95% CI=1.07–3.11). All strains were susceptible to gentamicin and only low frequencies of strains were found to be resistant to the quinolone antibiotics.

FIG. 4.

The average frequency distributions of antibiotic resistant Gram-negative isolates (N=1008) among zebrafish facilities. There was a trend of increased antibacterial resistance from bacteria isolated from Lab B, this was statistically significant by logistic regression model analysis (p<0.05, indicated by an asterisk) for some antibiotics.

Discussion

A comparison of the culturable TAB strain zebrafish gut bacterial microflora revealed diversity with varying levels of 13 defined species among the four different zebrafish facilities examined. The phenotypical homology found between the randomly selected groups was illustrated in the 16S rDNA phylogenetic tree. The variation of the composition of the intestinal microbiota of zebrafish between the various Labs underlines the clear impact of the management on the gut microbiota. Significant differences between the observed facilities suggest a correlation to differences in husbandry management. The most frequent strains from the study can be grouped as nonpathogenic/potential pathogenic or commensal bacteria. In particular, three genera (Aeromonas, Vibrio, Photobacterium, and Pseudomonas) of the identified bacterial species in the intestinal micobiota isolates of zebrafish in this study are potential freshwater fish pathogens: A. hydrophila,^33–36 A. sobria,³⁷ V. parahaemolyticus,³⁸ P. damselae,³⁹ P. fluorescens,^40–42 P. luteola,⁴⁴ and P. aeruginosa.^40,42,44 Many of the bacterial species are regular members of the normal microbiota.²⁰ However, these bacteria are also repeatedly reported as causative agents of infections in fish. Many bacterial infections, in particular known from mammals, are caused by bacteria normally inhabiting the various mucosal compartments of the animal.⁴⁵ Infections often develop secondary to factors as virus infections or stress among others.

Most bacteria causing diseases in fish are normal inhabitants of aquatic systems and normally do not cause disease symptoms. Even though pathogenicity and virulence screening were out of the scope of the study, through use of blood agar plates it was possible to observe variable hemolytic activities related to the primary growth of colonies. The GI tract microbiota composition is sensitive for physiological stress of the fish caused by poor system/tank water quality, including high levels of particulates, suboptimal photoperiod, conductivity, oxygen level, temperature, malnutrition, overstocking, recent transportation, handling, or other stress-causing conditions. Stress can affect the immune system of fish,⁴⁶ and change the ultrastructure of the intestinal mucosa.⁴⁷ The disturbance of the intestinal balance may cause establishment of several lethal infections by opportunistic pathogens.⁴⁸ Once an outbreak starts in a zebrafish facility, it often becomes difficult to control, since the rapid growth of bacteria may cause severe imbalance of the microbiota, leading to pathological and lethal effects.

The composition of bacteria isolated from the intestines thus potentially can contribute in describing the relation between the host intestinal immune system and the intestinal microbiota, and indirectly indicate the status of the overall health, growth, development, and performance. The GI microflora composition thus can serve as a measure for susceptibility of disease as an effect of weak secondary infections and/or triggered by environmental stress factors.

Within various groups of species, the microbiota changes more or less with the age of the host.⁴⁹ Humans have a relatively stable intestinal microbiota throughout adulthood,^50,51 although higher numbers of enterobacteria and lower numbers of anaerobic bacteria, including bifidobacteria were reported in elderly people.⁵² Comparing age-dependent effects revealed that sexually immature (juvenile) fish had a higher level of gut microbiota than elderly adult fish. Such an age-related difference in the composition of the intestinal microbiota has, to our knowledge, not previously been demonstrated in fish. The reason for this bias is not known. Lam et al.⁵³ demonstrated that zebrafish have a functionally developed immune system able to produce specific humoral and cellular immune responses from 4 to 6 weeks post fertilization. Zebrafish are considered sexually mature from 15 weeks post fertilization. The impact of age on the intestinal microbiota found in this study may hypothetically be explained by a higher level of sex hormones in adult zebrafish (>6 months) compared to juvenile zebrafish (<4 months). The immune system of the intestine may interact differently with the microbiota in juvenile and mature zebrafish.

The observed higher levels of total bacterial growth of samples from the posterior intestinal segment of zebrafish were in agreement with increased bacterial counts on tryptic soy agar (TSA) that has been seen from fore-gut to the hind-gut of farmed Atlantic salmon (Salmo salar L.).⁵⁴

Our experimental data show no gender differences in zebrafish intestinal microbiota. From a human fecal flora based study, higher levels of aerobic bacteria in males than in females was found.⁵⁵

The focus on gut microbiota diversity in zebrafish is based on the need for a better understanding of how the facility-specific water and diet influence the bacterial composition. The observed different levels of bacterial growth of tank water sample isolates might be related to either insufficient UV light exposure and/or by biofilm formation in aquarium pipes and filters. Lab B reported use of an old UV bulb during the period of samplings which may be related to the observed higher bacterial levels. Biofilms may supply an ideal bacterial community environment for transfer of different antibiotic resistance genes.⁵⁶ A similar tendency of a link between the zebrafish facility lineage and the microbiota composition was recently reported by Roeselers et al.²⁰

Relatively few P. damselae and Aeromonas spp. were found in the artemia feed samples, and the detection of C. testosteroni, O. anthropi, and P. luteola from dry feed underlines the importance of feed hygiene and a safety chain (from origin of the feed, via storage to feeding the fish). Moreover, all detected feed related isolates (except nonidentified fungi that did not grow from fish intestine) were closely linked with the microbiota of the zebrafish intestine. Our finding was in agreement with a report proposing animal feed as being one of the sources of microorganisms to animal gut floras.⁵⁷ Herein, storage and handling with long term and short term storage conditions of animal feeds should be of major concern.

Although no registered antibiotic usage among zebrafish facilities was reported, a variety of resistance patterns was observed in our study. Observed resistance to more than one class of antibiotics has been reported in a number of studies of fish pathogenic and aquaculture environmental bacteria.^58–60 The lowest registered frequency of drug resistance in our study was against gentamicin which is not used among production animals in Norway. Transferable R-plasmids conferring resistance to various antimicrobials including tetracycline, sulfonamide, trimethoprim, and streptomycin associated with aquaculture were reported many years ago.^61,62 Future zebrafish infection model studies should take into account the distribution of bacteria population and drug resistance profiles. In comparison with other aquatic environmental reports,^63–65 sulfonamide/trimethoprim resistance was less frequent in our study. However, high resistance activity in Pseudomonas spp. might be linked with general activated efflux pump systems,⁶⁶ in particular observed in isolates from Lab B. Quinolone resistance (nalidixic acid and oxolinic acid) was low compared to other studies.^59,65,67,68

The GI tract is a favorable ecologic niche for a large number of microorganisms, and as in other animal groups, a wide range of microbes colonize the GI tract of fish.⁴⁵ Information on the type of bacterial composition in the GI tract of fish is controversial ⁶⁹ because of the methods applied. Here we present data from the zebrafish GI tract microbiota using conventional culture-based techniques, ^{6,8,40,70–73} although culture-independent methods are increasingly used for characterization of the GI microbiota of fish.^19,54,74,75 The introduction of gnotobiotic zebrafish¹⁸ offers an in vivo system for dissecting the interplay between the participating bacterial strains of the observed microbiotas and the host. The host factors seem to be similarly important when bacterial isolates establish in the gnotobiotic zebrafish intestine as in the gnotobiotic mouse intestine. By using conventional methods, we have managed to establish a first report on the diversity of culturable bacteria in the zebrafish intestinal microbiota among different research facilities. The observation of different levels of the various groups/species of bacteria and variation in drug resistance profiles between facilities makes a follow-up study based on culture independent methods interesting in order to determine the variations in nonculturable bacteria.

In conclusion, this study suggests that variations in facility management routines affect the numbers of bacteria and the composition of the GI tract microbiota and drug resistance profile. Based on this, we propose that a simplified culturable microbiota protocol could be introduced which would indicate fish health and facility standard in accordance to a sound management practice for zebrafish facilities. More research on the relation between the intestinal microbiota and the health/disease status in zebrafish facilities are necessary to further verify the usefulness of such a protocol.

Footnotes

Acknowledgments

The Norwegian Zebrafish Platform and ZNN have support from the RCN FUGE Program Grant 183344/S10. The authors thank Duncan Colquhoun for his assistance in reviewing the article. The authors also acknowledge the statistical assistance of Camilla Kielland.

LC contributed the ideas for the establishment of the collaborations, design, data collection, performed the sampling and all traditional bacterial culturing, molecular genetic studies, statistical data analysis, and drafting and writing of the manuscript. JR contributed to the establishment of the collaborations, design, data collection, drafting and writing the manuscript. PA contributed to funding, design, acquisition of the fish, and writing the manuscript. HS contributed to acquisition of funds, design, the supervision, drafting and writing the manuscript.

Disclosure Statement

No competing financial interests exist.

References

Metchnikoff

. Surlaflore du corps. Human Member Protocol of Manchester Literature and Philosophical Society, 1901; 45:1–38.

Clapper

, Meade

. Normal flora of the nose, throat, and lower intestine of dogs. J Bacteriol, 1963; 85:643–648.

Zoetendal

, Collier

, Koike

, Mackie

, Gaskins

. Molecular ecological analysis of the gastrointestinal microbiota: A review. J Nutr, 2004; 134:456–472.

Savage

. Microbial ecology of the gastrointestinal tract. Ann Rev Microbiol, 1977; 31:107–133.

Trust

, Sparrow

RAH

. The bacterial flora in the alimentary tract of freshwater salmonid fishes. Canad J Microbiol, 1974; 20:1219–1228.

Horsley

. A review of bacterial flora of teleosts and elasmobranches, including methods for its analysis. J Fish Biol, 1977; 10:529–553.

González

, López-Díaz

JTM

, García-López

, Prieto

, Otero

. Bacterial microflora of wild brown trout (Salmo trutta), wild pike (Esox lucius), and aquacultured rainbow trout (Oncorhynchus mykiss) J Food Prot, 1999; 62:1270–1277.

Sugita

, Shibuya

, Hanada

, Deguchi

. Antibacterial abilities of intestinal microflora of the river fish. Fish Sci, 1997; 63:378–383.

Vergin

, Rappé

, Giovannoni

. Streamlined method to analyze 16S rRNA gene clone libraries. Bio Tech, 2001; 30:938–944.

10.

Dethlefsen

, McFall-Ngai

, Relman

. An ecological and evolutinary perspective on human-microbe mutualism and disease. Nature, 2007; 449:811–818.

11.

Goodman

, Gordon

. Our unindicated coconspirators: Human metabolism from a microbial perspective. Cell Metab, 2010; 12:111–116.

12.

Cheesman

, Guillemin

. We know you are in there: Conversing with the indigenous gut microbiota. Res Microbiol, 2007; 2–9.

13.

Gonzalez

, Stombaugh

, Lozupone

, Turnbaugh

, Gordon

, Knight

. The mind-body-microbial continuum. Dia Clin Neurosci, 2011; 13:55–62.

14.

Bosch

, McFall-Ngai

. Metaorganisms as the new frontier. Zoology, 2011; 114:185–189.

15.

Sukardi

, Ung

, Gong

, Lam

. Incorporating Zebrafish Omics into chemical biology and toxicology. Zebrafish, 2010; 7:41–52.

16.

Lieschke

, Currie

. Animal models of human disease: Zebrafish swim into view. Nature Rev Genet, 2007; 8:353–366

17.

Aleström

, Holter

, Nourizadeh-Lillabadi

. Zebrafish in functional genomics and aquatic biomedicine. Trends Biotechnol, 2006; 24:15–21.

18.

Rawls

, Samuel

, Gordon

. Gnotobiotic zebrafish reveal evolutionary conserved responses to the gut microbiota. Proc Natl Acad Sci, 2004; 101:4596–4601.

19.

Rawls

, Mahowald

, Ley

, Gordon

. Reciprocal gut microbiota transplants from zebrafish and mice to germ-free recipients reveal host habitat selection. Cell, 2006; 127:423–433.

20.

Roeselers

, Mittge

, Stephens

, Parichy

, Cavanaugh

, Guillemin

, Rawls

. Evidence for a core gut microbiota in the zebrafish. ISME J, 2011; 1:14.

21.

Verner-Jeffreys

, Welch

, Schwarz

, Pond

, Woodward

, Haig

, Rimmer

, Roberts

, Morrison

, Baker-Austin

. High prevalence of multidrug-tolerant bacteria and associated antimicrobial resistance genes isolated from ornamental fish and their carriage water. PLoS One, 2009; 21:4:e8388.

22.

Whitehead

. The Design and Analysis of Sequential Clinical Trials. Wiley: Chichester, 1997.

23.

Pocock

. Group sequential methods in the design and analysis of clinical trials. Biometrika, 1977; 64:191–199.

24.

Wallace

, Akhter

, Smith

, Lorent

, Pack

. Intestinal growth and differentiation in zebrafish. Mech Dev, 2005; 122:157–173.

25.

Oehlers

, Flores

, Chen

, Hall

, Crosier

. Topographical distribution of antimicrobial genes in the zebrafish intestine. Develop Comp Immun, 2011; 35:385–391.

26.

Hucker

, Conn

. Methods of Gram Staining. N.Y. State Agr Exp Sta Tech Bull, 1923; 93.

27.

Kloos

, Schleifer

. Bergey's manual of systematic bacteriology. Gram-Positive Cocci. Sneath

PHA

, Mair

, Sharpe

. Willians & Wilkins, 1986; 1015–1022.

28.

MacFaddin

. Biochemical Tests for Identification of Medical Bacteria. Lippincott Williams & Wilkins: Philadelphia, 2000; 912.

29.

Mahon

, Manuselis

. Diagnostic Microbiology. W.B. Saunders Company: Philadelphia, 2000; 1165.

30.

Noakes

, Wallace

, Smith

. Bacterial flora of the uterus of cows after calving on two hygienically contrasting farms. Vet Rec, 1991; 128:440–442.

31.

Devereux

, Haeberli

, Smithies

. A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Res, 1984; 11:387–395.

32.

Roblero

, Posadas

, Serna

JZD

, Garcia

, Rodriguez

. Phylogenetic characterization of bacterial consortia obtained of corroding gas pipelines in Mexico. World J Microbiol Biotechnol, 2008; 24:1775–1784.

33.

Chang

, Thompson

, Rodman

, Bylander

, Thomas

. Pathogenic analysis of Aeromonas hydrophila septicemia. Ann Clin Lab Sci, 1997; 27:254–259.

34.

Rodriguez

, Novoa

, Figueras

. Immune response of zebrafish (Danio rerio) against a newly isolated bacterial pathogen Aeromonas hydrophila. Fish Shellfish Immunol, 2008; 25:239–249.

35.

Pullium

, Dillehay

, Webb

. High mortality in zebrafish (Danio rerio), Contemporary Topics. Lab Animal Sci, 1999; 38:80–83.

36.

Janda

. Recent advances in the study of the taxonomy, pathogenocity and infectious syndromes associated with the genus Aeromonas. Clin Microbiol Rev, 1991; 4:397–410.

37.

Olivier

, Lallier

, Larivière

. A toxigenic profile of Aeromonas hydrophila and Aeromonas sobria isolated from fish. Can J Microbiol, 1981; 27:330–333.

38.

Alcaide

, Amaro

, Todolí

, Oltra

. Isolation and characterization of Vibrio parahaemolyticus causing infection in Iberian tooth carp Aphanius iberus. Dis Aquat Organ, 1999; 35:77–80.

39.

Stephens

, Raidal

, Buller

, Jones

. Infection with Photobacterium damselae subspecies damselae and Vibrio harveyi in snapper, Pagrus auratus with bloat. Aust Vet J, 2006; 84:173–177.

40.

Cahill

. Bacterial of fishes: A review. Microbiol Ecol, 1990; 19:21–41.

41.

Vey

, Boemare

, Vago

. Recherches sur les maladies bacteriennes de l'écrevisse Austropotamobius pallipes Lereboullet. Freshwater Crayfish, 1975; 2:287–297.

42.

Alderman

, Polglase

. Freshwater crayfish–Biology, management and exploitation. Pathogens, Parasites and Commensals. Holdich

, Lowery

. Croom Helm: London, 1988; 167–212.

43.

Altinok

, Balta

, Capkin

, Kayis

. Disease of rainbow trout caused by Pseudomonas luteola. Aquaculture, 2007; 273:393–397.

44.

Colwell

. The bacterial flora of Puget Sound fish. J Appl Bacteriol, 1962; 28:147–158.

45.

Nayak

. Role of gasrointestinal microbiota. Aquaculture, 2010; 41:1553–1573.

46.

Timothy

. Modulation of the immune system of fish by their environment. Fish Shellfish Immunol, 2008; 25:373–383.

47.

Peters

. The effect of stress on the stomach of the European eel, Anguilla anguilla L. J Fish Biol, 1982; 21:497–512.

48.

Sekirov

, Finlay

. The role of the intestinal microbiota in enteric infection. J Physiol, 2009; 587:4159–4167.

49.

Holzapfel

, Haberer

, Snel

, Schillinger

, Huis in't Veld

. Overview of gut flora and probiotics. Int J Food Microbiol, 1998; 41:85–101.

50.

Vanhoutte

, Huys

, De Brandt

, Swings

. Temporal stability analysis of the microbiota in human feces by denaturing gradient gel electrophoresis using universal and group-specific 16S rRNA gene primers. FEMS Microbiol Ecol, 2004; 48:437–446.

51.

Leser

, Molbak

. Better living through microbial action: The benefits of the mammalian gastrointestinal microbiota on the host. Environ Microbiol, 2009; 11:2194–2206.

52.

Mitsuoka

, Hayakawa

. The fecal flora in man. I. Composition of the fecal flora of various age groups. Zentbl Bakteriol Orig, 1973; 223:333–342.

53.

Lam

, Chua

, Gong

, Lam

, Sin

. Development and maturation of the immune system in zebrafish, Danio rerio: A gene expression profiling, in situ hybridization and immunological study. Dev Comp Immunol, 2004; 28; 9–28.

54.

Hovda

, Lunestad

, Fontanillas

, Rosnes

. Molecular characterization of the intestinal microbiota of farmed Atlantic salmon (Salmo salar L.) Aquaculture, 2008; 272:581–588.

55.

Mueller

, Saunier

, Hanisch

, Norin

, Alm

, Midtvedt

, Cresci

. et al. Differences in fecal microbiota in different European study populations in relation to age, gender, and country: A cross-sectional study. Appl Environ Microbiol, 2006; 72:1027–1033.

56.

Stewart

, Costerton

. Antibioitc resistance of biofilms. Lancet J, 2001; 358:135–138,147.

57.

Ruff

. New methods of disease control. Feed Mix, 1992; 1:15–18.

58.

McPhearson

, DePaola

, Zywno

, Motes

, Guarino

. Antibiotic resistance in Gram-negative bacteria from cultured catfish and aquaculture ponds. Aquaculture, 1991; 99:203–211.

59.

Schmidt

, Bruun

, Dalsgaard

, Larsen

. Incidence, distribution, and spread of tetracycline resistance determinants and integron-associated antibiotic resistance genes among motile Aeromonads from a fish-farming environment. Appl Environ Microbiol, 2001; 67:5675–5682.

60.

Hatha

, Vivekanandam

, Joice

, Christol

. Antibiotic resistance pattern of motile aeromonads from farm raised freshwater fish. Int J Food Microbiol, 2005; 98:131–134.

61.

Aoki

, Egusa

, Kimura

, Watanabe

. Detection of R factors in naturally occurring Aeromonas salmonicida strains. Appl Microbiol, 1971; 22:716–717.

62.

Aoki

, Takahashi

. Class D tetracycline resistance determinants of R plasmids from the fish pathogens Aeromonas hydrophila, Edwardsiella tarda, and Pasteurella piscicida. Antimicrob Agents Chemother, 1987; 31:1278–1280.

63.

Schmidt

, Bruun

, Dalsgaard

, Pedersone

, Larsen

. Occurrence of antimicrobial resistance in fish-pathogenic and environmental bacteria associated with four Danish rainbow trout farms. Appl Environ Microbiol, 2000; 66:4908–4915.

64.

Mirand

, Zemelman

. Antimicrobial multiresistance in bacteria isolated from freshwater Chilean salmon farms. Sci Total Environ, 2002; 293:207–218.

65.

Chelossi

, Vezzulli

, Milano

, Branzoni

, Fabiano

, Ricarrdi

Antibiotic resistance of benthic bacteria in fish farm and control sediments of Western Mediterranean. Aquaculture, 2003; 219:83–97.

66.

Poole

. Multidrug efflux pumps and antimicrobial resistance in Pseudomonas aeruginosa and related organisms. J Mol Microbiol Biotechnol, 2001; 3:255–264.

67.

McKoen

, Calabrese

, Bissonnette

. Antibiotic resistant Gram-negative bacteria in rural ground water supplies. Water Res, 1995; 29:1902–1908.

68.

Guardabassi

, Petersen

, Olsen

, Dalsgaard

. Antibiotic resistance in Acinetobacter spp. isolated from sewer receiving waste effluent from hospital and a pharmaceutical plant. Appl Environ Microbiol, 1998; 64:3499–3502.

69.

Izekova

, Lapteva

. Microflora associated with the digestive-transport surfaces of fish and their parasitic cestodes. Russ J Ecol, 2004; 3:176–180.

70.

Onarheim

, Raa

. Characterization and possible biological significance of autochthonous flora in the intestinal mucosa of seawater fish. Microbiology in Poikilotherms. Lese

Elsevier: Amsterdam, The Netherlands, 1990; 197–201.

71.

Zaman

, Leong

. The Proceeding of the Third Asian Fisheries Forum. Intestinal Distribution of a Caryophyllidae Cestode, Lytocestus parvulus, furtado 1963, in Clarias batrachus (L.) from Malaysia. Chou

, Munro

, Lam

, Chen

, Cheong

LKK

, Ding

et al. Asian Fisheries Society: Manila, Philippines, 1994; 314–316.

72.

Ringø

, Strøm

, Tabachek

. Intestinal microflora of Salmonids. A review. Aquaculture Res, 1995; 26:773–789.

73.

O'Hara

, Weinstein

, Miller

. Manual and automated systems for detection and identification of microorganisms. Manual of Clinical Microbiology. Murray

, Baron

, Pfaller

, Jorgensen

, Yolken

. American Society for Microbiology: Washington, D.C., 2003; 185–207.

74.

Sandaa

, Magnesen

, Torkildsen

, Bergh

. Characterisation of bacterial communities associated with early life stages of cultured great scallop (Pecten maximus), using denaturing gradient gel electrophoresis (DGGE) System Appl Microbiol, 2003; 26:302–311.

75.

Ringø

, Sperstad

, Kraugerud

, Krogdahl

. Use of 16S rRNA gene sequencing analysis to characterize culturable intestinal bacteria in Atlantic salmon (Salmo salar) fed diets with cellulose or non-starch polysaccharides from soy. Aquaculture Res, 2008; 39:1087–1100.