Functional Genomic Analysis. RNA isolation. To compare gene expression in the digestive tracts of zebrafish reared under different treatments or at different developmental stages (i.e., CONV versus GF at 6 dpf, CONR versus GF at 6 dpf, 6-dpf versus 10-dpf CONR, and 10-dpf versus 20-dpf CONR), two pools of animals from each group (n = 30 individuals per pool per group) were analyzed. Their digestive tracts (intestine, liver, pancreas, gall bladder, swim bladder) were removed under a dissecting microscope and pooled, and each pooled collection was homogenized by repeated passage through a 20-gauge needle. Total RNA was then extracted (TRIzol reagent; Invitrogen). RNA quantity and quality were assessed by using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies) and Agilent 2100 Bioanalyzer (Agilent Technologies).

Fabrication of zebrafish DNA microarrays. A zebrafish oligonucleotide probeset containing 16,399 65-mer oligonucleotides representing 16,228 gene clusters and 171 control elements was obtained from Sigma Genosys (www.sigma-genosys.com/oligonucleotide_Zebrafish.asp). Oligonucleotides were solubilized (final concentration, 50 µM in 3× SSC/0.75 M betaine), and spotted onto epoxy-coated glass slides (MWG Biotech) with a linear servo arrayer (P. Brown; Stanford University).

Printed slides were placed in a 42ºC oven at 40% relative humidity for 12-14 h. After incubation, slides were cross-linked [UV Stratalinker 2400 (Stratagene); 700 µJ/cm²], submerged and shaken in 0.2% SDS for 2 min, followed by three separate washes in water (1 min per wash at 23ºC). Slides were then placed in a 50ºC water bath for 20 min and spun dry by centrifugation. Processed slides were placed in a light-proof slide box and stored in a dessicant cabinet at room temperature until use.

Synthesis of cDNA probes. First-strand cDNA was generated by reverse transcription utilizing the 3DNA Array 350 kit (Genisphere) and modified oligo(dT) primers containing a 5' fluorophore/dendrimer-specific sequence. The fluorophore-specific oligo(dT) primer (1 µl per reaction) was added to total RNA [4 µg per reaction in experiments comparing CONV versus GF at 6 dpf, and CONR versus GF at 6 dpf; 5 µg per reaction in experiments comparing 6-dpf versus 10-dpf CONR, and 10-dpf versus 20-dpf CONR; in 10 µl of diethyl pyrocarbonate (DEPC)-treated water]. The 11-µl solution was incubated at 80ºC for 10 min and cooled on ice for 2 min, and the following materials were added to a total volume of 20 µl: 4 µl of 5× first-strand buffer (Invitrogen), 1 µl of RNase inhibitor (Genisphere), 1 µl of dNTP mix (10 mM each dATP, dCTP, dGTP, and dTTP), 2 µl of 0.1 M DTT, and 1 µl of Superscript II RNase H-negative reverse transcriptase (10 units/µl; Invitrogen). Reverse transcription was carried out at 42ºC for 2 h and terminated by adding 0.5 M NaOH/50 mM EDTA (65ºC for 10 min), followed by neutralization with 1 M Tris•HCl, pH 7.5.

Hybridizations. To compare mRNA levels, competing samples were paired, then purified and concentrated with Microcon YM-30 microconcentrators (Millipore) according to the manufacturer’s protocol. Each sample pair was then resuspended in hybridization buffer (MWG Biotech). Four hybridizations were carried out in a sequential manner. The primary hybridization was performed by adding 40 µl of sample to the DNA microarray under a supported glass coverslip (Erie Scientific) with subsequent incubation at 42ºC for 16-20 h at high humidity in the dark. Before the secondary hybridization, slides were gently submerged in 2× SSC/0.2% SDS and transferred to 2× SSC and then to 0.2× SSC (12 min per wash; room temperature) with subsequent drying by centrifugation. Secondary hybridization was performed using the complementary capture reagents provided in the 3DNA Array 350 kit (Genisphere). Forty-microliter reaction mixtures were assembled containing 2.5 µl of 3DNA capture reagent with Cy3, 2.5 µl of 3DNA capture reagent with Cy5, 20 µl of MWG hybridization buffer (containing a 1:100 dilution Genisphere anti-fade reagent) plus 15 µl of DEPC-treated water, and then incubated in the dark at 80ºC for 10 min, followed by 15 min at 50ºC. Forty microliters of the treated solution was then added to the slide under a supported glass coverslip and incubated at 50ºC for 3 h at high humidity in the dark. After hybridization, the DNA microarray was washed in 2× SSC/0.2% SDS, followed by 2× SSC and 0.2× SSC (12 min per cycle; room temperature), and then dried by centrifugation.

Data collection and analysis. Slides were scanned immediately on a ScanArray Express HT scanner (Perkin Elmer) to detect Cy3 and Cy5 fluorescence at 543 and 633 nm, respectively. Laser power was kept constant for Cy3/Cy5 scans and photomultiplier tube values were set for each experiment based on background fluorescence. Gridding and analysis of images was performed with ScanArray software (Perkin Elmer). Each spot was defined on a pixel-by-pixel basis, using a modified Mann-Whitney statistical test. Normalization of Cy3/Cy5 channel intensities was performed with a LOWESS procedure. After normalization, expression ratios were calculated for each spot and transformed to log₂ values.

Each experiment consisted of pairwise competitive hybridizations from two treatment groups (CONV versus GF at 6 dpf, CONR versus GF at 6 dpf, 6-dpf CONR versus 10-dpf CONR, or 10-dpf CONR versus 20-dpf CONR), plus reciprocal dye-swap replicates (Fig. 4). Since biological duplicates were generated for each treatment group, a total of four DNA microarrays were utilized per comparison of two treatment groups. Oligonucleotide elements that (i) received "present" calls in all four microarrays and (ii) displayed >1.55 mean signal-to-noise ratio across both dye channels in all four microarray replicates were identified and all others were excluded. The log₂ ratio of median dye intensities for each remaining element was averaged across all four microarrays. To account for measurement variance among replicate microarrays within an experiment, standard deviations of the averaged log₂ ratios of all remaining elements were averaged to identify the standard deviation for the experiment (SDE; ref. 2).

When considering the results of an experiment, we defined differentially expressed genes as those displaying an average log₂ ratio with an absolute value of greater than 2 SDE, providing ~95% confidence (GF versus CONV 1 SDE = 0.501; GF versus CONR 1 SDE = 0.566). Differentially expressed genes identified in this manner are referred to by the names of their putative mouse or human homologs. Homologies were assigned by using the following methods: (i) previous zebrafish gene name assignment, (ii) EST assembly homology (http://zfish.wustl.edu), (iii) UniGene homology (www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=unigene), or (iv) Ensembl gene prediction homology based on corresponding genomic sequence (www.sanger.ac.uk/Projects/D_rerio). Functional classification of genes was based in part on the Gene Ontology Consortium database (www.geneontology.org). For microarray image files, analysis output files, derived measurement files, and MIAME documentation, see http://gordonlab.wustl.edu/.

RNA was harvested from the digestive tracts of 6-dpf zebrafish monoassociated with A. hydrophila or P. aeruginosa (10 individuals per pool, 2 biological duplicate pools per treatment) using methods described above. These RNAs, and the RNAs employed for DNA microarray analyses of GF, CONV, and CONR digestive tracts, were used as templates for generating cDNAs (Superscript II reverse transcriptase; random primers; Invitrogen). qRT-PCR assays were performed as described (3), except that each 25-µl reaction contained cDNA corresponding to 1 ng of total RNA, and either 900 nM gene-specific primers or 300 nM 18S rRNA-specific control primers (see Table 6 for a list of gene-specific primers). Assays were performed in triplicate with an ABI Prism 7700 Sequence Detector (Applied Biosystems). Data were normalized to 18S rRNA (D D C_T analysis).

qRT-PCR was also used to verify changes in gene expression observed in DNA microarray assays of CONV versus GF, and CONR versus GF 6-dpf digestive tract RNAs (13/14 and 9/10 genes confirmed, respectively; Table 6).

Digestive tracts were harvested from CONR zebrafish under a dissecting microscope and were pooled (two pools of 25 and 13 animals each at 6 dpf; two pools of 17 and 28 animals each at 10 dpf, two pools of 12 and 24 animals each at 20 dpf, two pools of 11 and 12 animals each at 30 dpf, and two pools of 2 and 3 animals each at 1 year post fertilization; sexes mixed at all time points). Digestive tracts were also recovered from 6-dpf CONV animals (13 animals pooled).

Pooled tissues from each time point were homogenized by repeated passage through a 20-gauge needle, and genomic DNA was extracted using the QIAmp DNA Stool Mini Kit (Qiagen). To amplify bacterial 16S rDNA genes from genomic DNA, PCRs were prepared containing 0.05 unit/μl Taq DNA polymerase (Invitrogen), 1× PCR buffer minus MgCl₂ (Invitrogen), 3 mM MgCl₂, 200 µM each deoxynucleoside triphosphate, 100 ng/μl purified bovine serum albumin (New England Biolabs), 3 ng/μl DNA template, and 400 nM each bacterial 16S rDNA gene primer [forward: 5'-AGAGTTTGATYMTGGCTCAG-3' (Escherichia coli 16S rDNA positions 8 to 27; refs. 4 and 5); reverse: 5'-ACGGYTACCTTGTTACGACTT-3' (E. coli 16S rDNA positions 1492 to 1512)]. PCR conditions were as follows: 10 min at 94ºC, followed by 20 cycles of 60 sec at 94ºC, 60 sec at 52ºC, and 105 sec at 72ºC, with a post-amplification extension of 15 min at 72ºC. PCR products were fractionated by agarose gel electrophoresis, and 1.6-kb 16S rDNA amplicons were recovered (QIAquick Gel Extraction Kit; Qiagen) and cloned into pCR II TOPO (TOPO TA Cloning Dual Promoter Kit; Invitrogen). Plasmid inserts were sequenced in BigDye dye terminator reactions (Perkin Elmer).

Sequences shorter than 450 bp (Phred quality value ≥ Q20) were excluded. Potential chimeric sequences were identified by using the program Bellerophon (http://foo.maths.uq.edu.au/~huber/bellerophon.pl; ref. 6) and also excluded. The remaining sequences were compared to the Ribosomal Database Project II (RDP; http://rdp.cme.msu.edu; ref. 7) by using BLAST (8). Sequences displaying ≥98% identity with their closest relative in the RDP were considered to be from the same species (9, 10). All 16S rDNA sequences have been submitted to GenBank under accession numbers AY536921-AY538099. For detailed 16S rDNA homology information, see http://gordonlab.wustl.edu/.

GF, CONV, CONR, and monoassociated larvae were killed at various developmental stages, fixed in Bouin’s solution overnight at 4ºC, and imbedded in paraffin, and serial 5-µm-thick sections were cut perpendicular to the cephalocaudal axis. Sections from segments 1-3 were stained with hematoxylin and eosin to characterize cellular morphology and tissue architecture.

To quantify cellular proliferation, 5-dpf GF, CONR, and CONV zebrafish were immersed in a solution of 5-bromo-2'-deoxyuridine (BrdUrd; 160 µg/ml of filter-sterilized GZM) and 5-fluoro-2'-deoxyuridine (16 µg/ml of GZM) for 24 h before killing. Animals were fixed in Bouin’s solution, then imbedded in paraffin and sectioned as described above. Sections were deparaffinized and stained with goat anti-BrdUrd [final dilution = 1:400 in blocking buffer (PBS/1% BSA/0.3% Triton X-100) (11)], Alexa Fluor 594-conjugated donkey anti-goat Ig (1:400, Molecular Probes), and bisbenzimide (100 ng/ml of PBS; Sigma).

To quantify cell death, 6-dpf animals were killed, fixed overnight at 4° C in 3.7% formaldehyde (prepared in PBS), and then imbedded in paraffin. Sections were prepared as above, deparaffinized, and stained (TUNEL In Situ Cell Death Detection Kit; Roche). The number of BrdUrd- or TUNEL-positive nuclei was divided by the total number of nuclei in a given section. Using bright-field microscopy as a guide, cells in the intestinal epithelium were scored separately from cells in the underlying mesenchyme and muscle layers (n = 19-31 sections per animal for BrdUrd experiments; 11-23 sections per animal for the TUNEL experiment; ≥6 animals per treatment group; all treatments performed on at least two separate occasions). Tests for significance were carried out with unpaired Student’s t tests.

Animals were killed at 6 dpf, fixed in 2.5% glutaraldehyde in 0.1 M PBS, pH 7.4 (1 h at 25ºC), followed by 1% osmium tetroxide in 0.1 M PBS, pH 7.4 (1 h at 25ºC) and 1% uranyl acetate (1 h at 25ºC). Zebrafish were then dehydrated in ethanol and propylene oxide and embedded in resin (EMbed, Electron Microscopy Sciences). Seventy-five nanometer-thick sections were prepared and viewed with a JEOL Model 100C electron microscope.