• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of plosonePLoS OneView this ArticleSubmit to PLoSGet E-mail AlertsContact UsPublic Library of Science (PLoS)
PLoS One. 2010; 5(8): e12459.
Published online Aug 27, 2010. doi:  10.1371/journal.pone.0012459
PMCID: PMC2929194

Gut Microbial Gene Expression in Mother-Fed and Formula-Fed Piglets

Stefan Bereswill, Editor

Abstract

Background

Effects of diet on the structure and function of gut microbial communities in newborn infants are poorly understood. High-resolution molecular studies are needed to definitively ascertain whether gut microbial communities are distinct in milk-fed and formula-fed infants.

Methodology/Principal Findings

Pyrosequencing-based whole transcriptome shotgun sequencing (RNA-seq) was used to evaluate community wide gut microbial gene expression in 21 day old neonatal piglets fed either with sow's milk (mother fed, MF; n = 4) or with artificial formula (formula fed, FF; n = 4). Microbial DNA and RNA were harvested from cecal contents for each animal. cDNA libraries and 16S rDNA amplicons were sequenced on the Roche 454 GS-FLX Titanium system. Communities were similar at the level of phylum but were dissimilar at the level of genus; Prevotella was the dominant genus within MF samples and Bacteroides was most abundant within FF samples. Screened cDNA sequences were assigned functional annotations by the MG-RAST annotation pipeline and based upon best-BLASTX-hits to the NCBI COG database. Patterns of gene expression were very similar in MF and FF animals. All samples were enriched with transcripts encoding enzymes for carbohydrate and protein metabolism, as well as proteins involved in stress response, binding to host epithelium, and lipopolysaccharide metabolism. Carbohydrate utilization transcripts were generally similar in both groups. The abundance of enzymes involved in several pathways related to amino acid metabolism (e.g., arginine metabolism) and oxidative stress response differed in MF and FF animals.

Conclusions/Significance

Abundant transcripts identified in this study likely contribute to a core microbial metatranscriptome in the distal intestine. Although microbial community gene expression was generally similar in the cecal contents of MF and FF neonatal piglets, several differentially abundant gene clusters were identified. Further investigations of gut microbial gene expression will contribute to a better understanding of normal and abnormal enteric microbiology in animals and humans.

Introduction

The principles that govern early microbial colonization in the mammalian intestine are poorly understood. Environmental exposures such as early infant diet are believed to impact the development and function of gut microbial consortia [1]. However, attempts to identify similarities and differences among gut microbes from breast-fed and formula-fed infants have generated conflicting and controversial results. For decades, culture-based studies of enteric microbiology supported the notion that the benefits of breast milk derive partly from the growth of beneficial probiotic bacteria in milk fed infants [1], [2]. Recently, this idea has been called into question by studies that used culture-independent methodologies [3][6]. Nonetheless, it is well accepted that some microbe-mediated diseases including neonatal necrotizing enterocolitis (NEC) are less common in mother-fed (MF) infants than in formula-fed (FF) infants [7], [8]. For this reason, high-resolution studies are needed to better characterize both the fine-scale structure and function of gut microbial communities in MF and FF neonates.

Cultivation-independent techniques have made it possible to identify many or most of the gut microbes present within biological specimens such as fecal samples or gut mucosal biopsies [9], [10]. However, partly due to the complexity of gut microbial consortia, little is known about which microbial genes and proteins are expressed in vivo within the gastrointestinal tract. To date, functional analysis of gut microbes has largely been limited to predictions of functional potential based upon DNA sequence annotations, e.g. with metagenomic analyses [9], [11]. While a handful of studies have used unbiased shotgun approaches to measure gut microbial proteins [12], [13] and metabolites [14], [15], only two recent studies [16], [17] have begun the essential work of characterizing community wide gene expression (the metatranscriptome) in the GI tract, as has been accomplished in studies of marine [18][21] and soil microbes [22], [23]. High-throughput investigations of gene expression make it possible to study how the phenotypes of natural microbial communities change over time in response to environmental exposures.

RNA-sequencing (RNA-seq) [24], [25] is a promising experimental approach to profiling gene expression with high-throughput sequencing technologies. Also referred to as whole transcriptome shotgun sequencing, this technique involves directly sequencing cDNA libraries to learn about the RNA content within a biologic sample. In contrast to most array-based techniques, RNA-seq allows for the characterization of both known and unknown gene transcripts. Furthermore, data output in the form of discrete nucleotide sequences rather than hybridization signals makes it possible to compare results from different laboratories and to study fine-scale variations in transcript sequences. Originally described in studies of eukaryotic gene expression, RNA-seq has now been used to profile gene expression in microbial isolates [26] and also within natural mixed population communities of the ocean and soil [18][23]. This approach holds particular promise for studying the function of complex gut microbial communities without knowing a priori which organisms are present.

Here, an unbiased RNA-Seq approach utilizing massively parallel pyrosequencing was used to study microbial gene expression in cecal contents from 21-day-old mother-fed (MF) and formula-fed (FF) neonatal piglets. Because nutrient availability is tightly linked to global transcriptional control and upregulation of bacterial virulence programs [27], [28], we hypothesized that patterns of gene expression would be distinct in MF and FF animals. To test this hypothesis, four piglets receiving sow's milk and four piglets receiving an artificial piglet formula were sacrificed at 21 days of life. Both DNA and RNA from the cecal contents of each animal were harvested and cryopreserved at the time of sacrifice. 16S ribosomal RNA gene sequences within the DNA libraries were amplified, sequenced, and characterized. In parallel, we isolated RNA from each sample and constructed corresponding whole genome cDNA libraries. The cDNA libraries were subsequently pyrosequenced and, where possible, annotated. Differentially abundant features between the MF and FF animals were identified with Metastats, a recently developed statistical method for comparing high-throughput sequencing data in discrete treatment populations on the basis of count data [29].

Results

DNA and cDNA-based Community Profiling

Two parallel approaches were used to identify the microbial organisms present within cecal contents from the 8 piglets studied. We amplified and sequenced 16S ribosomal DNA sequences from the DNA samples, and we also characterized unamplified fragments of 16S rRNA sequences present within the whole genome cDNA libraries (Table 1).

Table 1
Sequencing statistics of DNA and cDNA libraries from mother-fed and formula-fed piglets.

Analysis of 16S rDNA amplicons demonstrated the presence of microbes from 11 total phyla. However, 98% of 16S rDNA amplicons classified at the level of phylum were derived from the phyla Bacteroidetes (56.8% of total), Firmicutes (36.6%), and Proteobacteria (4.8%). At the level of phylum, no taxa were differentially abundant within either the MF or FF group. However, several genera were differentially abundant in one of the two animal groups (Figure 1a and Table S1). The MF data sets were enriched (p<0.01) with sequences attributed to the genera Prevotella, Oscillibacter, and Clostridium; the FF data sets were enriched (p<0.01) with sequences from Bacteroides, Parabacteroides, and Alistipes. Prevotella represented the most abundant genus in MF animals but was nearly absent in FF animals. Conversely, Bacteroides was the most abundant genus in FF animals but was largely absent in MF animals. The genus Lactobacillus was relatively enriched within MF samples (MF mean relative abundance 0.09 vs FF mean relative abundance 0.02), although this discrepancy did not reach statistical significance (p = 0.13). Bifidobacterium sequences were not observed in any sample. Principal component analysis of this data set (Figure S1) demonstrated that samples from the MF and FF groups cluster separately at the level of genus.

Figure 1
Microbial community structure in cecal contents from mother-fed and formula-fed piglets at 21 days of life.

The sequenced cDNA libraries (after one cycle of mRNA enrichment) contained 80.6% ribosomal RNA sequences, and taxonomic assignments were successfully made for 4.0% of these sequences (25,398 total sequences) using the Ribosomal Database Project classification algorithm [30] (Figure 1B). Assessment of microbial community structure with this approach was quite similar, particularly for abundant organisms, to the assessments made by analyzing PCR amplicons of 16S rDNA sequences. Both analyses revealed a similar distribution of sequences from the three major phyla; further, both analyses demonstrate the enrichment of Prevotella, Oscillibacter, and Clostridia sequences in the MF samples and the enrichment of Bacteroides sequences in the FF samples. The relative agreement of results in Figure 1A and Figure 1B indicate that RNA-seq of total RNA isolated from mixed population consortia is an accurate means of determining community structure in the GI tract; as a prior study also indicated (23), analysis of 16S rDNA amplicons may not be required in this setting.

A core microbial metatranscriptome in the piglet cecum

While increasing attention has been paid recently to “core” indispensable genes and organisms that are present within mammalian gut microbial communities [9], [31], [32], little is known about how the expression of these genes varies in response to environmental conditions. To address this issue, barcoded cDNA libraries were constructed from microbial RNA extracted from the cecal contents of all 8 animals after performing one cycle of mRNA enrichment. Pyrosequencing of these libraries yielded 767,424 total reads with an average read length of 381 bp (281.3 Mb total sequence) (Table 1). Automated removal of ribosomal RNA sequences yielded 98,713 and 40,401 non-ribosomal RNA sequences from MF and FF animals, respectively (51.7 Mb total). Screened cDNA sequences were assigned functional annotations by MG-RAST [33] and additionally based upon best-BLASTX-hits to the NCBI COG database [34] (e-values<1e−5 for both methods of annotation). Taxonomic assignments for cDNA sequences were also made where possible.

Consistent with other studies of gene expression within marine and soil microbial consortia [18], [20][23], most transcripts could not be annotated. 31% of non-ribosomal RNA sequences could be matched to proteins in the SEED Subsystems database, and 30% of sequences were matched to the COG database (e-value cutoff of 1e−5 for both analyses) (Table 1). Similar to published metagenomic and metatranscriptomic studies of terrestrial and environmental microbial communities, over 30% of annotated transcripts were associated with either carbohydrate or protein metabolism (Figure 2a and Table 2). Sequences linked to amino acid metabolism, stress response, respiration, virulence, and cell wall metabolism were also well represented. Figure 2 demonstrates that the mean abundance of transcripts within MG-RAST subsystems was remarkably constant across all 8 animals; for all subsystems, the average standard deviation of the mean relative abundance from all animals was 1.0%. Projection of the global metabolic profiles onto the KEGG pathways using the iPath tool [35] further demonstrated the overall similarity of the MF and FF the gut microbial communities (Figure 2b).

Figure 2
A core microbial metatranscriptome in the piglet cecum.
Table 2
Annotations of highly expressed microbial genes in the piglet cecum.

Carbohydrate utilization profiles (Figure 3a) were dominated by transcripts linked to central carbohydrate and glycogen metabolism subsystems. The single most abundant COG hit was to glyceraldehyde-3-phosphate dehydrogenase (GAPDH; COG0057); other common transcripts encoded pyruvate phosphate dikinase (COG0574), phosphoglycerate kinase (COG0126), and glycogen synthase (COG0297) (Table 2). Sequences linked to utilization of mono- and oligosaccharides were also abundant, and were assigned to SEED subsystems for metabolism of sucrose, maltose, lactose, fructooligosaccharides, and sugar alcohols (hexitol). These sequences included transcripts for mannose-6-phosphate isomerase (COG0662), phosphomannomutase (COG1109), beta-galactosidase (COG3250; lactose utilization), and tagatose-bisphosphate aldolase (COG0191; galactose utilization). Sequences associated with ABC-type sugar transporters (COG1653, COG1879, COG3839, COG0395, COG1175, COG1129) and sugar-specific phosphotransferases (COG3444, COG3716, COG2893, COG1263) were commonly observed.

Figure 3
Universal expression of carbohydrate utilization and microbial virulence genes in the piglet cecum.

Transcripts associated with virulence, stress response, and cell wall metabolism were also relatively abundant across all animals (mean relative abundances 0.055±0.007, 0.047±0.01, and 0.044±0.01, respectively). The most abundant sequences classified within the SEED subsystem of virulence were associated with Ton and Tol transport systems, fluoroquinolone resistance, and other iron transport systems (Figure 3b). The third most commonly observed COG overall was rubrerythrin (COG1592), a protein that provides protection from oxidative stress via catalytic reduction of intracellular hydrogen peroxide [36]. Other oxidative stress response genes, such as peroxiredoxin (COG0450) and superoxide dismutase (COG0605), were also highly expressed. Abundant transcripts in the cell wall and capsule subsystem included numerous genes regulating sialic acid metabolism (COG4409, COG0363, COG0329, COG0774, COG1820, COG2942) and genes mediating utilization of plant-based polysaccharides via the cellulosome (e.g. SusC, which encodes an outer membrane protein involved in starch binding in Bacteroides species [37]).

Several of the commonly expressed microbial genes in the piglet cecum have crossover functions linking their primary function with binding to host epithelium, lipopolysaccharide (LPS) metabolism, oxidative stress response, and virulence. GAPDH, beta-galactosidase, enolase (COG0148; central carbohydrate metabolism), phosphoglycerate mutase (COG0588; central carbohydrate metabolism) and elongation factor Tu (COG 0050) have all been noted to mediate binding to host mucins, fibrinogen, and/or plasminogen [38][43]. Interestingly, it has been demonstrated that in vitro secretion of GAPDH by E coli varies according to the composition of nutrients in the growth medium [44]. Mannose-6-phosphate isomerase [45] and glutamate dehydrogenase [46] (GDH; COG0334) each have been shown to contribute to microbial survival in the face of oxidative stress. Mannose-6-phosphate isomerase, periplasmic ABC transporters, and phosphomannomutase are involved with processing of extracapsular polysaccharides such as LPS [45], [47]. Another commonly observed transcript, phosphoenolpyruvate phosphotransferase (COG1080) was recently demonstrated to jointly control carbohydrate utilization, biofilm formation, and intestinal colonization in a Vibrio cholerae model [48]. Finally, decreased pathogenicity has been observed in models with mutagenesis or inhibition of GAPDH and beta galactosidase [38], [49]. Because these genes were highly expressed in all samples, it is likely that these genes represent core features of microbial communities in the piglet cecum.

Taxonomic assignments for cDNA sequences

An advantage of digital studies of the community transcriptome is that the data can inform explanations of how metabolic tasks are divided among the members of gut microbial communities. Little is currently known about whether fundamental processes such as carbohydrate utilization are performed by a range of organisms in vivo or rather by a limited group of specialized community members. Furthermore, although it is widely recognized that numerous low abundance bacterial species reside in the GI tract, little is known about the contribution of such organisms to community metabolism and function. To begin to address these questions, taxonomic assignments were made, where possible, for each gene transcript by identifying the best BLASTN hit against an in-house database of 1054 finished microbial genomes (Table S2). In both MF and FF groups, non-ribosomal transcripts from the phylum Bacteroidetes were most abundant (Figure 4a). In general, the abundance of transcripts from a given phylum was similar to the population wide abundance of that phylum as determined by the 16S- based methods (Figure 1). In Figure 4b, the two most abundant general COG and two most abundant individual COG hits are listed along with the taxa linked to the transcripts.

Figure 4
Relationship between gut microbial community structure and function.

These data highlight the importance of available reference microbial genomes when attempting to profile mixed population microbial communities without knowing in advance which organisms are present. The taxonomic origin could be assigned for only 27.4% of all non-ribosomal sequences. In the MF group, which was dominated by organisms from the genus Prevotella, only 16% of sequences could be linked to a taxonomic origin (Table 1). Most likely, the gene content of the strains present in these samples differs substantially from the three Prevotella genomes present in our database (P. intermedia, P. ruminocola, and P. melaninogenica). During our initial attempts to make taxonomic assignments for cDNA sequences, we used a database that did not contain any Prevotella genomes. By adding Prevotella genomes, we modestly increased the relative proportion of sequences that could be assigned to a genus (data not shown). The abundance of annotated transcripts from the phylum Bacteroidetes was significantly higher in the FF group. This may represent a higher level of transcriptional activity in FF animals from species in this phylum, as recently reported by Turnbaugh et al. [17]. Alternatively, however, this result may indicate a closer alignment between the genomes of Bacteroides species in our database (B. fragilis NCTC 9343, B. fragilis YCH46, and B. thetaiotamicron) and the Bacteroides species within the cecal contents of the animals studied.

Differential gene expression in MF and FF piglets

Gene expression among gut microbes was relatively similar in MF and FF animals, despite taxonomic differences at the level of genus in the observed microbial communities. MG-RAST generated assignments of protein-coding nucleotide sequences to 29 broad functional categories of SEED subsystems and 637 narrow SEED subsystems (Level 1 and Level 3, respectively). In 24 of the 29 major subsystems, the abundance of assigned sequences was not statistically different in the MF and FF groups. Sequences assigned to the Level 1 subsystem of carbohydrates were similarly abundant in the two groups (p = 0.129). Remarkably, the abundance of MF and FF sequences was not significantly different in 94% of Level 3 Subsystems (600 of 637).

Significant differences were observed in five Level 1 SEED subsystems: prophage (p = 0.005), amino acids and derivatives (p = 0.009), potassium metabolism (p = 0.017), respiration (p = 0.018), and motility and chemotaxis (p = 0.04) subsystems. Table 3 and Table S3 list hits to Level 1 and Level 3 SEED Subsystems that were differentially abundant in either the MF group or the FF group; differentially abundant COGs are listed in Table S4. Only two carbohydrate utilization SEED subsystems were differentially represented within the two data sets. Transcripts for utilization of L-arabinose (p = 0.013), a sugar known to accumulate in the distal intestine of FF pigs due to the hydrolysis of plant-based non-starch polysaccharide additives [50], and for utilization of the sugar alcohol mannitol (p = 0.008) were each observed more commonly in the FF group. Two individual COGs, galactose mutarotase (COG2017; p = 0.011) and transketolase (COG0021; p = 0.006) were observed more frequently in the MF group. Galactose mutarotase is an essential enzyme in the metabolism of lactose [51], and transketolase is an enzyme involved in central carbohydrate metabolism that has been shown to link carbon availability with phage resistance, LPS metabolism, and flagellar motor function [52], [53]. Interestingly, in a recent in vitro study of Bifidobacterium gene expression, transketolase expression was [54] upregulated after exposure to breast milk.

Table 3
Differential abundance of gut microbial transcripts assigned to SEED Subsystems.

Significant differences between the MF and FF groups were evident in the abundance of transcripts related to amino acid metabolism. The MF group was markedly enriched in sequences encoding for enzymes that contribute to arginine metabolism, e.g. arginine deaminase (COG2235; p = 0.001) and ornithine aminotransferase (COG4992; p = 0.003). In neonates, arginine is synthesized in the gut from proline [55]; we found that transcripts for proline biosynthesis were equally abundant in the two groups (p = 0.56) but transcripts for proline and 4-hydroxyproline utilization were more abundant in the MF group (p = 0.007). The MF group was also enriched in sequences associated with glycine cleavage (p = 0.02), alanine biosynthesis (p = 0.02), and histidine degradation (p = 0.03), whereas the FF group was enriched with sequences associated with glycine and serine utilization (p = 0.016) and aromatic amino acid degradation (p = 0.02).

The MF data set was enriched with sequences assigned to the oxidative stress subsystem (p = 0.029). This may be consistent with long-held claims of the antioxidant properties of maternal milk in human neonates [56], [57]. Additional sequences enriched within the MF data set included transcripts encoding bacterial chemotaxis proteins CheA, CheV, and CheY (p = 0.005), as well as enzymes that contribute to luminal methanogenesis (p = 0.017). In the FF group, transcripts encoding Ton and Tol type receptors, which are important for iron-mediated bacterial virulence in the gut [58], [59], were significantly more abundant than in the MF group (p = 0.020).

Discussion

It is now well established that gut microbes contribute to mammalian physiology [60], [61]. Pioneering research has clarified the roles of microbes in the gut in normal development and homeostasis [62], and a new generation of investigations seeks to clarify the role of microbes in disease pathogenesis [63][65]. Although a subset of diseases can be linked to the presence of a single causative pathogen, there is a growing recognition of the need to study complete or near complete microbial communities [66], [67]. Prior to the advent of next-generation technologies in nucleotide sequencing and mass spectrometry, molecular studies of entire microbial consortia (rather than selected organisms of interest) were not technically feasible. Although bold advances have been made in understanding the metabolic potential of intestinal microbes based upon a massive amount of DNA sequencing of bacterial isolates and in metagenomic studies [9], [11], few functional studies have been performed characterizing community wide gene and protein expression in the gut.

Here we have presented a metatranscriptomic evaluation of intestinal bacteria in 8 neonatal piglets. In this study, we used an RNA pyrosequencing platform to profile gut microbial gene expression in MF and FF piglets without prior knowledge of which organisms were present. To our knowledge, this study represents the largest number of independent samples (8 subjects) used to date for analysis of community wide gene expression in the gut. Several methodologic considerations of this study warrant mention. First, we performed a single step for mRNA enrichment prior to construction of cDNA libraries, and cDNA sequencing results indicated that the degree of enrichment was modest (19.7% of all RNA sequences were non-ribosomal). In the future, this step in sample processing may not be necessary given the volume and length of sequencing reads available with current sequencing platforms. Second, amplification of cDNA sequences was not necessary in this study, although it has been required in prior metatranscriptomic studies due to low yields of RNA [18], [21]. In the piglet, cecal biomass is plentiful and samples can be immediately cryopreserved after collection without exposure to ambient oxygen. As these techniques are inevitably extended to studies of human fecal samples, the need for signal amplification will need to be re-evaluated depending upon RNA yield. Third, we studied the structure of the microbial communities both by sequencing 16S rDNA amplicons and by characterizing unamplified 16S rRNA sequences within the cDNA libraries. In agreement with results from Urich et al. [23], our analyses indicate that taxonomic assignments based upon DNA 16S amplicons correspond well with assignments derived from ribosomal sequences within cDNA libraries. Fourth, gene transcripts were studied without the presence of corresponding metagenomic data sets. The importance of studying paired mRNA and DNA data sets has not yet been clarified and past work in the field has, in fact, demonstrated low levels of homology between metatranscriptomes and metagenomic scaffolds. However, inclusion of whole genome libraries would have likely partially mitigated the large number of unannotated transcripts that we observed in the current study.

A subset of specific transcripts was relatively abundant in all samples studied, indicating that we have begun to define a core neonatal gut microbial metatranscriptome. As expected, a preponderance of mRNA transcripts corresponded to genes related to metabolism of carbohydrates and proteins. These results align well with recent papers that have defined the nature of carbohydrate utilization genes within a core metagenome (9,31,32). Additionally, commonly observed gut microbial transcripts in our study encode for proteins that enable binding to host epithelium, regulate processing of extracellular polysaccharides, and mediate microbial stress response. The validity of our results is supported by a recent proteomic-based study that characterized circulating antibodies against gut bacteria in human subjects [68]. The most commonly observed antibodies were active against GroEL, enolase, and elongation factor Tu, which were each observed frequently in our gene expression study. The significance of how and why antibodies are formed against gut bacterial antigens requires further study.

A primary goal of this study was to identify differences in the gut microbial communities of MF and FF neonatal piglets. MF samples were enriched with 16S sequences from the taxa Prevotella, Oscillibacter, and Clostidium, whereas the FF samples were enriched with sequences from Bacteroides, Parabacteroides, and Alistipes. Because the animals in the MF and FF piglets did not originate in the same litter of piglets, it is possible that observed differences between the two groups reflected a litter effect. However, studies from our laboratories have shown that litter effects on the gut flora of piglets are reproducibly overshadowed by diet effects (Wang et al., submitted).

Profiles of gene expression were similar. The abundance of sequences from more than 90% of SEED subsystems and COG clusters did not differ between MF and FF datasets was not statistically different. These results suggest that sow's milk and artificial formula, although chemically distinct, induce relatively subtle changes in the gut microbial gene expression. However, several important differences were noted in the transcriptomes of the MF and FF animals. Marked differences were noted in the expression of genes involved in amino acid metabolism. Interestingly, we observed a clear abundance of enzymes linked to arginine metabolism in the MF group of animals. This finding may have clinical relevance because several reports have indicated that an abnormally low serum concentration of arginine, a precursor for nitric oxide production, confers an increased risk for NEC [69], [70], and it is widely accepted that the incidence of NEC is lower in low birthweight infants receiving breast milk [7], [8]. Additionally, the MF data sets were significantly enriched in oxidative stress response genes, which is interesting in light of the oft-stated belief that breast milk provides antioxidant protection for newborns. By contrast, the FF data sets were significantly enriched in sequences encoding the Ton and Tol transport proteins, which have been associated with iron-mediated microbial virulence in animal models of infection [58], [59].

Recent advances in completing microbial genomes and completing metagenomic surveys have demonstrated the vast metabolic potential of the bacteria and archaea present within the mammalian intestine. As DNA-based studies continue, parallel studies of gut microbial function will be essential. Functional studies of gene expression, protein expression, and metabolite production will make it possible to define what is “normal” in the field of enteric microbiology. Eventually, continued progress in this area will allow us to better understand the contributions of microbes to diseases such as NEC, Crohn's disease, and obesity.

Materials and Methods

Experimental animals

Animals were managed throughout the study in accordance with requirements of the Institutional Animal Care and Use Committee at the University of Illinois (IACUC protocols 08070 and 08015), in accordance with approved NIH guidelines. Vaginally-delivered piglets were allowed to suckle for 48 h postnatal to obtain colostrum. Four piglets (mother fed, MF) from a single litter remained with the sow throughout the study. Four additional piglets (formula fed, FF), from two separate litters, were fed with formula for the remainder of the study. FF piglets were transported to the animal facilities and housed individually in metabolism cages with 12 h light/dark cycle as previously described [71]. Room temperature was maintained at 25°C with supplemental heat was provided through of radiant heaters suspended above the cages to maintain a local temperature between 30 to 32°C. The FF piglets were fed with a cow milk-based sow milk replacer (Advance Baby Pig LiquiWean; Milk Specialties, Dundee, IL). Formula was prepared fresh each morning at a concentration of 18.3% solids and was offered 14-times daily via a pump into a bowl at a rate of 360 mL/kg body weight. Piglets were weighed each morning and monitored three times per day for general health.

Sample Collection

Samples were collected on postnatal day 21. Piglets were first sedated with an intramuscular injection of Telazol (7 mg/kg body weight; Fort Dodge Animal Health, Fort Dodge, IA) and then euthanized by intracardiac administration of sodium pentobarbital (Fatal Plus: 72 mg/kg body weight; Vortech Pharmaceuticals, Dearborn, MI). The large intestine was excised and separated into cecum and colon at the cecocolic junction. Tissues and luminal contents from cecum were immediately collected into sterile tubes and snap frozen in liquid nitrogen. The samples were stored at −80°C until processing.

PCR amplification and analysis of 16S rDNA sequences

DNA was isolated from frozen cecal contents according to QIAamp DNA Stool mini Kit (Qiagen, Alameda, CA) with modifications [72]. Variable regions 1–4 of the 16S rRNA gene were amplified using TaKaRa Ex Taq PCR mixture (TAKARA Bio USA, Madison, WI ) and barcoded PCR primers A-788R and B-27F with the structure as follow: A-788R CGTATCGCCTCCCTCGCGCCATCAGGGACTACCAGGGTATCTAA, B-27F CTATGCGCCTTGCCAGCCCGCTCAG-MID-AGAGTTTGATCCTGGCTCAG. The MIDs used in these experiments were: MID4 AGCACTGTAG, MID5 ATCAGACACG, MID6 ATATCGCGAG, MID7 CGTGTCTCTA, MID9 TAGTATCAGC, MID10 TCTCTATGCG, MID11 TGATACGTCT, MID12 TACTGAGCTA. PCR program was set as follows 95°C 10 min and 30 cycles of 95°C 1 min, 50°C 1 min, 72°C 1.5 min followed by 72°C for 10 minutes. PCR products were purified using AMPure Kit (Agencourt Bioscience, Beverly, MA) and sequenced on GS Titanium 70×75 picotiter plate from adaptor B according to the manufacturer's protocols for GS FLX (Roche Applied Science, Indianapolis, IN). The resulting reads were sorted by barcode using SFF software tools, which also trims the barcode sequence in each read after sorting (Roche Applied Science, Indianapolis, IN). Sequences longer then 100bp were taxonomically annotated to the level of genus using the Ribosomal Database Project Classifier tool [30] with a bootstrap cutoff of 50%.

RNA isolation and cDNA library construction

The frozen ceca were cut in discs ~0.5–1 cm of thickness using a cold sterile pruner. Residual intestinal tissue was removed with the pruner and a handheld drill (Dremel, Racine, WI). The samples were kept deeply frozen during the procedure. Total microbial RNA isolation was performed using the RiboPure-Bacteria Kit (Ambion, Austin, TX). The MICROBExpress Bacterial mRNA Purification Kit (Ambion) was used to deplete the pool of 16S and 23S rRNA molecules present in the sample. cDNA libraries were constructed according to random primer cDNA synthesis protocol implemented in Double-Stranded cDNA Synthesis Kit (Invitrogen, Carlsbad, CA). Libraries were size selected in 1% low melting point agarose gel. The region of 250–800bp was removed from the gel and purified. Approximately 1µg of randomly-primed, size selected cDNAs were blunt-ended, adaptor ligated and converted to a single-stranded template DNA library using the GS Titanium General Library Prep Kit (Roche Applied Science, Indianapolis, IN). Libraries were prepared using barcode-containing adaptors in place of the standard Titanium adaptors, following Roche's instructions for preparation barcoded adaptors. The barcode sequences used for each library are listed above (MID 1–8). Libraries were quantified using Qubit reagents (Invitrogen, CA) and average fragment sizes were determined by analyzing 1 µl of the sized cDNA samples on the Bioanalyzer (Agilent, CA) using a DNA 7500 chip. The libraries were pooled in equimolar concentration into a single library. Processing for emulsion PCR, titration and sequencing on a GS FLX was done following the manufacturer's protocols (Roche Applied Science) on GS Titanium 70×75 picotiter plate. The resulting library reads were sorted by barcode using SFF software tools (Roche Applied Science).

Metatranscriptome sequence analysis

A total pool of initial 454 pyrosequencing reads was first screened for length (>100 bp) and rRNA contamination. Replicate reads were not removed, as a preliminary analysis of COG hits and taxonomic assignments indicated that roughly 1% of sequences were perfect replicates and that their presence did not significantly affect the relative abundance of transcripts (data not shown). Sequences were searched against an in-house database of 18,283 annotated long subunit (LSU) and short subunit (SSU) rRNA sequences compiled from the SILVA database [73] and the Ribosomal Database Project (RDP) [74]. Sequences were removed from further analysis based on BLASTN hits to the rRNA database that met the following criteria: (i) e-value<1e-5, and (ii) alignment length ≥40% of the sequence length. Screened cDNA sequences were assigned functional annotations based upon MG-RAST annotations using the SEED database and also upon best-BLASTX-hits to the NCBI COG database (e-values<1e-5). Taxonomic assignments (from phylum to genus) were made according to best-BLASTN-hits (requiring e-values<1e-5) to an in-house database (Table S2) of 1054 completed genomes from NCBI and the J. Craig Venter Institute.

Statistical analysis

Statistical tests for differentially abundant COG functions, cDNA-based taxonomic annotations, and 16S-based taxonomic annotations between populations (e.g. 21-day MF, 21-day FF) were made using the Metastats methodology with 1000 permutations to compute nonparametric p-values. Thresholds for significance were determined according to the specific comparison. For general COG categories and Level I and II SEED subsystems, the p-value threshold was 0.05. For individual COGs and level III SEED subsystems, the significance thresholds were 0.02 and 0.03, respectively.

Supporting Information

Figure S1

Principal component analysis of genus-level taxonomic assignments in MF and FF animals. MF animals (n = 4) represented by blue dots; FF animals (n = 4) represented by black dots.

(0.21 MB PDF)

Table S1

Gut microbial community structure in MF and FF piglets. Relative abundance of organisms was determined by amplification and sequencing of 16S rDNA present within microbial DNA isolated from the cecal contents of each animal. Relative abundance of sequences represents the number of sequences assigned to a given taxa for an individual animal divided by the total number of sequences for that animal. Mean relative abundance represents the average of these values in either the MF or FF groups.

(0.03 MB XLS)

Table S2

In-house collection of bacterial genomes used for taxonomic assigment of cDNA sequences.

(0.08 MB XLSX)

Table S3

Complete list of differentially abundant transcripts assigned to SEED subsystems in MF and FF piglets. Relative abundance of transcripts represents the number of sequences assigned to a subsystem for an individual animal divided by the total number of sequences for that animal. Mean relative abundance represents the average of these values in either the MF or FF treatment groups. A treatment group was considered to be enriched in transcripts assigned to a SEED subsystem if the p value was less than 0.05 for Level 1 subsystems and less than 0.03 for Level 3 subsystem. Subsystems with a mean relative abundance less than 0.0005 in both MF and FF groups were excluded.

(0.03 MB XLS)

Table S4

Differentially abundant COG hits in MF and FF piglets. Relative abundance is defined as the number of hits to an individual COG divided by the total number of hits to individual COGs for each individual subject. COGs were excluded if the mean relative abundance in both MF and FF groups was less than 0.0005. A treatment group was considered to be enriched in transcripts from an individual COG if the p value was less than 0.02.

(0.03 MB XLS)

Acknowledgments

The authors wish to thank Alvaro Hernandez and the staff of the W.M. Keck Center for Comparative and Functional Genomics at The University of Illinois (Urbana -Champaign) for sequencing support, the Initiative in Bioinformatics staff at the Computation Institute of The University of Chicago, and Vincent Denef, PhD for important insights regarding analysis of community gene expression data.

Footnotes

Competing Interests: SD has received research funding from Bristol Myers Squibb. This company played no role in the planning, conduct, or analysis of the experiments in this manuscript. This relationship does not alter the authors' adherence to PLoS ONE policies on sharing data and materials.

Funding: This work was supported by funding from the Surgical Infection Society (MM) and the March of Dimes Foundation (MM), as well as Bristol Myers-Squibb (SMD) and NIH R01 HD61929 (SMD). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1. Mackie RI, Sghir A, Gaskins HR. Developmental microbial ecology of the neonatal gastrointestinal tract. Am J Clin Nutr. 1999;69:1035S–45S. [PubMed]
2. Penders J, Thijs C, Vink C, Stelma FF, Snijders B, et al. Factors influencing the composition of the intestinal microbiota in early infancy. Pediatrics. 2006;118:511–21. [PubMed]
3. Palmer C, Bik EM, Digiulio DB, Relman DA, Brown PO. Development of the human infant intestinal microbiota. PLoS Biol. 2007;5:e177. [PMC free article] [PubMed]
4. Lundequist B, Nord CE, Winberg J. The composition of the faecal microflora in breastfed and bottle fed infants from birth to eight weeks. Acta Paediatr Scand 1985. 1985;74:45–51. [PubMed]
5. Hopkins MJ, Macfarlane GT, Furrie E, Fite A, Macfarlane S. Characterisation of intestinal bacteria in infant stools using real-time PCR and northern hybridisation analyses. FEMS Microbiol Ecol. 2005;54:77–85. [PubMed]
6. Penders J, Vink C, Driessen C, London N, Thijs C, et al. Quantification of bifidobacterium spp., escherichia coli and clostridium difficile in faecal samples of breast-fed and formula-fed infants by real-time PCR. FEMS Microbiol Lett. 2005;243:141–7. [PubMed]
7. Morales Y, Schanler RJ. Human milk and clinical outcomes in VLBW infants: How compelling is the evidence of benefit? Semin Perinatol. 2007;31:83–8. [PubMed]
8. Lin PW, Stoll BJ. Necrotising enterocolitis. Lancet. 2006;368:1271–83. [PubMed]
9. Qin J, Li R, Raes J, Arumugam M, Burgdorf KS, et al. A human gut microbial gene catalogue established by metagenomic sequencing. Nature. 2010;464:59–65. [PMC free article] [PubMed]
10. Andersson AF, Lindberg M, Jakobsson H, Backhed F, Nyren P, et al. Comparative analysis of human gut microbiota by barcoded pyrosequencing. PLoS One. 2008;3:e2836. [PMC free article] [PubMed]
11. Dinsdale EA, Edwards RA, Hall D, Angly F, Breitbart M, et al. Functional metagenomic profiling of nine biomes. Nature. 2008;452:629–32. [PubMed]
12. Klaassens ES, de Vos WM, Vaughan EE. Metaproteomics approach to study the functionality of the microbiota in the human infant gastrointestinal tract. Appl Environ Microbiol. 2007;73:1388–92. [PMC free article] [PubMed]
13. Verberkmoes NC, Russell AL, Shah M, Godzik A, Rosenquist M, et al. Shotgun metaproteomics of the human distal gut microbiota. ISME J. 2009;3:179–89. [PubMed]
14. Jansson J, Willing B, Lucio M, Fekete A, Dicksved J, et al. Metabolomics reveals metabolic biomarkers of crohn's disease. PLoS One. 2009;4:e6386. [PMC free article] [PubMed]
15. Saric J, Wang Y, Li J, Coen M, Utzinger J, et al. Species variation in the fecal metabolome gives insight into differential gastrointestinal function. J Proteome Res. 2008;7:352–60. [PubMed]
16. Turnbaugh P, Ridaura VK, Faith JJ, Rey FE, Knight R, et al. The effect of diet on the human gut microbiome: A metagenomic analysis in humanized gnotobiotic mice. Science Translational Medicine. 2009;1:6. [PMC free article] [PubMed]
17. Turnbaugh PJ, Quince C, Faith JJ, McHardy AC, Yatsunenko T, et al. Organismal, genetic, and transcriptional variation in the deeply sequenced gut microbiomes of identical twins. Proc Natl Acad Sci U S A. 2010;107:7503–8. [PMC free article] [PubMed]
18. Gilbert JA, Field D, Huang Y, Edwards R, Li W, et al. Detection of large numbers of novel sequences in the metatranscriptomes of complex marine microbial communities. PLoS ONE. 2008;3:e3042. [PMC free article] [PubMed]
19. Shi Y, Tyson GW, DeLong EF. Metatranscriptomics reveals unique microbial small RNAs in the ocean's water column. Nature. 2009;459:266–9. [PubMed]
20. Poretsky RS, Hewson I, Sun S, Allen AE, Zehr JP, et al. Comparative day/night metatranscriptomic analysis of microbial communities in the north pacific subtropical gyre. Environ Microbiol. 2009;11:1358–75. [PubMed]
21. Frias-Lopez J, Shi Y, Tyson GW, Coleman ML, Schuster SC, et al. Microbial community gene expression in ocean surface waters. Proc Natl Acad Sci. 2008;105:3805–10. [PMC free article] [PubMed]
22. Leininger S, Urich T, Schloter M, Schwark L, Qi J, et al. Archaea predominate among ammonia-oxidizing prokaryotes in soils. Nature. 2006;442:806–9. [PubMed]
23. Urich T, Lanzen A, Qi J, Huson DH, Schleper C, et al. Simultaneous assessment of soil microbial community structure and function through analysis of the meta-transcriptome. PLoS ONE. 2008;3:e2527. [PMC free article] [PubMed]
24. Wang Z, Gerstein M, Snyder M. RNA-seq: A revolutionary tool for transcriptomics. Nat Rev Genet. 2009;10:57–63. [PMC free article] [PubMed]
25. van Vliet AH. Next generation sequencing of microbial transcriptomes: Challenges and opportunities. FEMS Microbiol Lett. 2009;302:1–7. [PubMed]
26. Yoder-Himes DR, Chain PS, Zhu Y, Wurtzel O, Rubin EM, et al. Mapping the burkholderia cenocepacia niche response via high-throughput sequencing. Proc Natl Acad Sci. 2009;106:3976–81. [PMC free article] [PubMed]
27. Shelburne SA, 3rd, Keith D, Horstmann N, Sumby P, Davenport MT, et al. A direct link between carbohydrate utilization and virulence in the major human pathogen group A streptococcus. Proc Natl Acad Sci. 2008;105:1698–703. [PMC free article] [PubMed]
28. Gorke B, Stulke J. Carbon catabolite repression in bacteria: Many ways to make the most out of nutrients. Nat Rev Micro. 2008;6:613–24. [PubMed]
29. White JR, Nagarajan N, Pop M. Statistical methods for detecting differentially abundant features in clinical metagenomic samples. PLoS Comput Biol. 2009;5:e1000352. [PMC free article] [PubMed]
30. Wang Q, Garrity GM, Tiedje JM, Cole JR. Naive bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol. 2007;73:5261–7. [PMC free article] [PubMed]
31. Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, et al. A core gut microbiome in obese and lean twins. Nature. 2009;457:480–4. [PMC free article] [PubMed]
32. Kurokawa K, Itoh T, Kuwahara T, Oshima K, Toh H, et al. Comparative metagenomics revealed commonly enriched gene sets in human gut microbiomes. DNA Res. 2007;14:169–81. [PMC free article] [PubMed]
33. Meyer F, Paarmann D, D'Souza M, Olson R, Glass EM, et al. The metagenomics RAST server - a public resource for the automatic phylogenetic and functional analysis of metagenomes. BMC Bioinformatics. 2008;9:386. [PMC free article] [PubMed]
34. Tatusov RL, Koonin EV, Lipman DJ. A genomic perspective on protein families. Science. 1997;278:631–7. [PubMed]
35. Letunic I, Yamada T, Kanehisa M, Bork P. iPath: Interactive exploration of biochemical pathways and networks. Trends Biochem Sci. 2008;33:101–3. [PubMed]
36. Kurtz DM., Jr Avoiding high-valent iron intermediates: Superoxide reductase and rubrerythrin. J Inorg Biochem. 2006;100:679–93. [PubMed]
37. Martens EC, Koropatkin NM, Smith TJ, Gordon JI. Complex glycan catabolism by the human gut microbiota: The bacteroidetes sus-like paradigm. J Biol Chem. 2009;284:24673–7. [PMC free article] [PubMed]
38. Terra VS, Homer KA, Rao SG, Andrew PW, Yesilkaya H. Characterization of novel {beta}-galactosidase activity that contributes to glycoprotein degradation and virulence in streptococcus pneumoniae. Infect Immun. 2010;78:348–57. [PMC free article] [PubMed]
39. Ge J, Catt DM, Gregory RL. Streptococcus mutans surface alpha-enolase binds salivary mucin MG2 and human plasminogen. Infect Immun. 2004;72:6748–52. [PMC free article] [PubMed]
40. Candela M, Biagi E, Centanni M, Turroni S, Vici M, et al. Bifidobacterial enolase, a cell surface receptor for human plasminogen involved in the interaction with the host. Microbiology. 2009;155:3294–303. [PubMed]
41. Candela M, Bergmann S, Vici M, Vitali B, Turroni S, et al. Binding of human plasminogen to bifidobacterium. J Bacteriol. 2007;189:5929–36. [PMC free article] [PubMed]
42. Granato D, Bergonzelli GE, Pridmore RD, Marvin L, Rouvet M, et al. Cell surface-associated elongation factor tu mediates the attachment of lactobacillus johnsonii NCC533 (La1) to human intestinal cells and mucins. Infect Immun. 2004;72:2160–9. [PMC free article] [PubMed]
43. Sanchez B, Schmitter JM, Urdaci MC. Identification of novel proteins secreted by Lactobacillus plantarum that bind to mucin and fibronectin. J Mol Micro Biotech. 2009;17:158–162. [PubMed]
44. Egea L, Aguilera L, Gimenez R, Sorolla MA, Aguilar J, et al. Role of secreted glyceraldehyde-3-phosphate dehydrogenase in the infection mechanism of enterohemorrhagic and enteropathogenic escherichia coli: Interaction of the extracellular enzyme with human plasminogen and fibrinogen. Int J Biochem Cell Biol. 2007;39:1190–203. [PubMed]
45. Lerner A, Castro-Sowinski S, Valverde A, Lerner H, Dror R, et al. The azospirillum brasilense Sp7 noeJ and noeL genes are involved in extracellular polysaccharide biosynthesis. Microbiology. 2009;155:4058–68. [PubMed]
46. Mailloux RJ, Singh R, Brewer G, Auger C, Lemire J, et al. Alpha-ketoglutarate dehydrogenase and glutamate dehydrogenase work in tandem to modulate the antioxidant alpha-ketoglutarate during oxidative stress in pseudomonas fluorescens. J Bacteriol. 2009;191:3804–10. [PMC free article] [PubMed]
47. Goldberg JB, Hatano K, Pier GB. Synthesis of lipopolysaccharide O side chains by pseudomonas aeruginosa PAO1 requires the enzyme phosphomannomutase. J Bacteriol. 1993;175:1605–11. [PMC free article] [PubMed]
48. Houot L, Chang S, Absalon C, Watnick PI. Vibrio cholerae phosphoenolpyruvate phosphotransferase system control of carbohydrate transport, biofilm formation, and colonization of the germfree mouse intestine. Infect Immun. 2010;78:1482–94. [PMC free article] [PubMed]
49. Fugier E, Salcedo SP, de Chastellier C, Pophillat M, Muller A, et al. The glyceraldehyde-3-phosphate dehydrogenase and the small GTPase rab 2 are crucial for brucella replication. PLoS Pathog. 2009;5:e1000487. [PMC free article] [PubMed]
50. Schutte JB, de Jong J, van Weerden EJ, Tamminga S. Nutritional implications of L-arabinose in pigs. Br J Nutr. 1992;68:195–207. [PubMed]
51. Vaillancourt K, Bedard N, Bart C, Tessier M, Robitaille G, et al. Role of galK and galM in galactose metabolism by streptococcus thermophilus. Appl Environ Microbiol. 2008;74:1264–7. [PMC free article] [PubMed]
52. Sasajima K, Kumada T. Deficiency of flagellation in a bacillus subtilis pleiotropic mutant lacking transketolase. Agric Biol Chem. 1983;47:1375–6.
53. Eidels L, Osborn MJ. Lipopolysaccharide and aldoheptose biosynthesis in transketolase mutants of salmonella typhimurium. Proc Natl Acad Sci. 1971;68:1673–7. [PMC free article] [PubMed]
54. Gonzalez R, Klaassens ES, Malinen E, de Vos WM, Vaughan EE. Differential transcriptional response of bifidobacterium longum to human milk, formula milk, and galactooligosaccharide. Appl Environ Microbiol. 2008;74:4686–94. [PMC free article] [PubMed]
55. Bertolo RF, Burrin DG. Comparative aspects of tissue glutamine and proline metabolism. J Nutr. 2008;138:2032S–9S. [PubMed]
56. Aycicek A, Erel O, Kocyigit A, Selek S, Demirkol MR. Breast milk provides better antioxidant power than does formula. Nutrition. 2006;22:616–9. [PubMed]
57. Friel JK, Martin SM, Langdon M, Herzberg A, Gene R, et al. Milk from mothers of both premature and full-term infants provides better antioxidant protection than does infant formula. Pediatr Res. 2002;51:612–8. [PubMed]
58. Tsolis RM, Baumler AJ, Heffron F, Stojilkovic I. Contribution of TonB- and Feo-mediated iron uptake to growth of Salmonella typhimurium in the mouse. Infect Immun. 1996;64:4549–56. [PMC free article] [PubMed]
59. Torres AG, Redford P, Welch RA, Payne SM. Ton-B dependent systems of uropathogenic Escherichia coli: aerobactin and heme transport and TonB are required for virulence in the mouse. Infect Immun. 2001;69:6179–85. [PMC free article] [PubMed]
60. Hooper LV, Gordon JI. Commensal host-bacterial relationships in the gut. Science. 2001;292:1115–8. [PubMed]
61. Sears CL. A dynamic partnership: Celebrating our gut flora. Anaerobe. 2005;11:247–51. [PubMed]
62. Hooper LV, Wong MH, Thelin A, Hansson L, Falk PG, et al. Molecular analysis of commensal host-microbial relationships in the intestine. Science. 2001;291:881–4. [PubMed]
63. Wu S, Rhee KJ, Albesiano E, Rabizadeh S, Wu X, et al. A human colonic commensal promotes colon tumorigenesis via activation of T helper type 17 T cell responses. Nat Med. 2009;15:1016–22. [PMC free article] [PubMed]
64. Ley RE, Backhed F, Turnbaugh P, Lozupone CA, Knight RD, et al. Obesity alters gut microbial ecology. Proc Natl Acad Sci. 102:11070–5. [PMC free article] [PubMed]
65. Frank DN, St Amand AL, Feldman RA, Boedeker EC, Harpaz N, et al. Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proc Natl Acad Sci U S A. 104:13780–5. [PMC free article] [PubMed]
66. Frank DN, Pace NR. Gastrointestinal microbiology enters the metagenomics era. Curr Opin Gastroenterol. 2008;24:4–10. [PubMed]
67. Allen EE, Banfield JF. Community genomics in microbial ecology and evolution. Nat Rev Microbiol. 2005;3:489–98. [PubMed]
68. Prangli AL, Utt M, Talja I, Sepp E, Mikelsaar M, et al. Antigenic proteins of lactobacillus acidophilus that are recognised by serum IgG antibodies in children with type 1 diabetes and coeliac disease. Pediatr Allergy Immunol. Epub ahead of print Jul 2 2009 [PubMed]
69. Shah P, Shah V. Arginine supplementation for prevention of necrotising enterocolitis in preterm infants. Cochrane Database Syst Rev. 2007;3:CD004339. [PubMed]
70. Amin HJ, Zamora SA, McMillan DD, Fick GH, Butzner JD, et al. Arginine supplementation prevents necrotizing enterocolitis in the premature infant. J Pediatr. 2002;140:425–31. [PubMed]
71. Houle VM, Schroeder EA, Odle J, Donovan SM. Small intestinal disaccharidase activity and ileal villus height are increased in piglets consuming formula containing recombinant human insulin-like growth factor-I. Pediatr Res. 1997;42:78–86. [PubMed]
72. Zoetendal EG, Heilig HG, Klaassens ES, Booijink CC, Kleerebezem M, et al. Isolation of DNA from bacterial samples of the human gastrointestinal tract. Nat Protoc. 2006;1:870–3. [PubMed]
73. Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig W, et al. SILVA: A comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucl Acids Res. 2007;35:7188–96. [PMC free article] [PubMed]
74. Cole JR, Wang Q, Cardenas E, Fish J, Chai B, et al. The ribosomal database project: Improved alignments and new tools for rRNA analysis. Nucl Acids Res. 2009;37:D141–145. [PMC free article] [PubMed]

Articles from PLoS ONE are provided here courtesy of Public Library of Science

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...