- Research article
- Open Access
Bifidobacteria grown on human milk oligosaccharides downregulate the expression of inflammation-related genes in Caco-2 cells
BMC Microbiologyvolume 15, Article number: 172 (2015)
Breastfed human infants are predominantly colonized by bifidobacteria that thrive on human milk oligosaccharides (HMO). Two predominant species of bifidobacteria in infant feces are Bifidobacterium breve (B. breve) and Bifidobacterium longum subsp. infantis (B. infantis), both of which include avid HMO-consumer strains. Our laboratory has previously shown that B. infantis, when grown on HMO, increases adhesion to intestinal cells and increases the expression of the anti-inflammatory cytokine interleukin-10. The purpose of the current study was to investigate the effects of carbon source—glucose, lactose, or HMO—on the ability of B. breve and B. infantis to adhere to and affect the transcription of intestinal epithelial cells on a genome-wide basis.
HMO-grown B. infantis had higher percent binding to Caco-2 cell monolayers compared to B. infantis grown on glucose or lactose. B. breve had low adhesive ability regardless of carbon source. Despite differential binding ability, both HMO-grown strains significantly differentially affected the Caco-2 transcriptome compared to their glucose or lactose grown controls. HMO-grown B. breve and B. infantis both downregulated genes in Caco-2 cells associated with chemokine activity.
The choice of carbon source affects the interaction of bifidobacteria with intestinal epithelial cells. HMO-grown bifidobacteria reduce markers of inflammation, compared to glucose or lactose-grown bifidobacteria. In the future, the design of preventative or therapeutic probiotic supplements may need to include appropriately chosen prebiotics.
Milk is a unique biological fluid consumed by mammalian infants. It contains many macro- and micro-nutrients that are essential for the growth and development of the newborn [1, 2]. In addition, a diverse cocktail of bioactive factors, such as oligosaccharides, antibodies and nucleotides in milk, play immune, prebiotic and protective functions in the infant gut [2–4]. Oligosaccharides are the third most abundant component in human milk and they are present as lactose-derived free forms and protein and lipid bound glycoconjugates . Milk oligosaccharides can withstand the pH of the stomach and virtually all of them resist enzymatic digestion in the gastro-intestinal tract . Recent studies on human milk oligosaccharides (HMO) and glycoconjugates demonstrate both local and systemic beneficial effects to the suckling neonate [7–9]. Milk oligosaccharides provide protection against enteric pathogen infections by antibacterial activity, competing with pathogen binding sites and enhancing the binding of IgA with pathogens . Another protective function of milk oligosaccharides is that the intact oligosaccharides serve as a prebiotic substrate enabling enrichment of Bifidobacterium species in the infant gut thereby consuming available nutrients and lowering the gut pH .
Bifidobacterium species were first observed over 100 years ago in feces of breastfed infants and later research suggested breast milk contains special molecules defined as “bifidus factors” that stimulate the growth of bifidobacteria [12, 13]. Culture-based studies over the years and high-throughput metagenomic studies have demonstrated that Bifidobacterium is a commonly enriched member of the intestinal microbiota of breastfed infants [14, 15]. Research in the last decade has provided a mechanistic basis for that enrichment whereby HMOs and glycoconjugates serve as prebiotics selectively promoting bifidobacteria . Indeed, genomic analysis of a prototypical infant borne bifidobacteria, Bifidobacterium longum subsp. infantis (B. infantis) which grows well on HMO, revealed a single cluster of genes encoding milk oligosaccharide metabolism suggesting co-evolution of this strain with human milk [17, 18]. Analysis of other infant-borne strains of B. bifidum, B. longum subsp. longum (B. longum) and B. breve shown to grow on HMO [19, 20] also possessed specific milk glycan transporters and glycosyl hydrolases linked to milk glycan consumption [16, 19].
The ability of bifidobacteria to bind and interact with the intestinal epithelium plays an important role in gut colonization and modulation of host immune system [21, 22]. Previous research has proven that different species of bifidobacteria exhibit different capacities to adhere to the intestinal epithelium and to stimulate the gastrointestinal immunity [23–25]. Our research group recently showed that the adhesion rate of bifidobacteria to the intestinal epithelial cells (IECs) changes according to the carbon source supplied in their growth medium . When comparing B. infantis ATCC 15697 grown in HMO and lactose, the HMO grown B. infantis had a significantly higher rate of adhesion to both Caco-2 and HT-29 cells . This work was subsequently confirmed by Kavanaugh et al. showing that growth of B. infantis ATCC 15697 on 6’sialyllactose (an HMO component structure) also resulted in increased adherence to cultured IECs . Previous research also showed the ability of bifidobacteria to induce the anti-inflammatory capacity of IECs. For example, colon organ cultures exposed to B. infantis showed reduced production of pro-inflammatory cytokine IL-17 . In another study, interferon gamma was reduced in the Peyer’s patches of mice fed B. longum .
There are few documented studies on changes in the interaction between bifidobacteria and IECs as a function of the carbon source of the bacterial growth medium. Microarray studies in our laboratory have previously shown that incubation of HMO with B. infantis altered gene expresssion in Caco-2 cells . Chichlowski et al. observed that HMO-fed B. infantis tightens cell-cell junctions, increases the level of cytokine IL-10 while decreasing the level of pro-inflammatory TNFα in Caco-2 cells .
In the current study, we sought to understand the effects of HMO-fed bifidobacteria on the gene expression of intestinal cells on a genome-wide basis. We investigated the effects of two bifidobacterial strains fed different carbon sources—HMO, glucose (GLU), or lactose (LAC)—on the gene expression of Caco-2 cells using RNA sequencing (RNA-Seq). Due to their dominance in breastfed infant feces  and efficient consumption of HMOs, B. infantis and B. breve were selected for the study. The two strains selected were B. infantis ATCC 15697, which is the current model to study genetic adaptations to growth on human milk  and B. breve SC95, a strain that also grows vigorously on HMOs . Our group has also shown that HMO consumption by B. infantis triggers expression of surface binding proteins that interact with intestinal cell surface structures . We hypothesized that HMO-grown bifidobacteria, relative to those grown on LAC or GLU will have enhanced adhesion and will alter gene expression in Caco-2 cells consistent with a protective modulatory mechanism in the host intestine.
Adhesion of bifidobacteria grown in different carbon sources to Caco-2 cells
Several studies have demonstrated the ability of B. infantis strains to grow on HMO as the sole carbon source  and a recent study in our laboratory identified three strains of B. breve (SC95, SC154, and ATCC 1570) that can also efficiently consume HMO . Based on the results of this study and their exclusive presence in infant feces, B. breve SC95 was selected for the adhesion assay. Different adhesion percentages were observed between the two bifidobacterial species. The levels of adhesion in B. infantis ATCC 15697 ranged from 1.1 to 9.6 % and in B. breve SC95 it ranged from 0.9 to 1.3 % (Fig. 1). Binding percentages obtained for B. infantis ATCC 15697 were very similar to that reported previously [25, 26]. However, the current study included the GLU-grown B. infantis ATCC 15697, which showed the lowest percentage of adhesion. Compared to B. infantis ATCC 15697, B. breve SC95 presented lower binding efficiency to Caco-2 cells. The highest binding percentage (1.3 %) of B. breve SC95 was observed for LAC-grown bacteria and there were numerical trends among treatments and the adhesion efficiency, however no statistical significant differences were observed among them. Studies of different B. breve strains grown in MRS media have shown very low adhesion percentages to Caco-2 cells . In the current study, we have observed lower binding of B. breve SC95, regardless of the carbon source.
Gene expression analysis of Caco-2 cells in response to bifidobacteria grown on different substrates
In order to examine if bifidobacteria grown in different carbon sources elicit a change in the intestinal cellular responses, we examined the transcriptome of the Caco-2 cells using RNA-Seq. This resulted in 20–32 million reads per sample and only uniquely mapped reads were considered in the analysis. As described by Bentley et al.  and Ramskold et al.  a threshold RPKM (reads per kilo base per million mapped reads) value of 0.1 RPKM was established to balance the number of false positives and false negatives and to define potential meaningful gene expression. There were 15,613, 15,574, and 15,520 genes expressed in Caco-2 cells exposed to B. breve SC95 grown in GLU, HMO and LAC respectively. Caco-2 cells exposed to B. infantis grown in GLU, HMO and LAC had 15,561, 15,468 and 15,516 expressed genes respectively. Summaries of gene expression intensities for all transcripts in all samples are provided in the NCBI GEO repository, accessions GSE63950 and GSE64017.
Effect of carbon source consumption by bifidobacteria on Caco-2 gene expression
To determine the specific effect of carbon source on Caco-2 gene expression, differential expression between carbon sources were analyzed individually in each bifidobacteria strain experiment. As expected, bifidobacteria grown in different carbon sources elicited significant changes (p-value ≤0.05, FDR q ≤ 0.5, fold change ≥2) in expression of genes in Caco-2 cells. The number of differentially expressed genes in Caco-2 cells as a function of carbon source is summarized in Table 1. During co-incubation with B. breve SC95, the expression of 12 Caco-2 cell genes was upregulated and 61 genes were downregulated by HMO compared to GLU, while 196 genes were upregulated and 144 genes were downregulated by HMO compared to LAC. Comparison between GLU vs LAC grown samples of B. breve SC95 yielded 96 up-regulated genes and 426 downregulated genes by LAC.
Co-incubation of Caco-2 cells with B. infantis ATCC 15697 also modified Caco-2 gene expression as a function of carbon source (Table 1). There were 107 genes upregulated and 178 genes downregulated by HMO compared to GLU and upregulation of 21 genes and downregulation of 37 genes by HMO compared to LAC. When comparing GLU vs LAC grown B. infantis ATCC 15697, 148 genes were upregulated and 97 genes were downregulated by LAC (Table 1). List of these genes with significant changes in expression and their RPKM values are provided in Additional file 1 and Additional file 2.
To compare the differential effects of bifidobacterial strain and carbon source on the Caco-2 cell transcriptome, between-strain differential gene expression was examined for each of the three carbon sources. The number of Caco-2 genes differentially expressed in response to co-incubation with B. infantis compared with B. breve for each carbon source is listed in Fig. 2. For example, there are 2596 Caco-2 genes that are differentially expressed in response to GLU-fed B. infantis compared to GLU-fed B. breve that are not also differentially expressed in response to HMO- or LAC-fed strains. The carbon source of glucose elicits the biggest bifidobacteria strain effect on Caco-2 gene expression. Meanwhile, the use of HMO as a carbon source appears to reduce the difference between strains. If the choice of carbon source were unimportant, the intersection of the gene expression results (Fig. 2) should be quite high. However, that is not the case. The intersection of the three carbon sources is rather small (Fig. 2), suggesting that the choice of carbon source definitively affects Caco-2 response to different bifidobacteria strains. Thus, the differential response of Caco-2 cells to the two different bifidobacteria strains is highly dependent on the carbon source.
Predicted consequences of bifidobacterial carbon source on Caco-2 function
To investigate the functional consequences of bifidobacteria carbon source on Caco-2 cell gene expression, functional annotation clustering was performed using DAVID  for each input gene list (Table 1). These results were filtered to obtain terms with statistically significant enrichment for each comparison (Tables 2 and 3). In the co-incubation of B. breve SC95 with Caco-2 cells (Table 2), genes that were downregulated by HMO, relative to GLU, were enriched with the annotation term GO:0008009 ~ chemokine activity. Several other annotation terms related to inflammation and immunity were also significantly enriched among genes downregulated by HMO grown B. breve SC95 (Table 2). There was no significant enrichment of annotation terms for genes up-regulated by HMO grown, relative to GLU grown, B. breve SC95. Likewise, genes downregulated by HMO-fed B. breve SC95, relative to LAC-fed, were also enriched with the annotation term “chemokine activity.”
Functional enrichment analysis of the Caco-2 cell genes modulated by B. infantis (Table 3) also showed the modulation of “chemokine activity” by HMO. Relative to LAC-fed B. infantis, HMO-fed B. infantis downregulated Caco-2 cell genes associated with “chemokine activity.” However, this difference was not seen in the comparison between HMO-fed and GLU-fed B. infantis. There was no significant enrichment of annotation terms for genes upregulated by HMO-fed B. infantis compared to either GLU- or LAC-fed. In summary, consumption of HMO by two different bifidobacterial strains was associated with downregulation of inflammation-related signaling in Caco-2 cells.
Expression of inflammation related genes is reduced by HMO-grown bifidobacteria
Heat maps summarizing significantly differentially expressed inflammation–related genes in Caco-2 cells co-incubated with B. infantis and B. breve are shown in Figs. 3 and 4, respectively. Nearly all of these genes are downregulated by HMO feeding relative to either GLU or LAC or both.
Necrotizing enterocolitis (NEC) is a very common emergency occurring in pre-term infants. Previous studies have shown that NEC can be prevented by probiotics perhaps in part because they modulate the immune mediated gene expression in enterocytes . Given that the functional enrichment analyses and the heat maps (Figs. 3 and 4) pointed to chemokine-related changes in gene expression in the Caco-2 cells exposed to bifidobacteria grown in different carbon sources, we specifically investigated genes involved in inflammation. Expression of CXCL1, CXCL2 and CXCL3 were downregulated by HMO feeding of both strains of bifidobacteria (Fig. 5a and b). Another cell adhesion and chemotaxis target gene that has shown higher expression in NEC is ICAM1  and interestingly this gene was significantly downregulated in Caco-2 cells exposed to both B. infantis ATCC 15697 and B. breve SC95 grown in HMO (Figs. 3 and 4).
Aberrant modulation of gut bacteria, such as bifidobacteria, has been implicated in inflammatory bowel disease (IBD) [40, 41]. Therefore, we screened candidate genes for ulcerative colitis (UC) and Crohn’s disease for significant changes in expression in the RNA-Seq data. Numerous candidate genes were downregulated in Caco-2 cells exposed to B. infantis ATCC 15697 and/or B. breve SC95 grown on HMO (Table 4). ICAM1, which is known to be highly expressed in both Crohn’s disease and UC relative to controls , was downregulated in Caco-2 cells exposed to either HMO-fed bifidobacterial strain.
Commensal bifidobacteria have been associated with regulation of intestinal inflammation [38, 43, 44]. Bifidobacteria are normal residents of the human intestine, commonly found in the infant gut and particularly enriched in the intestine of breastfed infants . B. infantis and B. breve are often dominant members of the breastfed infant gut microbiota  and have been linked to a modulatory role in intestinal inflammation and regulation of the immune response at the gut associated lymphoid tissue . It is well accepted that bifidobacteria overrepresentation in nursing infants brings benefits to the neonate .
One of the major driving forces that underlie bifidobacterial predominance in infants is the ability to consume HMOs. In a previous study, we demonstrated that growth on HMOs significantly increased the binding of B. infantis, but not B. bifidum, to intestinal cells . Consistent with that study, we found that HMO-grown B. infantis showed significant adherence to Caco-2 cells. More recently, we have characterized unique strains of B. breve that grow well on HMO  which led us to hypothesize that these strains might also exhibit an HMO-induced binding phenotype. However, unlike B. infantis, B. breve SC95 showed low percentage of binding with Caco-2 cells, regardless of the growth medium. Previous studies demonstrate that the adherence of B. breve to Caco-2 cells is strain-specific and that adherence of a high-adhering strain of B. breve is almost totally abolished with trypsin or pronase treatment of the culture . In the current protocol we have discarded the spent culture and this may explain the low binding percentage of B. breve SC95. An interesting future study design would be to include the spent culture of bifidobacteria in Caco-2 binding assays. The transcriptomic changes of Caco-2 cells evoked by B. breve may result from the constitutive expression of surface structures or the production of soluble factors. There is evidence that bifidobacteria exert an indirect effect on intestinal inflammation, due to the release of soluble factors that reduce inflammation .
Taken together, both strain and carbon source affect the binding affinity of bifidobacteria to IECs. In a previous study, we found that HMO-grown B. infantis increased the expression of genes involved in promoting integrity of the barrier function . However, that study was a targeted approach in which only a few genes were probed. Here, we employed a high-throughput sequencing method to obtain an overview of all gene expression in Caco-2 cells upon incubation with B. infantis ATCC 15697 and B. breve SC 95 grown in different carbon sources. With each bifidobacterial strain, the supplied carbon source—GLU, LAC, or HMO—significantly affected gene expression in Caco-2 cells. Interestingly, even though the adhesive ability of B. breve was low, it still elicited a significant transcriptional response in Caco-2 cells, altering even more genes than B. infantis. These observations suggest that mechanisms additional to that of direct binding of B. breve are important to promote gene expression changes in Caco-2 cells. Indirect mechanisms, such as the production of soluble factors by B. breve during interaction with IECs, may underlie the gene expression changes and it should be investigated.
In a non-hypothesis driven assessment of Caco-2 gene expression in response to bifidobacteria grown on different carbon sources, we found that growth on HMOs substrate reduced inflammation-related signaling. The reduced expression of genes related to chemokine activity in the presence of HMO-grown bifidobacteria occurred independent of bifidobacterial strain, suggesting that prebiotic HMO primes bifidobacteria to elicit an anti-inflammatory state in IECs. RNA-Seq analysis showed significant downregulation of three chemokines in CXCL family: CXCL1, CXCL2 and CXCL3. These chemokines play an active role in development of gastrointestinal diseases marked by inflammatory response such as gastritis, necrotizing enterocolitis (NEC), ileitis, ulcerative colitis  and inflammatory bowel diseases (IBD) . Mouse studies have showed an association between higher incidence of necrotizing enterocolitis and elevated intestinal expression of CXCL1 mRNA . A recent study where B. infantis was administered to mouse pups that are prone to NEC has shown significant downregulation of CXCL1 gene in the intestinal epithelium . The in vivo blocking of CXCL2 has been shown to alleviate inflammation related bowel injuries . CXCL1, CXCL2 and CXCL3 are positively associated with tumor associated angiogenesis and depletion of these three chemokine factors have inhibited the tumor growth in mice .
Given the effect of HMO-grown bifidobacteria on inflammatory related signaling in Caco-2 cells in this study, we specifically reviewed the Caco-2 transcriptome data in the context of known candidate genes associated with NEC, Crohn’s diseases and ulcerative colitis. Several candidate genes associated with these bowel diseases are significantly regulated by HMO-grown bifidobacteria (Table 4). Expression of intracellular adhesion molecule-1 (ICAM1) was significantly downregulated by B. breve and B. infantis grown in HMO. ICAM1, located in the luminal surface of the intestinal epithelium, is a ligand for neutrophils. ICAM1 facilitates the transepithelial migration of neutrophils and their accumulation in the luminal surface of the intestine, which contributes to mucosal injury leading to conditions such as ulcerative colitis . Therefore the decreased expression of ICAM1 by B. breve and B. infantis grown in HMO may aid to reduce the risk of NEC in infants and inflammatory bowel disease in adults. These results highlight the therapeutic potential to prevent or possibly ameliorate IBD with synbiotics that include both bifidobacteria and milk oligosaccharides.
This study presents an initial assessment of the transcriptomic changes of intestinal epithelial cells evoked by commensal bifidobacteria grown on different sugars. We provide evidence that consumption of HMOs, which are abundant in the gastrointestinal tract of nursing infants, promotes a beneficial interaction between bifidobacteria and the intestinal epithelium. Consumption of free milk glycans by bifidobacteria leads to reduced expression of inflammatory genes, contributing to maintenance of the integrity of the intestinal mucosa. Our study represents a major step towards a better understanding of bifidobacteria-host interactions that take place in the infant gut, which can be applied to the use of bifidobacteria as probiotics to promote gut health. Moreover, the results presented herein provide support for the use of prebiotic HMOs as the best growth substrate for bifidobacteria, not only by supporting their growth to high densities and contributing to niche occupation, but also by inducing protective responses in the host.
Bacterial strains and culture conditions
Bifidobacterium longum subsp. infantis ATCC 15697 and B. breve SC95 were routinely grown for 48 h anaerobically at 37 °C in the semisynthetic de Man-Rogosa-Sharpe (MRS) broth or MRS agar (Becton Dickinson) supplemented with 1 % (wt/vol) L-cysteine. Single colony isolates were inoculated into modified MRS (mMRS) without sugar supplemented with 2 % of filter-sterilized (wt/vol) lactose (LAC) (Sigma Aldrich), 2 % glucose (GLU) (Fisher) or 2 % purified human milk oligosaccharide mixture (HMO) as the sole carbon source. HMO was kindly provided by the laboratory of Dr. Bruce German (Department of Food Science and Technology, University of California, Davis) and purified according to the method described in Gnoth et al. .
Caco-2 cell culture
Enterocyte-like human colon adenocarcinoma (Caco-2) cells were obtained from the American Type Culture Collection (ATCC® HTB37™). These cells undergo spontaneous differentiation during culture, expressing some transporters and metabolic enzymes normally present in the gut . Caco-2 cells were routinely cultured in 75-cm2 flasks at 37 °C in a 5 % CO2 constant-humidity environment with medium replaced every 2–3 days. Caco-2 cells were grown using in Dulbecco’s modified Eagle medium (DMEM) containing 20 % heat-inactivated fetal calf serum (FBS), 1 % nonessential amino acids, 50 IU/mL penicillin, and 50 μg/mL streptomycin. Cells were subcultured at 80 % confluence by adding 0.25 % trypsin/0.9 mM EDTA solution (Invitrogen, CA). For adhesion and gene expression assays, Caco-2 cells were seeded in 24-well plate (2 cm2/well; BD Falcon, Franklin Lakes, NJ) at 1 × 105cells per well. The viable cell number was obtained using a trypan blue and TC20™automated cell counter. Caco-2 cell monolayers were used fifteen days after confluence, a time when morphological and functional differentiation is complete .
B. infantis ATCC 15697 and B. breve SC95 from the exponentially grown 48 h-old cultures supplemented with HMO, GLU or LAC were collected by centrifugation (4000 g for 10 min), washed with sterile phosphate-buffered saline (PBS; pH 7.3), and resuspended in DMEM at approximately 1 × 108 cells/mL. For reference purposes (100 % values), 1 ml aliquots of the original bacterial cell suspensions used in the adhesion assay were centrifuged, the cells resuspended in 200 μl trypsin/EDTA plus 200 μl PBS and then frozen and stored at -20 °C until quantification of the bacteria. B. infantis ATCC 15697 and B. breve SC95 bacterial suspensions were incubated with a monolayer of fully differentiated Caco-2 cells at 37 °C, 5 % CO2 for 2 h. All incubations were performed in biologically independent triplicates. After 2 h of incubation, cell monolayers were gently washed three times with PBS, to remove unbound bacteria, and then detached from the plastic surface by incubation with 200 μl trypsin/EDTA per well (10 min, 37 °C). To perform quantification of adherent bacteria, cell suspensions were incubated at 37 °C for 30 min in Gram-positive lysis buffer (20 mM Tris–HCl, 2 mM sodium EDTA, 1.2 % Triton X-100 and 20 mg/ml lysozyme). Quantification of adherent bacteria was performed by quantitative PCR targeting the 16S rRNA gene. The primers employed were as follows: Bif F (5′-TCGCGTCTGGTGTGAAAG-3′) and Bif R (5′-CCACATCCAGCGTCCAC-3′) for B. infantis . B. breve was analyzed using primers BiBRE-F (5′- CCGGATGCTCCATCACAC-3′) and BiBRER (5′- ACAAAGTGCCTTGCTCCCT-3′). A standard curve for quantification of bifidobacterial strains was generated from serial dilutions of bacterial DNA and used to calculate numbers of bacterial copies. Estimates of the number of bifidobacterial genome copies in the standard were based on a genome size (1.75–2.8 Mb) of the individual strain. Quantitative PCR was performed in a 7500 Fast Real-Time PCR System (Applied Biosystems, CA) using SYBR Green fluorophore. The PCR reactions and melting curves were performed in 20 μl containing 1 μl of each primer, 10 μl SYBR Green PCR Master Mix 2× (Applied Biosystems, CA), and 2 μl of bacterial DNA. The PCR reaction was incubated at 95 °C for 10 min, followed by 40 cycles consisting of 20 s at 95 °C, 20 s 56 °C, and 30 s at 60 °C. Bacterial adhesion was expressed as the number of adherent bacteria divided by total number of bacteria added, multiplied by 100 .
RNA-Seq experiment and data analysis
B. infantis ATCC 15697 and B. breve SC95 cells were incubated with fully differentiated Caco-2 cell monolayers as described above for the adhesion assay. All incubations were performed in biologically independent triplicates that consisted of independent bacterial cultures and Caco-2 cells that are of different passages. After 2 h of incubation, Caco-2 cell monolayers were gently washed three times with PBS and RNA was extracted from Caco-2 cells using the Trizol method according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). RNA was quantified by an ND-1000 spectrophotometer (Fisher Thermo, Wilmington, MA), and the quality and integrity was assessed by the spectrophotometer 260/280 ratio, gel electrophoresis and capillary electrophoresis with an Experion bio-analyzer (Bio-Rad, Hercules, CA).
Gene expression analysis was conducted using Illumina RNA-Seq technology. Messenger RNA was isolated and purified using a TruSeq RNA Sample Prep Kit v2 (Illumina, San Diego, CA). The fragments were sequenced at Vincent J. Coates Genomics Sequencing Laboratory of University of California Berkeley using the Illumina HiSeq2000. Sequence reads of 100 bp were assembled and analyzed in RNA-Seq and expression analysis application of CLC Genomics Workbench 5.5.1 (CLC Bio, Aarhus, Denmark). The human genome, H. sapiens Build 37.1 (ftp://ftp.ncbi.nih.gov/genomes/H_sapiens/ARCHIVE/BUILD.37.1/) was utilized as the reference genome for the assembly. The following criteria were used to filter the unique sequence reads: minimum length fraction of 0.9; minimum similarity fraction of 0.8; maximum number of two mismatches. Data were normalized by calculating the reads per kilo base per million mapped reads (RPKM = total exon reads/mapped reads in millions × exon length in kb)  for each gene and annotated with Ensembl human genome assembly GRCh37.p11 (57,412 total genes).
Significant gene expression changes in Caco-2 cells exposed to B. infantis ATCC 15697 or B. breve SC95 grown in HMO, GLU or LAC were analyzed using t-tests on log2 transformed data (0.5 was added to each number before log transformation to deal with zero counts). Analyses were conducted, 1) on the same strain grown on different sugars 2) between the two strains grown on same sugars. Genes with p-value ≤0.05, FDR q ≤ 0.5 and fold change ≥2 were considered to be statistically significant.
Genes with significant change in expression levels were further analyzed using the functional analysis clustering tool of The Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.7 . High classification stringency, enrichment score ≥1.3, p-value (EASE score) ≤0.05 and globally corrected enrichment Benjamini p-value (to control for family-wide false discovery rate) ≤0.05 were the statistical parameters used to cluster functionally similar annotation terms associated with the input gene list.
Inflammation is a valuable protection system in the body against harmful external and internal stimuli . However the dysfunctional regulation of the intestinal immune system is a contributory factor to many diseases such as necrotizing enterocolitis (NEC) , inflammatory bowel disease (IBD) , celiac disease, and Crohn’s disease . Loza et al. assembled 1027 inflammation related genes using literature survey and Ingenuity pathway analysis . This gene list was used to extract the inflammation related genes that showed significant changes in expression in our RNA-Seq analysis. Genes with significant changes in expression in the RNA-Seq experiment were also compared to lists of genes associated with Crohn’s disease , inflammatory bowel disease and ulcerative colitis .
Availability of supporting data
The raw sequencing data and processed data have been deposited in NCBI’s GEO database. The RNA-Seq experiments involving the co-incubation of Caco-2 cells with B. infantis and B. breve are deposited as accessions GSE63950 and GSE64017, respectively.
Human milk oligosaccharides
Intestinal epithelial cells
German JB, Freeman SL, Lebrilla CB, Mills DA. Human milk oligosaccharides: evolution, structures and bioselectivity as substrates for intestinal bacteria. Nestle Nutr Workshop Ser Pediatr Program. 2008;62:205–18. discussion 218–222.
German JB, Dillard CJ, Ward RE. Bioactive components in milk. Curr Opin Clin Nutr Metab Care. 2002;5(6):653–8.
Kunz C, Rudloff S. Health promoting aspects of milk oligosaccharides. Int Dairy J. 2006;16(11):1341–6.
Severin S, Xia WS. Milk biologically active components as nutraceuticals: Review. Crit Rev Food Sci. 2005;45(7–8):645–56.
Casado B, Affolter M, Kussmann M. OMICS-rooted studies of milk proteins, oligosaccharides and lipids. J Proteomics. 2009;73(2):196–208.
Engfer MB, Stahl B, Finke B, Sawatzki G, Daniel H. Human milk oligosaccharides are resistant to enzymatic hydrolysis in the upper gastrointestinal tract. Am J Clin Nutr. 2000;71(6):1589–96.
Gustafsson A, Kacskovics I, Breimer ME, Hammarstrom L, Holgersson J. Carbohydrate phenotyping of human and animal milk glycoproteins. Glycoconj J. 2005;22(3):109–18.
Cleary TG. Human milk protective mechanisms. Adv Exp Med Biol. 2004;554:145–54.
Bode L, Beermann C, Mank M, Kohn G, Boehm G. Human and bovine milk gangliosides differ in their fatty acid composition. J Nutr. 2004;134(11):3016–20.
Pacheco AR, Barile D, Underwood MA, Mills DA. The impact of the milk glycobiome on the neonate gut microbiota. Ann Rev Anim Biosci. 2015;3:419–45.
Harmsen HJ, Wildeboer-Veloo AC, Raangs GC, Wagendorp AA, Klijn N, Bindels JG, et al. Analysis of intestinal flora development in breast-fed and formula-fed infants by using molecular identification and detection methods. J Pediatr Gastroenterol Nutr. 2000;30(1):61–7.
Moro M. Biologische Untersuchungen der Darmbakterien des Säuglings. Jahrb. f, Kinderheilk III Folge. 1905. p. 11.
Gyorgy P, Norris RF, Rose CS. Bifidus factor. I. A variant of Lactobacillus bifidus requiring a special growth factor. Arch Biochem Biophys. 1954;48(1):193–201.
Yatsunenko T, Rey FE, Manary MJ, Trehan I, Dominguez-Bello MG, Contreras M, et al. Human gut microbiome viewed across age and geography. Nature. 2012;486(7402):222–7.
Subramanian S, Huq S, Yatsunenko T, Haque R, Mahfuz M, Alam MA, et al. Persistent gut microbiota immaturity in malnourished Bangladeshi children. Nature. 2014;510(7505):417–21.
Garrido D, Dallas DC, Mills DA. Consumption of human milk glycoconjugates by infant-associated bifidobacteria: mechanisms and implications. Microbiology. 2013;159(Pt 4):649–64.
Sela DA, Mills DA. Nursing our microbiota: molecular linkages between bifidobacteria and milk oligosaccharides. Trends Microbiol. 2010;18(7):298–307.
Ward RE, Ninonuevo M, Mills DA, Lebrilla CB, German JB. In vitro fermentation of breast milk oligosaccharides by Bifidobacterium infantis and Lactobacillus gasseri. Appl Environ Microbiol. 2006;72(6):4497–9.
Ruiz-Moyano S, Totten SM, Garrido D, Smilowitz JT, German JB, Lebrilla CB, et al. Variation in consumption of human milk oligosaccharides by infant-gut associated strains of Bifidobacterium breve. Appl Environ Microbiol. 2013;79(19):6040–9.
Locascio RG, Ninonuevo MR, Kronewitter SR, Freeman SL, German JB, Lebrilla CB, et al. A versatile and scalable strategy for glycoprofiling bifidobacterial consumption of human milk oligosaccharides. Microb Biotechnol. 2009;2(3):333–42.
Tuomola E, Crittenden R, Playne M, Isolauri E, Salminen S. Quality assurance criteria for probiotic bacteria. Am J Clin Nutr. 2001;73(2 Suppl):393S–8.
Turroni F, Serafini F, Foroni E, Duranti S, O’Connell Motherway M, Taverniti V, et al. Role of sortase-dependent pili of Bifidobacterium bifidum PRL2010 in modulating bacterium-host interactions. Proc Natl Acad Sci U S A. 2013;110(27):11151–6.
Crociani J, Grill JP, Huppert M, Ballongue J. Adhesion of different bifidobacteria strains to human enterocyte-like Caco-2 cells and comparison with in vivo study. Lett Appl Microbiol. 1995;21(3):146–8.
Candela M, Seibold G, Vitali B, Lachenmaier S, Eikmanns BJ, Brigidi P. Real-time PCR quantification of bacterial adhesion to Caco-2 cells: competition between bifidobacteria and enteropathogens. Res Microbiol. 2005;156(8):887–95.
Kavanaugh DW, O’Callaghan J, Butto LF, Slattery H, Lane J, Clyne M, et al. Exposure of subsp. To milk oligosaccharides increases adhesion to epithelial cells and induces a substantial transcriptional response. PLoS One. 2013;8(6):e67224.
Chichlowski M, De Lartigue G, German JB, Raybould HE, Mills DA. Bifidobacteria isolated from infants and cultured on human milk oligosaccharides affect intestinal epithelial function. J Pediatr Gastroenterol Nutr. 2012;55(3):321–7.
Tanabe S, Kinuta Y, Saito Y. Bifidobacterium infantis suppresses proinflammatory interleukin-17 production in murine splenocytes and dextran sodium sulfate-induced intestinal inflammation. Int J Mol Med. 2008;22(2):181–5.
Sheil B, MacSharry J, O’Callaghan L, O’Riordan A, Waters A, Morgan J, et al. Role of interleukin (IL-10) in probiotic-mediated immune modulation: an assessment in wild-type and IL-10 knock-out mice. Clin Exp Immunol. 2006;144(2):273–80.
LoCascio RG. Glycomic and genetic characterization of the metabolism of human milk oligosaccharides by Bifidobacterium species. Davis, CA: University of Calfornia, Davis; 2009.
Lewis ZT, Totten SM, Smilowitz JT, Popovic M, Parker E, Lemay DG, et al. Maternal fucosyltransferase 2 status affects the gut bifidobacterial communities of breastfed infants. Microbiome. 2015;3:13.
Sela DA, Chapman J, Adeuya A, Kim JH, Chen F, Whitehead TR, et al. The genome sequence of Bifidobacterium longum subsp. infantis reveals adaptations for milk utilization within the infant microbiome. Proc Natl Acad Sci U S A. 2008;105(48):18964–9.
Garrido D, Kim JH, German JB, Raybould HE, Mills DA. Oligosaccharide binding proteins from Bifidobacterium longum subsp. infantis reveal a preference for host glycans. PLoS One. 2011;6(3):e17315.
Ward RE, Ninonuevo M, Mills DA, Lebrilla CB, German JB. In vitro fermentability of human milk oligosaccharides by several strains of bifidobacteria. Mol Nutr Food Res. 2007;51(11):1398–405.
Gleinser M, Grimm V, Zhurina D, Yuan J, Riedel CU. Improved adhesive properties of recombinant bifidobacteria expressing the Bifidobacterium bifidum-specific lipoprotein BopA. Microb Cell Fact. 2012;11:80.
Bentley DR, Balasubramanian S, Swerdlow HP, Smith GP, Milton J, Brown CG, et al. Accurate whole human genome sequencing using reversible terminator chemistry. Nature. 2008;456(7218):53–9.
Ramskold D, Wang ET, Burge CB, Sandberg R. An abundance of ubiquitously expressed genes revealed by tissue transcriptome sequence data. PLoS Comput Biol. 2009;5(12):e1000598.
Dennis Jr G, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, et al. DAVID: database for annotation, visualization, and integrated discovery. Genome Biol. 2003;4(5):3.
Ganguli K, Meng D, Rautava S, Lu L, Walker WA, Nanthakumar N. Probiotics prevent necrotizing enterocolitis by modulating enterocyte genes that regulate innate immune-mediated inflammation. Am J Physiol Gastrointest Liver Physiol. 2013;304(2):G132–41.
Chan KY, Leung KT, Tam YH, Lam HS, Cheung HM, Ma TP, et al. Genome-wide expression profiles of necrotizing enterocolitis versus spontaneous intestinal perforation in human intestinal tissues: dysregulation of functional pathways. Ann Surg. 2014;260(6):1128–37.
Bai AP, Ouyang Q. Probiotics and inflammatory bowel diseases. Postgrad Med J. 2006;82(968):376–82.
Jostins L, Ripke S, Weersma RK, Duerr RH, McGovern DP, Hui KY, et al. Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease. Nature. 2012;491(7422):119–24.
Gillberg L, Varsanyi M, Sjostrom M, Lordal M, Lindholm J, Hellstrom PM. Nitric oxide pathway-related gene alterations in inflammatory bowel disease. Scand J Gastroenterol. 2012;47(11):1283–97.
Underwood MA, Arriola J, Gerber CW, Kaveti A, Kalanetra KM, Kananurak A, et al. Bifidobacterium longum subsp. infantis in experimental necrotizing enterocolitis: alterations in inflammation, innate immune response, and the microbiota. Pediatr Res. 2014;76(4):326–33.
Jeon SG, Kayama H, Ueda Y, Takahashi T, Asahara T, Tsuji H, et al. Probiotic Bifidobacterium breve induces IL-10-producing Tr1 cells in the colon. PLoS Pathog. 2012;8(5):e1002714.
Hardy H, Harris J, Lyon E, Beal J, Foey AD. Probiotics, prebiotics and immunomodulation of gut mucosal defences: homeostasis and immunopathology. Nutrients. 2013;5(6):1869–912.
Bernet MF, Brassart D, Neeser JR, Servin AL. Adhesion of human bifidobacterial strains to cultured human intestinal epithelial cells and inhibition of enteropathogen-cell interactions. Appl Environ Microbiol. 1993;59(12):4121–8.
Kozol RA. Neutrophil recruitment to the gastrointestinal tract. J Surg Res. 1992;53(3):310–5.
Kucharzik T, Walsh SV, Chen J, Parkos CA, Nusrat A. Neutrophil transmigration in inflammatory bowel disease is associated with differential expression of epithelial intercellular junction proteins. Am J Pathol. 2001;159(6):2001–9.
Lu J, Pierce M, Franklin A, Jilling T, Stafforini DM, Caplan M. Dual roles of endogenous platelet-activating factor acetylhydrolase in a murine model of necrotizing enterocolitis. Pediatr Res. 2010;68(3):225–30.
Han XB, Liu X, Hsueh W, De Plaen IG. Macrophage inflammatory protein-2 mediates the bowel injury induced by platelet-activating factor. Am J Physiol Gastrointest Liver Physiol. 2004;287(6):G1220–6.
Dhawan P, Richmond A. Role of CXCL1 in tumorigenesis of melanoma. J Leukoc Biol. 2002;72(1):9–18.
Sumagin R, Robin AZ, Nusrat A, Parkos CA. Transmigrated neutrophils in the intestinal lumen engage ICAM-1 to regulate the epithelial barrier and neutrophil recruitment. Mucosal Immunol. 2014;7(4):905–15.
Gnoth MJ, Rudloff S, Kunz C, Kinne RK. Investigations of the in vitro transport of human milk oligosaccharides by a Caco-2 monolayer using a novel high performance liquid chromatography-mass spectrometry technique. J Biol Chem. 2001;276(37):34363–70.
Chantret I, Barbat A, Dussaulx E, Brattain MG, Zweibaum A. Epithelial polarity, villin expression, and enterocytic differentiation of cultured human colon carcinoma cells: a survey of twenty cell lines. Cancer Res. 1988;48(7):1936–42.
Mahler GJ, Shuler ML, Glahn RP. Characterization of Caco-2 and HT29-MTX cocultures in an in vitro digestion/cell culture model used to predict iron bioavailability. J Nutr Biochem. 2009;20(7):494–502.
Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat Methods. 2008;5(7):621–8.
Huang DW, Sherman BT, Tan Q, Kir J, Liu D, Bryant D, et al. DAVID bioinformatics resources: expanded annotation database and novel algorithms to better extract biology from large gene lists. Nucleic Acids Res. 2007;35(Web Server issue):W169–75.
Loza MJ, McCall CE, Li L, Isaacs WB, Xu J, Chang BL. Assembly of inflammation-related genes for pathway-focused genetic analysis. PLoS One. 2007;2(10):e1035.
Neuman MG. Immune dysfunction in inflammatory bowel disease. Transl Res. 2007;149(4):173–86.
James SP. Prototypic disorders of gastrointestinal mucosal immune function: Celiac disease and Crohn’s disease. J Allergy Clin Immunol. 2005;115(1):25–30.
Franke A, McGovern DP, Barrett JC, Wang K, Radford-Smith GL, Ahmad T, et al. Genome-wide meta-analysis increases to 71 the number of confirmed Crohn’s disease susceptibility loci. Nat Genet. 2010;42(12):1118–25.
Micallef L, Rodgers P. EulerAPE: drawing area-proportional 3-Venn diagrams using ellipses. PLoS One. 2014;9(7):e101717.
This work was supported by the Bill & Melinda Gates Foundation; and the National Institutes of Health awards R21AT006180, R01AT007079 and R01AT008759. D.A.M. acknowledges support as the Peter J. Shields Endowed Chair in Dairy Food Science.
D.A.M. is a co-founder of Evolve Biosystems, a company focused on diet-based manipulation of the gut microbiota.
SW and DAM designed the research. SW conducted the experiment. SW, ARP, DGL and DAM analyzed the data. SW, ARP, DGL, and DAM wrote the paper. All authors read and approved the final manuscript.
A Microsoft Excel file that contains spreadsheets of Caco-2 genes differentially expressed in response to B. infantis grown on GLU, LAC, or HMO. (XLSX 86 kb)
A Microsoft Excel file that contains spreadsheets of Caco-2 genes differentially expressed in response to B. breve grown on GLU, LAC, or HMO. (XLSX 177 kb)