During the last decades the role of the diet in development, as well as in prevention and management, of many diseases has been subjected to intense research. The term ‘functional food’ has been adapted to denote foods that may provide a health benefit beyond basic nutrition (Saris et al. Reference Saris, Asp and Bjorck1998). The oldest and probably best-known functional food products are health-promoting bacteria or probiotics, defined as live microbial dietary supplements that beneficially affect consumers through their effects in the intestinal tract (Roberfroid, Reference Roberfroid2000). At present, probiotics, most often belonging to the genera Lactobacillus and Bifidobacteria, are almost exclusively consumed as fermented dairy products, such as yoghurt or freeze-dried cultures. Several health-related effects associated with the intake of probiotics have been reported in different animal models as well as in human studies (Roberfroid, Reference Roberfroid2000). However, the scientific evidence is still scarce and the mechanisms by which probiotics influence the host organism are only beginning to be explored.
The initial step in the characterization of mechanism of action for functional food products is the identification of a specific interaction between the active component of this food and an effect in the host organism that is potentially beneficial for health. One approach to investigate these interactions is to map the changes in transcription profile of the host organism caused by nutrient intake. Recent development of expression profiling by the use of microarray technology has made it possible to monitor the expression of thousands of genes simultaneously, allowing systematic analysis of complex biological processes and offering an advantage of reducing bias in data collection, compared with the candidate gene-based approaches. In the present report, we have studied the interactions between the intake of two strains of probiotic bacteria and regulation of intestinal gene expression of the host organism. Oligonucleotide microarrays were applied to compare global transcriptional profiles in the distal ileum of mice receiving Lactobacillus paracasei ssp. paracasei F19 (L. F19) or L. acidophilus NCFB 1748 with the control group of mice receiving a placebo product. The effects of the two Lactobacillus strains were evaluated both in germ-free (GF) or normal microflora (NMF) mice. L. F19 as well as L. acidophilus NCFB 1748 has been used in products branded Cultura and Dofilus (Arla Foods, Stockholm, Sweden). Both strains have good survival in the gastrointestinal tract, which is considered an important characteristic for the health-promoting activity of probiotics (Mättö et al. Reference Mättö, Fondén, Tolvanen, von Wright, Vilpponen-Salmela, Satokari and Saarela2006). Previously, L. F19 and L. acidophilus NCFB 1748, in combination with Bifidobacterium lactis Bb12, have been shown to efficiently restore the intestinal microflora during antibiotic treatment (Sullivan et al. Reference Sullivan, Barkholt and Nord2003). The objective of this study was to provide a comparative insight into the molecular mechanisms by which L. F19 and L. acidophilus NCFB 1748 interact with the host organism in a gnotobiotic environment v. in the context of the complex gut microflora.
Materials and methods
Animal experiments
GF and NMF male mice of Swiss Webster strain aged 6–8 weeks (Taconic, Lille Skensved, Denmark) were maintained under a standard 12-h light cycle regime, the relative humidity was between 45–55 %, the temperature was kept at 20°C. The animals had ad libitum access to purified ingredient diet D12450B (Research Diets Inc., New Brunswick, NJ, USA) and water. NMF mice were housed in individually ventilated standard cages. GF mice were housed in gnotobiotic isolators and handled according to established procedures; isolators were daily monitored for sterility by culturing animal faeces and isolator interiors for bacteria. After 1 week of acclimatization, the mice were divided into two test groups, which received L. F19 (n 6) or L. acidophilus NCFB 1748 (n 6) bacteria in acidified ultra high-temperature processed-milk (approximately 5 × 108 colony forming units/ml) and a control group (four in NMF and five in GF), which received acidified milk only. In NMF conditions, the products were administrated twice per d during a 10-d period: the first dosing was performed by oral gavage (1 ml per mouse) while the second dosing was done by sublingual injection (100 μl per mouse). To minimize the risk of compromising the sterility of the gnotobiotic isolators, the number of product administration events was reduced in GF conditions: the products were fed to GF mice once per d in 8 out of 10 d by oral gavage (1 ml per mouse). Following the period of product administration, the mice were killed by cervical dislocation. To avoid diurnal variations, all the mice were killed between 11.00 and 13.00 hours and approximately 4 h after the last dosing of Lactobacillus. The distal part of the ileum (1·5 cm) was excised for RNA extraction. Additional samples of ileum and colon were collected from each mouse for bacteriological analysis. To analyse the presence of L. F19 and L. acidophilus NCFB 1748 in the groups receiving an active product and the non-presence of these strains in the control groups, the samples were assayed on Rogosa agar (Merck, Darmstadt, Germany). Colonies were isolated and the presence of the two probiotic strains was confirmed by randomly amplified polymorphic DNA-PCR using the primers LBC-19 (5′-AGT AGC CAC-3′) and OPA-05 (5′-TGC CGA GCT G-3′) for screening and OPA-02 (5′-TGC CGA GCT G-3′) and OPA-13 (5′-CAG CAC CAC-3′) for confirmation of L. F19 and L. acidophilus NCFB 1748, respectively. Slanetz Bartley agar (Merck) was used for checking for the possible presence of enterococci. To confirm the absence of bacteria other than the administrated strains in ex-GF mice, these samples were spread on blood agar base (Oxoid, Basingstoke, Hants., UK) with defibrinated blood added to a final concentration of 7 %, and the plates were incubated both aerobically and anaerobically. The present study was performed after prior approval from the local ethical committee for animal experimentation.
Test products
Test products were produced from ultra high-temperature processed milk containing 1·5 % fat. Probiotic organisms, L. F19 or L. acidophilus NCFB 1748, were added to a final concentration of 1 × 109 colony forming units/ml and the milk was acidified to pH 4·5 by addition of glucono-δ-lactone to a final concentration of 1·6 %. The placebo product was identical to the active product despite no addition of probiotic organisms. The number of probiotic bacteria in the products was assayed by plating on MRS agar, pH 5·4 (de Man Rogosa Sharp; Oxoid). The absence of contaminating bacteria was confirmed by plating on count agar sugar free FIL-IDF (Fédération internationale de laiterie/International Dairy Federation) (Merck) and on plate count agar (Oxoid) with skimmed milk added to a final concentration of 0·1 %, for the active and placebo products, respectively.
RNA extraction
Tissues collected for RNA preparation were immediately submerged in RNAlater RNA Stabilization Reagent (Qiagen, Hilden, Germany) to preserve the quality and quantity of RNA. Total RNA was isolated from the intestinal tissue samples using the RNeasy Mini Kit (Qiagen) followed by a DNase digestion step (RNase-Free DNase Set; Qiagen) according to the manufacturer's instructions. The RNA yield was quantified by spectrophotometric analysis and the RNA purity was determined based on the A260:A280 ratio. All RNA samples were analysed by agarose gel electrophoresis to check for integrity of 18S and 28S rRNA. Further, the quality of the RNA was verified by Agilent 2100 Bioanalyzer analysis (Agilent Technologies, Palo Alto, CA, USA) using the RNA 6000 Nano Assay Kit (Agilent Technologies).
Expression profiling by Affymetrix
Total RNA spiked with poly-A controls (pGIBS-TRP, –THR and –LYS; American Type Culture Collection, Manassas, VA, USA) was converted into cDNA utilizing a T7 promoter-polyT primer (Genset, Paris, France) and the RT Superscript II (Invitrogen Life Technologies, Paisley, UK) followed by a second strand cDNA synthesis (Invitrogen Life Technologies). ds cDNA was in vitro transcribed into biotinylated cRNA (Enzo Life Sciences, Farmingdale, NY, USA). Finally, fragmented cRNA (35–200 bases) was used as a target for hybridization. Hybridization spike controls (oligonucleotide B2 and a cRNA cocktail (BioB, BioC, BioD and Cre; GeneChip Eukaryotic Hybridization Control Kit; Affymetrix, Santa Clara, CA, USA)) were used as hybridization quality controls. Aliquots of each sample were hybridized (16 h at 45°C) to the Murine Genome Array U74Av2 (Affymetrix). The arrays were subsequently washed, stained and scanned according to the manufacturer's instructions (GeneChip Expression Analysis Technical Manual; Affymetrix). Data were analysed using robust multi-chip analysis in GeneTraffic UNO version 2.6–25 (Stratagene, La Jolla, CA, USA) and Spotfire DecitionSite for Functional Genomics version 7.1 (Spotfire Inc., Göteborg, Sweden). The log2 fold change for each sample and probe set v. the control group was calculated using the formula: log2 fold change = log2 (probe set intensity/mean probe set intensity in the control group). The mean log2 fold change was calculated for each Lactobacillus strain v. the control group. Statistical significance of the difference in gene expression was determined using two-sided Student's t test. A transcript was considered differentially expressed if the mean absolute log2 fold change was >0·5 (corresponding to a mean absolute fold change >1·41) and the P-value was < 0·05. In addition, the mean intensity in the group showing the highest expression should be >75. To view the microarray data in a biological context, Ingenuity Pathway Analysis Software tool (Ingenuity Systems, Redwood City, CA, USA; http:\\www.ingenuity.com) was used to generate a network connecting the differentially regulated targets and all other gene products, based on known mammalian gene product events (such as protein–protein, protein–nucleic acid interactions) defined in Ingenuity System's database.
Expression profiles of genes differentially regulated in GF mice in response to Lactobacillus administration were compared with the corresponding profiles in GF mice of the NMRI/KI strain mono-colonized by Bacteroides thetaiotaomicron, using the data published by Hooper et al. (Reference Hooper, Wong, Thelin, Hansson, Falk and Gordon2001). The .CEL files from this study (Mu11K and Mu19K Affymetrix array sets, duplicate microarray hybridizations performed on pooled ileal RNA samples corresponding to GF v. mono-colonized mice) kindly provided by Professor L.V. Hooper, were re-analysed using Microarray Suite Version 5.0 (MAS5.0) (Affymetrix) and Spotfire DecitionSite for Functional Genomics version 7.1 (Spotfire Inc.). The overall intensity across each array was scaled to a target intensity of 150. A transcript was considered differentially expressed if the mean absolute log2 fold change was >0·5 (corresponding to a mean absolute fold change >1·41). Additionally, the mRNA should be called Present by the MAS5.0 software in either GF or colonized mice, and the differential expression should be observed in all four comparisons performed on duplicate microarray hybridizations (see Hooper et al. (Reference Hooper, Wong, Thelin, Hansson, Falk and Gordon2001) for details concerning the design of the microarray experiment).
Quantitative real-time PCR
Expression profiles for selected targets were confirmed using quantitative real-time PCR with the ABI PRISM 7000 Sequence Detector System (Applied Biosystems, Foster City, CA, USA), according to the manufacturer's instructions. Briefly, total RNA, isolated for the expression profiling, was converted into cDNA by utilizing the Superscript™ first-strand synthesis system (Invitrogen Life Technologies). PCR reactions (25 μl) contained each PCR primer (400 nm) designed by using PRIMER EXPRESS 2.0 software (Applied Biosystems) and 1x SYBR® Green PCR Master Mix (Applied Biosystems). The forward and reverse primer sequences are presented in Table 1. All reactions were performed in duplicate and a dissociation curve was completed for every PCR run to control the specificity of the amplification reaction. The relative quantities of different mRNA transcripts were calculated after normalization of the data against an endogenous control – acidic ribosomal phosphoprotein P0 (Arbp) – using the standard curve method (Applied Biosystems, 1997).
Primer sequences used for quantitative real-time-PCR analysis
Results
Microarray analysis of the differential gene expression evoked by oral administration of Lactobacillus strains
We have used oligonucleotide microarray to analyse the host transcriptional responses caused by oral delivery of Lactobacillus bacteria. The test groups of mice received one strain of Lactobacillus – L. F19 or L. acidophilus NCFB 1748 – in acidified milk, while the control group of mice received acidified milk only. Experiments were performed using age-matched male mice raised under GF or NMF conditions. After the period of product administration of 10 d, the presence of the corresponding Lactobacillus strains in the intestinal contents of the test groups of mice, and the lack of these bacteria in the control group, were confirmed. Importantly, from the intestine of the test groups of mice raised under GF conditions, only the administrated strain could be re-isolated, while the corresponding control group remained GF during the whole study period. While previous studies have shown that the different strains of Lactobacillus bacteria are able to colonize the whole gastrointestinal tract of GF mice and are established in high numbers in both the small and large intestine (Wagner et al. Reference Wagner, Warner, Roberts, Farmer and Balish1997; Ibnou-Zekri et al. Reference Ibnou-Zekri, Blum, Schiffrin and von der Weid2003), we found that L. F19 and L. acidophilus NCFB 1748 were present in the ileum of the GF mice in considerably larger numbers than compared with the colon (data not shown). The gene expression profiles in the distal ileum of the test groups of mice were compared with the corresponding control group using Affymetrix gene arrays.
The global gene expression analysis demonstrated that genes altered by administration of Lactobacillus are involved in widely different functions. In the NMF mice, the expression of twenty-two probe sets was significantly (P < 0·05) changed by a factor of more than 41 % (absolute log2 fold change >0·5), in at least one test group, relative to the control group (Table 2). Administration of Lactobacillus to GF mice led to the differential expression of thirty probe sets in one or both test groups, relative to control mice (absolute fold change >1·41, P < 0·05; Table 3). From these transcripts, the identity of most of the genes is known and represents proteins of different functional classes, whereas three transcripts only show homology to sequences in the Expressed sequence tag (EST) or genomic databases. The vast majority of genes, which were significantly changed exclusively in mice receiving L. F19 or L. acidophilus NCFB 1748, tended to be regulated in the same manner in the other test group, even though this difference did not reach statistical significance and/or meet the fold change criteria. Importantly, transcriptional changes evoked by administration of Lactobacillus to NMF mice were different from the responses seen in GF mice. In fact, only one probe set, representing retinol binding protein 2, showed significant difference in response to the delivery of Lactobacillus bacteria both to NMF and GF mice (Tables 2 and 3).
Differentially expressed genes in normal microflora mice receiving Lactobacillus paracasei ssp. paracasei F19 (F19) or Lactobacillus acidophilus NCFB 1748 (NCFB) compared with the control group of mice receiving placebo product‡§
Values were significantly different, determined by two-sided Student's t test: (*P ≤ 0·05; **P ≤ 0·01).
† Transcripts were identified by similarity search using mean expression profile of Clu, C3, Bcl6, Ptprc, Serpina1, Laptm5 and Vcam1 as a template.
‡ Global mRNA expression pattern was characterized in distal ileum using Affymetrix (Santa Clara, CA, USA) gene arrays. The filtering criteria were set to a mean absolute fold change >1·41 (log2 fold change >0·5) and a P value < 0·05 in either of the two test groups, compared with the control group. In addition, the mean intensity in the group showing highest expression should be >75.
§ For details of procedures, see Materials and methods.
The colours represent the differential expression pattern. Green indicates down-regulation; red indicates up-regulation with an absolute log2 fold change >0·5. BCR, B cell receptor.
Differentially expressed genes in germ-free mice receiving Lactobacillus paracasei ssp. paracasei F19 (F19) or Lactobacillus acidophilus NCFB 1748 (NCFB) compared with the control group of mice receiving placebo product†‡
Values are significantly different, determined by two-sided Student's t test (*P ≤ 0·05; **P ≤ 0·01; ***P ≤ 0·005).
† Global mRNA expression pattern was characterized in distal ileum using Affymetrix (Santa Clara, CA, USA) gene arrays. The filtering criteria were set to a mean absolute fold change >1·41 (log2 fold change >0·5) and a P value < 0·05 in either of the two test groups, compared with the control group. In addition, the mean intensity in the group showing highest expression should be >75.
‡ For details of procedures, see Materials and methods.
The colours represent the differential expression pattern. Green indicates down-regulation; red indicates up-regulation with an absolute log2 fold change >0·5. CREB, cyclic AMP-response element-binding; HLA, histocompatibility locus antigen.
Quantitative real-time PCR validation of differentially expressed genes
To minimize erroneous conclusions due to technical variability and multiple testing effects inherent to the microarray technology, quantitative real-time PCR analysis was applied to validate expression profiles of eleven genes selected on the basis of biological interest (Fig. 1). For all the genes examined, quantitative real-time PCR data are in good agreement with the gene array results with regard to the direction of observed changes. Furthermore, individual animal-to-animal comparison of the expression profiles for these genes showed good correlation comparing the two techniques (data not shown).
Expression analysis by quantitative real-time PCR in (A) normal microflora and (B), (C) germ-free mice receiving Lactobacillus paracasei ssp. F19 (L. F19; 
) or Lactobacillus acidophilus NCFB 1748 (L. NCFB 1748; ■) compared with the control group of mice (□) receiving placebo product. The relative quantities of different mRNA transcripts were calculated after normalization of the data against an endogenous control – acidic ribosomal phosphoprotein P0 (Arbp). The results are shown as the fold change of expression in two test groups relative to the corresponding control group, with the expression level in the control groups set to 1 (A),(B) or − 1 (C). Quantitative real-time PCR analysis was performed on the same set of samples that was used in the gene array experiment. Statistical differences were determined by two-sided Student's t test: *P ≤ 0·05; #P < 0·06. For details of procedures, see Materials and methods.
Administration of Lactobacillus to normal microflora mice modulates the immune phenotype expression
Several genes with a potential role in the intestinal immune system – Clu, C3, Bcl6, Ptprc, Serpina1, Laptm5 and Vcam1 – were significantly up-regulated in the NMF mice receiving L. acidophilus NCFB 1748 and showed identical transcription profiles. Searching for similar expression patterns using Profile Search (Spotfire DecitionSite for Functional Genomics version 7.1; Spotfire Inc.) with correlation as similarity measure (0·92 as cut off) identified thirty-five additional probe sets up-regulated by a factor of more than 41 % in at least one test group, compared with the control mice (Table 2). Ingenuity Pathway Analysis (Ingenuity Systems) reveals that these targets form an integrated functional network controlling different aspects of immune system development and function and are regulated by common upstream factors (Fig. 2(A)). Most of the transcripts in this cluster are expressed predominantly or exclusively in B cells, e.g. the components of the B cell receptor for antigen including Ig molecule (encoded by Ighs), Cd79a and Cd79b, as well as factors involved in downstream signalling from the B cell receptor, such as Ptprc (alias Cd45), Cd19, Blk, Lck and Prkcb1. A key mediator responsible for the organization of B cells within lymphoid structures, Ltb, together with three additional markers potentially involved in this function, Cxcl13, Vcam1 and Bcl6, were up-regulated in response to Lactobacillus administration. Additionally, genes with potential function in phagocytosis (Mfge8 and Coro1a) and complement function (C3 and Clu) were increased in the test groups. Several probe sets supported an up-regulation in Serpina1 transcript, encoding for α1-antitrypsin, in response to Lactobacillus administration.
The Affymetrix Probe Set IDs (Santa Clara, CA, USA) for genes listed in Tables 2 and 3 were imported into the Ingenuity Pathway Analysis software (Ingenuity Systems, Redwood City, CA, USA), to generate interactome and transcriptional networks connecting the query genes and all other gene products. The networks generated by the Ingenuity Pathway Analysis tool are ranked by score reflecting on how relevant they are to the genes within the dataset. For normal microflora dataset, the two networks with the highest score contained a number of common focus genes and were merged into a single network (A). For germ-free dataset, the network with the highest score containing fourteen focus genes is presented (B). Red and green nodes are input genes with red indicating up-regulation and green indicating down-regulation in mice receiving Lactobacillus compared with the control group of mice receiving placebo product (Tables 2 and 3). The white nodes indicate genes not part of the dataset file. The common biological function for selected groups of nodes is shown. Some peripheral nodes and connection genes were removed for simplicity. The displayed network features only genes with functional interactions. BCR, B cell receptor. For details of procedures, see Materials and methods.
Administration of Lactobacillus strains in germ-free mice evokes coordinated changes in expression profiles of genes regulating energy homeostasis and host defence
Three transcripts encoding for key regulators of fat and sugar metabolism and insulin sensitivity at the whole body level – adipsin (Cfd), adiponection (Adipoq) and resistin like β (Retnlb) – were differentially expressed in the test groups of mice compared with the control group (see Table 3 for statistical significance notes). Adipsin and adiponectin, an insulin-sensitizing hormone (Pajvani & Scherer, Reference Pajvani and Scherer2003), were up-regulated, while resistin like β, known to induce insulin resistance (Rajala et al. Reference Rajala, Obici, Scherer and Rossetti2003), was down-regulated, in response to Lactobacillus administration. Transcripts for three cytosolic proteins, Scd1, Car3 and Thrsp, with possible functions in lipid metabolism, were significantly up-regulated in the groups of mice receiving Lactobacillus (Table 3). Stearoyl-Coenzyme A desaturase 1 is an Fe-containing enzyme that catalyses a rate-limiting step in the synthesis of unsaturated fatty acids. Although very little is known about the specific function of Car3 and Thrsp, both genes are implicated in fatty acid metabolism (Stanton et al. Reference Stanton, Ponte, Coleman and Snyder1991; Kinlaw et al. Reference Kinlaw, Church, Harmon and Mariash1995). Additionally, several genes with a potential role in intestinal defence against bacterial infections – Mmp7, Lyzs, Pla2g2a, Sprr1a, Igh6 – were up-regulated in response to Lactobacillus administration in GF mice. Ingenuity Pathway Analysis (Ingenuity Systems) indicated that the differentially expressed targets involved in regulation of energy homeostasis as well as host defence form a complex network regulated by common upstream mediators, with TNF having a central regulatory role (Fig. 2(B)).
To evaluate if the changes in gene expression profile we describe are specific to Lactobacillus or can be elicited by inoculation of the gnotobiotic intestine with non-probiotic bacteria as well, we turned to the previously published study describing the effects of colonization with Bacteroides thetaiotaomicron, a prominent component of the normal intestinal microflora (Hooper et al. Reference Hooper, Wong, Thelin, Hansson, Falk and Gordon2001). In this report, the global gene expression profile in the distal ileum of mice inoculated with Bacteroides was compared with the age-matched mice remaining GF by Affymetrix microarrays, thus providing a good comparison to our dataset. We compared the transcription profiles of genes differentially regulated in response to Lactobacillus (Table 3) with the corresponding profiles in mice colonized by Bacteroides, using the raw data files kindly provided by Professor L.V. Hooper (see Materials and Methods for details on data analysis). The probe sets for several genes involved in regulation of energy homeostasis (Adipoq, Retnlb, Scd1) were represented and called Present on these arrays; however, none of these transcripts was regulated in response to Bacterioides in the manner similar to Lactobacillus. On the contrary, two probe sets supported an up-regulation of Retnlb by Bacteroides, while this gene was down-regulated in response to Lactobacillus. Interestingly, Rbp2, encoding a protein shown to participate in uptake and metabolism of vitamin A in the small intestine (Levin, Reference Levin1993; Lissoos et al. Reference Lissoos, Davis and Levin1995), was up-regulated in response to Bacteroides as well as Lactobacillus. Similar to Lactobacillus, administration of Bacteroides increased expression of genes implicated in host defence, including small proline like proteins (Sprr2a up-regulated by Bacteroides and Sprr1a by Lactobacillus) and Pla2g2a (data not shown).
Discussion
In the present study, we investigated the molecular effects of the two Lactobacillus strains – L. F19 and L. acidophilus NCFB 1748 – on global gene expression profile in the distal small intestine using the Affymetrix gene arrays. The changes in transcriptional profile caused by oral administration of Lactobacillus bacteria in NMF mice were compared with the effects evoked by administration of these strains in GF mice, with the GF mice representing a simplified model of interactions between gut commensals and their host.
Lactobacilli have long been acknowledged to promote the intestine's immunological barrier, particularly through enhancement of humoral immune responses, induction of germinal centre formation, activation of the phagocytosis by macrophages and alleviation of intestinal inflammatory responses (Wagner et al. Reference Wagner, Warner, Roberts, Farmer and Balish1997; Perdigon et al. Reference Perdigon, Vintini, Alvarez, Medina and Medici1999; Shu & Gill, Reference Shu and Gill2002; Ibnou-Zekri et al. Reference Ibnou-Zekri, Blum, Schiffrin and von der Weid2003; Galdeano & Perdigon, Reference Galdeano and Perdigon2006). Recent data indicate that differences may exist in the immune stimulatory effects of specific strains of probiotic bacteria (Perdigon et al. Reference Perdigon, Vintini, Alvarez, Medina and Medici1999; Ibnou-Zekri et al. Reference Ibnou-Zekri, Blum, Schiffrin and von der Weid2003). In line with the previous reports, the present study demonstrated that the administration of two Lactobacillus strains to NMF mice caused concerted increases in a cluster of genes involved in immune response (Table 2, Fig. 2(A)). Several components of B cell receptor-signalling were up-regulated (Ighs, Cd79a, Cd79b, Ptprc (alias Cd45), Cd19, Blk, Lck, Prkcb1), suggesting mobilization of B-lymphocytes. Also, transcripts implicated in phagocytosis (Mfge8, Coro1a), complement function (C3, Clu) and architectural organization of B cells within lymphoid structures (Ltb, Cxcl13, Vcam1 and Bcl6) were increased in response to Lactobacillus administration. Previously, Di Caro et al. (Reference Di Caro, Tao, Grillo, Elia, Gasbarrini, Sepulveda and Gasbarrini2005) demonstrated by global gene expression profiling in duodenal mucosa that administration of L. rhamnosus to human subjects induced expression of a number of genes involved in immune response, including Ltb, Cxcl13, C3 and Ms4a1, which were also up-regulated by the two strains used in the present study. We did not detect any qualitative differences comparing the effect of L. F19 v. L. acidophilus NCFB 1748 on the expression profile of immune response-related genes. However, the mean fold change of the increased signal for this group of transcripts was higher in mice receiving L. acidophilus NCFB 1748. Notably, immune stimulatory effect in response to Lactobacillus administration was not observed in mice raised under GF conditions. The intestinal microflora has a large impact on the development of gut-associated lymphoid tissue. Also, the response to probiotic bacteria depends on the immunological state of the host (Falk et al. Reference Falk, Hooper, Midtvedt and Gordon1998). Therefore, we speculate that under the conditions used, the intestinal tissue of gnotobiotic mice might have been incompetent to respond to the stimulation by lactic acid bacteria as observed in NMF mice.
Lactobacillus bacteria have been shown to improve clinical symptoms of inflammatory bowel disease, a major clinical problem in the field of gastroenterology (Matsumoto et al. Reference Matsumoto, Watanabe, Imaoka and Okabe2001; Saggioro, Reference Saggioro2004). In this context, it is interesting to note that Serpina1, encoding α1-antitrypsin, was up-regulated in the NMF mice in response to Lactobacillus administration (Table 2). The concentration of α1-antitrypsin is increased in the intestine in connection with inflammatory bowel disease, where its release is thought to protect the mucosa from proteolytic damage (Faust et al. Reference Faust, Raschke, Hormann, Milovic and Stein2002). We hypothesize that increase in Serpina1 transcription may be part of the mechanism for Lactobacillus bacteria to reduce the severity of inflammatory bowel disease, which remains to be addressed in further studies.
Recently, the role of factors secreted from the gut in regulation of food intake and whole body energy partitioning has been acknowledged (Bloom et al. Reference Bloom, Wynne and Chaudhri2005). Also, connection between gut microflora and energy homeostasis of the host organism has become recognized. Colonization of GF mice with normal gut microflora was shown to increase body fat and cause insulin resistance (Backhed et al. Reference Backhed, Ding, Wang, Hooper, Koh, Nagy, Semenkovich and Gordon2004). Conversely, obesity affects the composition of the gut microbiota in mice (Ley et al. Reference Ley, Backhed, Turnbaugh, Lozupone, Knight and Gordon2005). Interestingly, consumption of dairy products supplemented with Lactobacillus bacteria has been shown to decrease serum cholesterol and LDL-cholesterol (Akalin et al. Reference Akalin, Gonc and Duzel1997). However, the mechanisms responsible for directing these changes remain largely unknown. In this context, one of the most interesting findings of the present study was the coordinated differential regulation of transcripts for several secreted factors controlling whole body lipid and glucose metabolism, in response to the administration of Lactobacillus in GF mice. Adiponectin (Adipoq) and adipsin (Cfd), both up-regulated in the test groups of mice, are known to be decreased in overweight human subjects and/or in animal models of obesity (Flier et al. Reference Flier, Cook, Usher and Spiegelman1987; Lowell et al. Reference Lowell, Napolitano, Usher, Dulloo, Rosen, Spiegelman and Flier1990; Shillabeer et al. Reference Shillabeer, Hornford, Forden, Wong, Russell and Lau1992; Hu et al. Reference Hu, Liang and Spiegelman1996; Yamauchi et al. Reference Yamauchi, Kamon and Waki2001). Interestingly, pharmacological adiponectin treatment in rodents has been shown to increase insulin sensitivity (Pajvani & Scherer, Reference Pajvani and Scherer2003). Resistin like β (Retnlb), on the other hand, was down-regulated in the test groups of mice. Recently, elevated serum levels of resistin like β, attributable to increased mRNA and protein production in intestine and bone marrow, were reported in mice fed a high-fat diet as well as in animal models of type 2 diabetes (Shojima et al. Reference Shojima, Ogihara and Inukai2005). Infusion of resistin like β has been shown to induce insulin resistance (Rajala et al. Reference Rajala, Obici, Scherer and Rossetti2003). In the light of this knowledge, an increased signal for adiponectin and adipsin in combination with reduced expression of resistin like β, observed in GF mice in response to the delivery of Lactobacillus (Table 3), suggests the possibility for improved insulin sensitivity of the host organism. The present report does not answer the question if similar host responses are elicited by other components of the gut microflora. Notably, colonization of GF mice with normal gut microflora has been shown to decrease, rather than increase, insulin sensitivity (Backhed et al. Reference Backhed, Ding, Wang, Hooper, Koh, Nagy, Semenkovich and Gordon2004). One factor suppressed in the ileum in response to conventionalization of gnotobiotic mice, implicated in the promotion of adiposity and insulin resistance, is Fiaf (Backhed et al. Reference Backhed, Ding, Wang, Hooper, Koh, Nagy, Semenkovich and Gordon2004). Probe sets for Fiaf were represented on the Affymetrix microarray used in this study. However, the mRNA level for this gene was unaltered in response to Lactobacillus administration (data not shown). Interestingly, mono-colonization of GF mice with common gut bacterium Bacteroides thetaiotaomicron (Hooper et al. Reference Hooper, Wong, Thelin, Hansson, Falk and Gordon2001) did not evoke effects similar to Lactobacillus on the expression of genes involved in regulation of energy homeostasis. This suggests that the gut microbiota influence the expression of genes important for energy metabolism differently depending on the microbial composition and that Lactobacilli may influence transcription of genes involved in regulation of energy homeostasis in a favourable way. However, notice should be taken that different strains of GF mice have been used in the reports compared (C57BL/6J, NMRI/KI or Swiss Webster strain was used by Backhed et al. Reference Backhed, Ding, Wang, Hooper, Koh, Nagy, Semenkovich and Gordon2004, Hooper et al. Reference Hooper, Wong, Thelin, Hansson, Falk and Gordon2001 or in the present study, respectively) and, therefore, we cannot exclude the possibility that the differences in the transcriptional changes described in response to the different bacterial strains are at least partly related to the variation in the genetic background of the host.
Adiponectin, adipsin and resistin like β were differentially regulated in response to Lactobacillus administration in GF but not in NMF mice. The gut microbiota are known to regulate the intestinal motility, differentiation, assembly of gut-associated lymphoid tissue, etc., resulting in substantial differences comparing the morphology and function of gnotobiotic v. NMF intestine (Falk et al. Reference Falk, Hooper, Midtvedt and Gordon1998). Additionally, Lactobacillus bacteria have been shown to colonize the intestine of both GF and NMF mice (Bateup et al. Reference Bateup, McConnell, Jenkinson and Tannock1995; Filho-Lima et al. Reference Filho-Lima, Vieira and Nicoli2000). However, the colonization efficiency might differ comparing the gnotobiotic v. NMF gut. Therefore, it is not surprising that the changes in gene expression pattern caused by Lactobacillus administration in GF mice differ from the responses in NMF mice. Further studies are required to clarify if the differential expression of Adipoq, Cfd and Retnlb observed in GF mice can be evoked in NMF conditions applying different product administration regimes and/or dose of probiotics. Also, the influence of genetic background and/or species differences of the host on the effect of Lactobacillus administration remains to be addressed.
Gene array approaches are limited by multiple comparison caveats and, therefore, a certain number of false–positive results can be expected. Also, several of the alterations in gene expression that we describe are of relatively low magnitude and of marginal significance. However, the biological relevance of the differentially regulated targets highlighted in the present report is supported by the fact that a number of genes from the same biological pathway are coordinately changed as illustrated by Ingenuity Pathway Analysis (Fig. 2). Furthermore, the two genetically close Lactobacillus strains used in the study evoke similar changes in the expression profile of the genes belonging to these functional groups, which further emphasizes the biological interpretation of the data. Finally, the differential expression of selected targets was confirmed by an alternative approach, quantitative real-time PCR, which largely supported the conclusions from the gene array experiments (Fig. 1).
In summary, the present study characterizes global transcriptional responses to administration of two probiotic strains – L. F19 and L. acidophilus NCFB 1748 – in vivo. The results reveal that Lactobacillus bacteria modulate expression of host genes participating in different fundamental physiological processes in the intestine. Several changes in gene expression that we describe were expected, being consistent with the earlier publications in the field, e.g. stimulation of the immune system in NMF mice and increased expression of host defence markers in GF mice. Additionally, we report differential expression of several genes previously not known to be regulated by Lactobacillus, such as transcripts involved in regulation of energy homeostasis. The differential response in NMF v. GF mice that we describe underlines the fact that host–bacterial interactions are both complex and dynamic, and any impact of Lactobacillus feeding is likely to be affected by factors such as age, sex, health status, already existing gut microflora, etc. Consequently, additional molecular and physiological studies are required to characterize the functional impact of the changes in transcriptional profile presented in the present report.
Acknowledgements
The study was supported by Arla Foods, Innovation and Arexis Research and Development Fund. The authors are grateful to Professor Lora V. Hooper and Professor Jeffrey I. Gordon for kindly providing an access to the raw data files from their previously published study (Hooper et al. Reference Hooper, Wong, Thelin, Hansson, Falk and Gordon2001). Dr Robert Brooksbank (Cambridge Biotechnology Ltd.) is acknowledged for generating the Ingenuity Pathway Analysis maps.
