Iron is an essential micronutrient for most organisms and is taken up by high affinity transport systems [1]. Two oxidized forms of iron, ferric (Fe³⁺) and ferrous iron (Fe²⁺), occur in nature with ferrous iron being to be most abundant in the intestine [2]. Iron is limited in most environments and a battle for iron occurs in many microbial ecosystems, including the human gastro-intestinal (GI) tract [3, 4]. Withholding iron is therefore a competitive and defense trait in many Gram-positive and Gram-negative bacteria [5, 6]. Predominant members of an ecosystem frequently possess efficient iron-scavenging systems which enable them to outcompete other microorganisms by depriving them from iron [7]. In addition, restricting iron to pathogens in the GI-tract has been coined as nutritional immunity phenomenon and is usually associated with efficient iron-sequestration systems.

Bifidobacteria are non-sporeforming, nonmotile, anaerobic, Gram-positive bacteria with a high G + C content that are residents of the GI-tract and predominant in infants [8]. Bifidobacteria have been shown to beneficially modulate the composition and activity of the intestinal microbiota, to prevent bacterial infections, and to exert anti-inflammatory and immunomodulation activities [9]. These beneficial traits depend strongly on the ability of a strain to survive and to adapt in the GI-tract [10]. Adaptation and survival involve the use of efficient and diverse nutrient uptake systems, enzymes, stress proteins, and factors that interact with the host and with other members of the gut microbiota [11].

Bifidobacteria require iron for growth and produce extracellular binding proteins that are involved in iron uptake [12]. Such iron-binding proteins may also be implicated in iron withholding thereby limiting the availability to pathogens in the GI-tract [13]. Additionally, the essential micronutrient zinc might have a similar role in ecosystems as iron, and mechanisms for zinc sequestration have been recently reported to further contribute to nutritional immunity by similar mechanisms as iron sequestration [14–16].

Bacterial extracellular proteins are either actively transported through the cytoplasmic membrane into the environment or simply shed from the surface. The composition of the extracellular proteome, also known as the secretome, strongly depends on the nutrient preferences of the bacterium [17]. Bifidobacterial secretomes have been used to study diversity and physiology of the genus, and to identify differences in nutrient uptake and stress response [18, 19]. Extracellular proteins of bifidobacteria are pivotal for host interactions and -adaptations, for nutrient uptake, adhesion, and stress sensing [20–24]. Further, extracellular proteins of bifidobacteria are potentially directly involved in mechanisms beneficial to the host [25], and the secretome is therefore an ideal target to understand interactions and responses of bifidobacteria in the GI-tract.

Functional genomics is powerful to identify bifidobacterial mechanisms active in the gut, such as genes involved in host-microbe interactions, the degradation of human milk oligosaccharides, or pili encoding genes [10, 23]. Combined genomic and proteomic analyses of bifidobacteria revealed mechanisms of adaptation to the GI-tract and genetic functions that mediate specific host-microbe and microbe-microbe interactions [26, 27]. In contracts to genomics, proteomic detects the functional gene products that are present under specific conditions [28, 29]. Proteomic also allows comparison of strain specific features under similar conditions and is therefore well suited for studying features which may not be accessed by genomics [28]. However, analyses of proteomic mass spectra (MS) rely upon homology with pre-established protein sequences derived from genus related databases [30]. Such approach will inevitably fail to identify strain-unique protein sequences [22] and because many bifidobacterial properties are strain specific, homology-driven proteomics has limits to identify important phenotypic features [19]. A combined genomic-proteomic approach in which the genome is sequenced and used to build a protein database is therefore more suitable for the analyses of the proteomic MS data form bifidobacteria.

Bifidobacterium kashiwanohense PV20–2 and Bifidobacterium pseudolongum PV8–2 were isolated from the feces of breast-fed, anemic Kenyan infants. These strains exhibit high iron sequestration mechanisms and their whole genomes have been sequenced [12, 31, 32]. In this study, the genomes of B. kashiwanohense PV20–2 and B. pseudolongum PV8–2 were compared to other bifidobacterial genomes to identify genes potentially involved in iron metabolism. Further, we identified the secretome of both strains under iron limiting conditions using a combined genomic-proteomic approach. In this approach, the predicted coding sequences were used to identify MS/MS-peptides obtained via a 1D gel-based shotgun proteomic approach.

In this study we analyzed two bifidobacteria strains that had the highest iron sequestration capacity among 56 bifidobacterial isolates from stools of iron - deficient Kenyan infants [12]. Using a comparative genomics and a combined proteomic-genomic approach we identified in both strains specific adaptations to the low iron concentration encountered in the GI-tract of iron-deficient infants.

The comparative genome analyses revealed adaptations to low iron conditions and to a specific diet, thus the genomic content reflects the source of the strains, as already proposed previously for bifidobacteria [36]. B. kashiwanohense PV20–2 encodes genes putatively involved in bacillibactin and pyoverdines syntheses that were not present in strain DSM 21854. Bacillibactins and pyoverdines are both siderophores [34, 35] and the corresponding genes in PV20–2 could encode for the synthesis of a bifidobacterial siderophore. The identification in the secretome of an ABC-transporter genetically linked to the putative siderophore genes suggest that this transporter is involved in the uptake of the iron-siderophore complex. In the B. pseudolongum PV8–2 strain, iron related genome content included a ferric iron transporter clearly inserted in the genome (Fig. 2). The GC-content of 59.4% of the insertion suggests it has been acquired from a high-GC-content bacterium. It is, however, not known when the insertion occurred and it is therefore also possible that the sequence is already adapted to the host chromosome for its GC-content. The origin of this insertion is therefore not clear. The presence of a siderophore operon in strain PV20–2 and an additional iron transporter in PV8–2 seems directly related to the selective pressure mediated by the low-iron abundance in the gut of the iron-deficient infants.

Arabinogalactan are pectin derived sugars that can be utilized by some bifidobacteria [37] and the corresponding genes have been described in B. breve [38]. Arabinogalactan is found in plant cell wall and is a potential prebiotic with some selectivity for bifidobacteria [39] The occurrence of an arabinogalactan machinery in the human isolate PV8–2, and not in the rumen isolates DSM 20092 and AGR 2145 might be related to the specific diet of the Kenyan infant from whom feces the strain was isolated. The infant weaning diet is composed of mother’s milk and maize porridge cooked in water, and therefore containing arabinogalactan-rich source [40].

Approximately 75% of the CDSs that were unique for PV8–2 and PV20–2 in the direct comparison to strains of their own species were found in other bifidobacteria. This suggests that the corresponding genes have been acquired from other bifidobacteria and that horizontal gene transfer (HGT) occurs between members of the phylum. In addition, the ferric iron operon of PV20–2 is inserted in a conserved genome region and likely acquired via HGT. The high cell densities and strong selective pressure in the GI-tract facilitate HGT and outgrowth of novel variants [41, 42]. In addition, costs of gene transfer are lower if the genes fit well in the recipient’s system and transfer occurs therefore easier between related species [43]. The occurrence of a whole set of genes horizontal through the genus strongly suggests that HGT events occur frequently between bifidobacteria. The unique genes of PV20–2 and PV8–2 encoded mostly hypothetical or phage related functions, paralleling an observation in the B. breve taxon [44].

Bacteria, including bifidobacteria, secrete proteins to ensure the acquisition of essential nutrients, cell to cell communication, and competition in their respective niches [45]. Extracellular proteins related to iron metabolism, adhesion and antimicrobial activity were identified in the secretome of PV8–2 and PV20–2. Both strains apparently use therefore a combination of strategies to efficiently compete and interact with the environment. This observation is in agreement with recent observations that several mechanisms are used by bacteria to compete in different environments [46–49]. Furthermore, the ability of bacteria to occupy specific environments mainly relies on their ability to obtain adequate supplies of nutrients that are indispensable for their growth, such as iron [50, 51]. In the exoproteome of B. kashiwanohense PV20–2 three ferrous iron transporters were identified which seems directly related to the high iron internalization activity of the strain [12]. In contrast, in the exoproteome of PV8–2 a ferric iron transporter was identified. This suggests different iron acquisition strategies in both strains. Remarkably, the ferric transporter of PV8–2 is likely acquired via HGT and its occurrence in the secretome shows that it is active. The absence of ferrous transporter proteins in the secretome suggests that the acquired ferric transporter has taken over the function for iron uptake. The competition for iron within the gut microbiota and the limited availability of iron in the intestine, leads to a constant battle for iron between the enteric microbiota and the host [7, 52, 53]. Thus efficient iron acquisition mechanisms are likely pivotal for proliferation and persistence in the gut [14, 16, 54]. Production of iron binding proteins by bifidobacteria may also limit the iron available for pathogens thus potentially being implicated in a so called nutritional immunity phenomenon [55–57]. In competition, inhibition, and displacement assays using HT29-MTX cell lines, strain PV8–2 reduces adhesion of Salmonella Typhimurium and Escherichia coli O157:H45 while strain PV20–2 only reduced adhesion of S. Typhimurium [58]. This competitive strength of especially PV8–2 might be in part due to iron binding mechanism limiting iron availability for pathogens.

The zinc ABC transporter ZnuA was identified in the exoproteome of both strains. Zinc metal transporters may facilitate, although to a lesser extent, the uptake of other divalent ions, such as iron [59]. Alternatively, their detection may be the result of a common micronutrient stress sensing. Further, ZinT was identified in PV8–2. ZinT forms a complex with ZnuA to enhance its activity. ZinT was identified in the extracellular fraction of enterohemorrhagic E. coli where it was important for colonization [60, 61]. Zinc competition in the intestine may also play a role in nutritional immunity and control of enteropathogens growth, but further research is needed to confirm the beneficial effects of zinc sequestration in the gut environment [14, 47].

Elongation factors have been found in the secretome lactobacilli and are involved in the adhesion Lactobacillus johnsonii to human intestinal cells and mucins [62, 63]. Their presence in the secretome and the high adhesion properties of both strains [64] suggests a similar role for these factors in bifidobacteria. Similarly, glyceraldehyde-3-phosphate dehydrogenases, and the stress proteins GroEL and DnaK were previously shown to be involved in adhesion of lactobacilli [65–68], and their occurrence in the secretome of PV20–2 and PV8–2 suggests same functionality.

The vast majority of the proteins identified in both secretomes possess an N-terminal signal peptide, and are predicted to be secreted, or were already shown to be extracellular (Additional file 4: Table S4 and S5). In addition, many proteins were predicted to be lipoproteins, while lipoproteins of bifidobacteria were reported to stimulate the immune system and were involved in adhesion [69, 70]. However, the implication of these proteins in more than one function has to be further tested. Furthermore, the lysozyme M1 might be involved in inhibition against competing microorganisms. Lysozymes or glycoside hydrolases such as N-acetylmuramidases are reported to be involved in competition and antimicrobial activity because they can act as bacteriolysins [71].

In our study we used genomics and proteomics to characterize the response to low iron conditions of B. kashiwanohense PV20-2 and B. pseudolongum PV8–2, selected for their high iron sequestration properties. The expression of ferric and ferrous binding proteins and the different responses to iron limitation reflect strain specific characteristics, which is in agreement with high strain variation among bifidobacteria. In addition, to the ferrous iron operon identified in the genome of both studied strains, PV8-2 was shown to have a specific ferric iron operon. The iron binding proteins identified in this study are likely involved in iron metabolism. By combining functional assays and genomics and proteomic tools we could uncover iron-sequestration mechanisms that could provide important competition edge for colonization in the low iron environment of the infant gut. However, the specific function of the extracellular proteins disclosed in this study and their role in the GI-tract colonization remain to be investigated.