Characterization of GH2 and GH42 β-galactosidases derived from bifidobacterial infant isolates

Bifidobacteria are common members of the human gut microbiota, and are especially abundant in the gastrointestinal tract (GIT) of healthy, breast-fed infants (Milani et al. 2017). The presence of bifidobacteria in the gut is believed to provide a number of health benefits for the human host, although the mechanisms by which such benefits are delivered are currently not fully understood (Arboleya et al. 2016; Milani et al. 2017). Bifidobacteria are saccharolytic microorganisms and their ability to utilize complex dietary glycans or host-derived mucins is an important property to assist in their establishment and persistence into the GIT (Koropatkin et al. 2012; Milani et al. 2015; Riviere et al. 2016). The saccharolytic metabolism of bifidobacteria is facilitated by a large array of carbohydrate degrading enzymes, in particular glycosyl hydrolases (GHs), which provide the capacity to directly or indirectly (through syntropy) utilize a range of glycan substrates available in the gut (Milani et al. 2015; Turroni et al. 2018a). Bifidobacteria commonly represent the dominant component of the gut microbiota of healthy, breast-fed infants (Roger et al. 2010), a phenomenon that is believed to be due in part by their ability to metabolize human milk oligosaccharides (HMOs), which contain one or more β-linked galactose moieties (Fuhrer et al. 2010). Among the glycosyl hydrolases encoded by bifidobacteria (Milani et al. 2015) β-galactosidases (which are members of GH families 2 and 42) (van den Broek et al. 2008) have been described to participate to the utilization of (human) milk and milk-based substrates (i.e. lactose, HMOs as well as GOS) in Bifidobacterium bifidum, Bifidobacterium longum subsp. infantis and Bifidobacterium breve (Garrido et al. 2013; Goulas et al. 2007; James et al. 2016; Miwa et al. 2010; Moller et al. 2001; O’Connell Motherway et al. 2013; Yoshida et al. 2012). In addition, β-galactosidases in B. bifidum can also participate to the degradation of mucin (Turroni et al. 2010), while in B. breve such enzymes are involved in the degradation of the plant polymer galactan (O’Connell Motherway et al. 2011).

Human milk glycans can be quite diverse and they are composed of 13 core structures generated through the elongation of lactose at the reducing end with one or more β1,3-linked lacto-N-biose (type-I chain) and/or β1,3/6-linked N-acetyllactosamine units (type-II chain) (Urashima et al. 2012). These core structures (including lactose itself) can in turn be substituted at terminal positions by fucose connected via α1,2/3/4 links, and/or sialic acid residues attached by α2,3/6 links (Smilowitz et al. 2014). Notably, HMOs are especially rich in type-I chains and constitute a characteristic feature of human milk (Urashima et al. 2012).

Some of the glycosidic linkages and monosaccharides found in HMOs are also present in mucins found in the intestine, for example the presence of β-linked galactose and N-acetyl-glucosamine, as well as fucose and sialic acid residues which are substituted to mucin via α1,2/3/4 and α2,3/6 linkages (Tailford et al. 2015).

Recent genome analyses have highlighted metabolic pathways responsible for the degradation of HMOs and mucin-type O-glycans in certain infant-derived bifidobacteria. With regards to HMO utilization it has been shown that B. bifidum and B. longum subsp. infantis can utilize many different HMOs directly (Asakuma et al. 2011; Garrido et al. 2016; Sela 2011; Ward et al. 2007). In contrast, B. breve and B. longum subsp. longum can only metabolize a small number of these directly, thus relying on cross-feeding strategies (Asakuma et al. 2011; Egan et al. 2014; LoCascio et al. 2007; Ward et al. 2007). B. bifidum is the only bifidobacterial species described so far to be able to utilize mucin-type O-glycans (Turroni et al. 2010), and the products of (extracellular) mucin degradation constitute growth substrates for other (bifido)bacterial species, revealing that bifidobacteria can adopt a syntrophic strategy to access different substrates available in the gut (Egan et al. 2014; Turroni et al. 2018b).

The complex structural and functional heterogeneity of HMOs in mother’s milk is presumed to be essential in conferring the bifidogenic activity and associated health benefits to the host, yet it also represents the main limitation for the deliberate incorporation of HMOs in infant formulations (Akkerman et al. 2018). For this reason alternative ways of recreating the beneficial effects of human milk have focused on the enzymatic production of other non-digestible, lactose-derived prebiotics, such as GOS produced by β-galactosidases that have transgalactosylation activities (Macfarlane et al. 2008).

Based on the above, the relevance of diet-based, lactose-derived oligosaccharides (e.g. HMOs and GOS) in supporting bifidobacterial establishment and persistence in the gut is obvious, especially during early life when bifidobacteria are particularly abundant (Milani et al. 2017). The current study focused on a survey and subsequent hydrolytic characterization of (selected) β-galactosidases identified from sequenced strains of bifidobacterial species which have been previously shown to be members of the infant gut microbiota (i.e. B. breve, B. bifidum, B. longum subsp. longum and B. longum subsp. infantis) (Lewis and Mills 2017; Milani et al. 2017; Turroni et al. 2012). Following comparative genome analysis, our analysis focused on a number of selected β-galactosidases of which hydrolytic capabilities were assessed for various β-galactoside-containing substrates.

In the current work we investigated β-galactosidase enzymes belonging to four bifidobacterial species B. breve, B. bifidum, B. longum subsp. longum and B. longum subsp. infantis, all of which are commonly found in healthy, breast-fed infants. Our analysis clearly shows that multiple distinct β-galactosidases are encoded by these infant-derived bifidobacterial species (Additional file 1: Table S1) with a partial overlap between their hydrolytic activities (Tables 4 and 5). Notably, for some of the identified β-galactosidases the involvement in utilization of galacto-oligosaccharides and/or host-derived glycans has previously been reported (Additional file 1: Table S1). Of note, some of the clusters from our comparative analysis appeared to be species specific, suggesting that certain β-galactosidase encoding genes are only found within certain species. For example the enzymes belonging to Clusters 1, 2, 3, 6 and 9 contain a wide range of B. bifidum β-galactosidases which represent homologs of BbgIII (NCIMB 41171), BbgI (NCIMB 41171), BbgIV (NCIMB 41171), BIF2 (DSM 2015) and BIF1 (DSM 20215), respectively. Notably, these enzymes have previously been reported to be capable of GOS synthesis (Additional file 1: Table S1) (Goulas et al. 2007; Miwa et al. 2010; Moller et al. 2001).

Our comparative analysis was used as a guide to select 20 candidates to be used for further assessment of β-galactosidase activity towards a number of substrates including a range of selected β-galactosyl-containing structures. Cloning efforts of the selected genes resulted in the successful overexpression and purification of 17 presumed β-galactosidases, of which hydrolytic activity was further assessed on a variety of substrates. The obtained findings revealed that the majority of the expressed enzymes indeed exhibit β-galactosidase activity. However, these β-galactosidases clearly differ from each other in their substrate specificity, since some of the enzymes are active on nearly all tested substrates, while others appear to be quite specific and merely hydrolyse one or two substrates.

Of note, the enzymes that are capable of hydrolysing the broadest range of selected substrates are members of Cluster 15 and are also widespread across the four (sub)species (i.e. Bbr_0529, B216_08266, B8809_0415 and Blon_2016), suggesting that these genes encode conserved β-galactosidase activities. Besides, enzymes of Cluster 15 appear to be homologous to the previously identified BbgII (from B. bifidum NCIMB 41171), which is capable of both GOS hydrolysis and synthesis (Goulas et al. 2007; Miwa et al. 2010; Moller et al. 2001) (Fig. 1). Notably, some of the β-galactosidases represented by this cluster have previously been shown to be involved in the utilization of either GOS and HMOs (Additional file 1: Table S1). In particular, it has been shown that Blon_2016 from B. longum subsp. infantis ATCC 15697 represents a β-galactosidase possessing hydrolytic activity towards LNT and LNnT, as well as GOS (Garrido et al. 2013), which is indeed confirmed by our observations (Tables 4 and 5).

Another member of Cluster 15 is the β-galactosidase encoded by Bbr_0529, which is required for the utilization of GOS and certain HMOs by B. breve UCC2003 (James et al. 2016; O’Connell Motherway et al. 2013), being consistent with our observed hydrolytic activities towards the majority of tested substrates including galactobiose, galactotriose, but also the central moieties of type I and type II HMOs (Tables 3, 4). As type I chains represent the most abundant HMO core structure in human milk (Urashima et al. 2012), the prevalence of Cluster 15 members across all four infant-derived bifidobacterial (sub)species (Fig. 3b) and observed ability to cleave LNT (but also LNnT) (Table 4) supports their relevance to bifidobacteria in obtaining access to substrates derived from a human milk-based diet.

Surprisingly, one of the β-galactosidases whose product was shown to be incapable of hydrolysing lactose or lactulose is Bbr_0420, which was instead shown to be active towards d-galactotriose Galβ1-4Galβ1-4Gal (Table 4). Previous studies established that Bbr_0420, a β-galactosidase which is upregulated when B. breve UCC2003 is grown on purified GOS (O’Connell Motherway et al. 2013), seems to be dedicated to the hydrolysis of d-galactotriose, which in turn is produced from the extracellular degradation of potato galactan in this strain (O’Connell Motherway et al. 2011). The transcription of this gene is, however, not induced when B. breve UCC2003 is grown on lactose and certain HMOs (e.g. LNT or LNnT) (James et al. 2016). Bbr_0420 from B. breve UCC2003 and its homologue B8809_0321 from B. longum subsp. longum NCIMB 8809 (both belonging to Cluster 13) were shown here to hydrolyse galactobiose (Galβ1-4Gal) and galactotriose (Galβ1-4Gal-β1-4Gal), but incapable of hydrolysing either LNT or LNnT. This suggests that they represent β-galactosidases with a narrow substrate specificity, which is directed towards β1,4 galactosidic links (Tables 4, 5). Interestingly, a recent study has associated a homolog corresponding to this β-galactosidase and located within the galactan cluster of B. longum AH1206 (BL1206_0411) with the persistence of this strain in the GIT tract (Maldonado-Gomez et al. 2016). This finding represents another indication of the role played by β-galactosidases and their dietary galactose-containing substrates in promoting bifidobacterial colonization.

Notably, purified Bbr_0010 from B. breve UCC2003 has previously been reported to hydrolyze lactose and LNnT (James et al. 2016), while Bbr_1552 represents a β-galactosidase with a broader substrate specificity, being involved in the hydrolysis of GOS, as well as lactose, LNT and LnNT in this strain (James et al. 2016; O’Connell Motherway et al. 2013), which is also consistent with our observations. Interestingly, a previous study employing a Tn5-based random mutagenesis system in B. breve UCC2003 identified Bbr_0010 as the main β-galactosidase responsible for lactose utilization in this strain (Ruiz et al. 2013). Based on our analysis this enzyme does indeed hydrolyse lactose (as well as LNnT) and the corresponding gene is conserved across B. breve (as Cluster 5 members).

Two interesting cases are constituted by B8809_0611 (Cluster 7) and B8809_1361 (Cluster 8) from B. longum subsp. longum NCIMB 8809 as these two β-galactosidases show similar hydrolytic profile and are particularly active towards LNnT (type II chain) despite being member of two different clusters (Tables 4 and 5). Cluster 7 member Blon_2334 from B. longum subsp. infantis ATCC 15697 has been described as a β-galactosidase responsible for degradation of lactose and type II HMOs, but not type I HMOs (Yoshida et al. 2012), being consistent with what we observed for B8809_0611. In contrast, B8809_1361, which shows a hydrolytic profile similar to Blon_2334 and B8809_0611, belongs to a different cluster (Cluster 8). Finally, B8809_1361 is only present in certain strains of B. longum subsp. longum, and perhaps constitutes an auxiliary β-galactosidase responsible for degradation of lactose and type II HMOs in this subspecies (Additional file 1: Table S1).

In conclusion, the information collected in this study highlights the importance of GHs in bifidobacterial saccharolytic metabolism, in particular β-galactosidases which are involved in the utilization of a range of substrates such as lactose, HMOs, and GOS, found as part of the milk-based infant diet. The qualitative assay presented in this study provides a clear insight on the diversity of β-galactosidases in terms of substrate specificity. In fact, some appear to be more specialized towards milk-based substrates, while others are specific for plant-derived substrates. Therefore, the findings presented here constitute a solid foundation for future studies on bifidobacterial β-galactosidases and further investigation on the role of milk-derived substrates in establishing bifidobacterial predominance in the infant GIT.