Abstract
The infant gut microbiome is impacted by early-life feeding, as human milk oligosaccharides (HMOs) found in breastmilk cannot be digested by infants and serve as nutrients for their gut bacteria. While the vast majority of HMO-utilization research has focused on Bifidobacterium species, recent studies have suggested additional HMO-utilizers, mostly Bacteroides, yet their utilization mechanism is poorly characterized. Here, we investigate Bacteroides dorei isolates from breastfed-infants and identify that polysaccharide utilization locus (PUL) 33 enables B. dorei to utilize sialylated HMOs. We perform transcriptional profiling and identity upregulated genes when growing on sialylated HMOs. Using CRISPR-Cas12 to knock-out four PUL33 genes, combined with complementation assays, we identify GH33 as the critical gene in PUL33 for sialylated HMO-utilization. This demonstration of an HMO-utilization system by Bacteroides species isolated from infants opens the way to further characterization of additional such systems, to better understand HMO-utilization in the infant gut.
Similar content being viewed by others
Introduction
Following birth, the dynamic and complex process of the infant gut microbial colonization is initiated1,2,3,4. The main factors that determine the composition of the infant gut microbiome are delivery mode2,5,6,7 and infant feeding (formula vs. breast-milk)5,7,8,9. Specifically, the third-largest solid component in breast-milk is human milk oligosaccharides (HMOs), which are a family of glycans indigestible by infants10,11, left to be consumed by bacteria in the infant’s gut. HMOs have a variety of health benefits, such as being decoy receptors for pathogens, as they resemble receptors found on the infant’s intestinal epithelium, with which pathogens can interact12,13,14,15,16. HMOs can be divided into three main groups based on their structure: fucosylated, sialylated, or neutral sugars. Sialylated sugars have special importance in infant health and are able to shift metabolism to increase lean body mass and improve bone morphology17,18. Also, offspring of mice producing milk deficient in sialylated HMOs were found to have defects in nervous system development and neuronal patterning, particularly in areas associated with executive functions and memory19.
One of the most studied roles of HMOs in infancy is the promotion of a healthy infant gut microbial community, advantageous for future health20,21,22,23. HMOs support the colonization of beneficial gut bacteria that can break them down24 and most research in the past two decades has focused on Bifidobacterium species25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41, which are considered as the hallmark HMO-utilizers in the breastfed-infant gut microbiome. However, recent evidence suggests that Bifidobacterium species are not the only commensal HMO-utilizers in the infant gut. Studies investigating the infant gut microbiome composition report frequent cases (up to 90%) in which breastfed-infants do not harbor any Bifidobacterium species throughout their breastfeeding period2,42,43. There are also significant correlations between specific HMO levels and the relative abundance of other genera, such as Bacteroides, Parabacteroides, and Lactobacillus28,44,45,46. In addition, computational annotation of genomes of common infant gut microbes points towards some genes as potential HMO-utilizers2,43,47, mostly from the Bacteroides genus. Taken together, all these suggest that Bacteroides species are also candidates for HMO utilization.
Bacteroides are known degraders of complex versatile glycans48,49,50,51, using thousands of enzyme combinations52, namely glycoside hydrolases (GHs)53. Their glycan utilization repertoire is determined by the variety of polysaccharide utilization loci (PULs) present in their genome54,55. A PUL is a gene cluster containing jointly-regulated genes that encode the machinery necessary for the breakdown of a specific sugar. Bacteroides species have developed multiple strategies for utilizing HMOs with varying levels of efficiency, as was demonstrated for type strains of B. fragilis, B. thetaiotaomicron and B. vulgatus on both a pool of HMOs56 and individual fucosylated HMOs57. Some of the mechanisms have been further identified, for example, that B. thetaiotaomicron and B. fragilis utilize both host mucin-O-glycans and HMOs using a similar set of upregulated genes56. For fucosylated HMOs, fucosidases from Bacteroides were profiled and showed versatility in efficiency and substrate bond specificity58. Finally, for sialylated sugars, one of B. thetaiotaomicron’s GH33 sialidases was crystallized and characterized59, unraveling a wide binding groove accounting for a broad substrate specificity, including various sialylated HMOs60.
While these pioneering works lay the foundation for understanding HMO utilization by Bacteroides species, further research is needed to extend the mechanistic characterization to additional infant gut commensal strains and specific types of HMOs. Commonly, mechanistic analyses rely on genetic manipulation of tested strains, mainly used in type strains that have well-characterized genomes, functions, and systems, rather than on natural isolates that differ (to an unknown extent) from their closest type strains. Bacteroides research thus far has focused on the B. thetaiotaomicron type strain61, which requires a mutant parental strain62,63,64,65 for allelic-exchange-based genetic manipulation66,67,68. However, this commonly used method suffers from several limitations, mainly low efficiency (especially in isolates). Focusing on natural isolates, generating a parental strain for each isolate is unlikely, thus additional approaches should be implemented. Finally, the genetic tools developed in E. coli tend to be incompatible with the transcription-translation machinery of Bacteroides, making the development of novel genetic manipulation systems even more challenging. Luckily, recent adaptations of a CRISPR-based system for gene knockout in Bacteroides69,70 paved the way for specific, markerless, genomic editing of Bacteroides genomes, which is efficient not only for type strains but also for natural isolates.
Here, we implemented the recently developed CRISPR-Cas system to genetically manipulate natural isolates of Bacteroides from the infant gut. In B. dorei, we identified PUL33 as a necessary gene cluster for sialylated HMO utilization and specifically the GH33 as the critical gene for this purpose within this cluster. Finally, expressing B. dorei’s PUL33 genes in bacteria that cannot utilize sialylated HMOs, enabled them to grow on these sugars, demonstrating the sufficiency of this gene cluster.
Results
B. dorei and B. vulgatus grow on HMOs with varying efficiency
In our search for non-Bifidobacterium HMO-utilizers, we started with two B. dorei strains that we recently isolated, which were shown to grow on various HMOs71. Before we continued to mechanistically examine the HMO utilization process in these strains, we first explored the generality of this phenotype in additional natural isolates. Towards this end, we isolated two additional B. dorei strains and six B. vulgatus strains (which are closest taxonomically in the Bacteroides species tree72) from breastfed-infant stool samples. We grew the isolates on minimal media (MM) supplemented with a single HMO representing all three structural groups; Fucosylated HMOs: 2-Fucosyllactose (2’-FL) & Difucosyllactose (DFL); Sialylated HMOs: 3-Sialyllactose (3’-SL) & 6-Sialyllactose (6’-SL); and neutral HMOs: Lacto-N-Tetraose (LNT) & Lacto-N-neotetraose (LNnT). We examined the growth kinetics of all 10 isolates on these sugars compared to glucose and lactose, which served as a control.
Vast majority of isolates were able to utilize the tested HMOs with varying degrees of efficiency (Supplementary Fig. 1). For all isolates, growth on fucosylated HMOs was inferior compared to growth on sialylated or neutral HMOs, both in terms of the lag phase length and the peak-OD value. This finding was surprising as fucosidases from GH families 29 and 95 (that participate in fucosylated-HMO breakdown73) are more prevalent among B. dorei and B. vulgatus compared to specialized sialidases from the GH33 family (that participate in sialylated-HMO breakdown73). When comparing the average counts of enzymes per genome, we found there are 8.82 average GH29 fucosidases and 4.14 GH29 fucosidases per genome, compared to only 2.14 average GH33 per a B. dorei or B. vulgatus genome74 (see the “Methods” section).
Hereafter, we chose to focus on B. dorei isolates growing on sialylated HMOs due to a number of reasons. First, sialylated HMOs contribute to infant health and their effect is mediated by the gut microbiome17,18,19 (see the “Introduction” section). Second, the consistent growth dynamics on the two tested sialylated HMOs (3’-SL, 6’-SL), together with the unique terminal residue of sialic acid suggest a specialized pathway or mechanism to be discovered, which is specific for sialylated sugar structures. Lastly, B. dorei’s growth on sialylated HMOs is overall comparable to glucose and lactose, especially in contrast to the poor growth on fucosylated structures, suggesting that sialylated HMOs serve as an effective carbon source for the species B. dorei.
PUL33 is upregulated in B. dorei in response to sialylated HMOs
Differential expression of genes under various conditions can highlight the enzymes, mechanisms, and pathways enabling the utilization of specific carbon sources. Here, we re-analyzed previously published71 RNA-sequencing data of one of our B. dorei isolates when growing on MM supplemented with the individual HMOs described earlier. The whole genome transcriptional profiling analysis revealed that a specific gene cluster called PUL33 is significantly upregulated when growing on sialylated HMOs compared to glucose (fold-change > 2, padj < 0.001; Fig. 1A, B). Of the remaining 65 PULs in B. dorei’s genome74, none were completely upregulated as PUL33 was, rather only some of their genes were upregulated (Fig. 1A, Supplementary Fig. 2). The upregulation of PUL33 was specific to sialylated HMOs and was not evident on fucosylated or neutral HMOs (Fig. 1B, Supplementary Fig. 3), suggesting that PUL33 has a distinct role in the utilization of the sialylated HMOs tested (3’-SL, 6’-SL).
Differential gene expression data comparing minimal media supplemented with various HMOs (2’-FL, DFL, 3’-SL, 6’-SL, LNT, LNnT) to glucose. A Log2 fold-change of genes encoded in polysaccharide utilization locus (PUL) 33 and the previous PULDB annotation of PUL8 in B. dorei. The gene GKD17_02665 was not annotated as GH33 in the subsequent version. Asterisks indicate significance as outputted by DESeq2 using the Wald test, after correcting for multiple hypothesis using the Benjamini and Hochberg method (p_adjusted *<0.05; **<0.01; ***<0.001). For all genes in PUL33 p_adjusted values were lower than 8.26*10−10 when comparing expression on sialylated HMOs compared to glucose. B Volcano plots of gene expression on sialylated HMOs, showing the fold-change (log2, x-axis) vs. the differential significance (-log2(p_adjusted), y-axis), with highlighted genes from PUL33 and the previous annotation of PUL8. The statistical analysis is the same as in (A). C Schematic of sialylated HMO utilization by the core genes in B. dorei’s PUL33. GH: Glycoside hydrolase. Figure cartoon was created with BioRender.com [https://www.biorender.com/]. Source data are provided as a Source Data file.
To further pin-point the specific genes that play a role in the breakdown of sialylated HMOs, we examined the individual genes of PUL33. One of the first steps in utilizing 3’-SL and 6’-SL is releasing the sialic acid residue, making the lactose core available for further breakdown. This function is performed by sialidases belonging to the glycoside hydrolase (GH) family 3353. Indeed, the gene encoding a GH33 sialidase was among the most upregulated genes in the entire genome and specifically the highest in PUL33, with a 50.21 and 58.49-fold increase in expression when growing on 3’-SL and 6’-SL, respectively (Fig. 1A, B). In the current annotation of B. dorei, this is the only GH33 gene. However, previous versions of B. dorei’s assembly in the Polysaccharide-Utilization Loci DataBase (PULDB) included an additional GH33 sialidase named GKD17_02665 found in a previous version of PUL874. The expression of the previously-annotated GH33 was not upregulated at all in our data (Fig. 1A, B). The lack of upregulation of GH33-PUL8 under the tested growth conditions supports the notion that GH33-PUL33 in the main sialidase in the context of 3’-SL and 6’-SL, however, does not rule out its potential role in utilizing other sialylated HMOs not tested here.
PUL33 is necessary for sialylated HMOs utilization in B. dorei
In order to test if PUL33 is necessary for the utilization of sialylated HMOs, we aimed to assess the growth of B. dorei on minimal media containing 3’-SL and 6’-SL, while lacking specific genes in PUL33. It is not straightforward to knock-out a genomic region from a natural isolate of Bacteroides62,63,64,65,66,67 and for this purpose, we used a CRISPR-based system recently developed utilizing Cas12a (Fig. 2A)69. As the CRISPR-based system requires sensitivity to erythromycin, we implemented it on an erythromycin-sensitive B. dorei isolate from our strain collection and confirmed by RT-qPCR that it also upregulates PUL33 upon growth on sialylated HMOs (Supplementary Fig. 4A). Additionally, we performed whole genome sequencing of both isolates using long-read Oxford Nanopore Technology and generated a synteny map comparing them to the type strain B. dorei DSM 17855. The sequencing results revealed a high degree of similarity (average nucleotide identity range 98.92–99.04%) between the two isolates and the type strain, including the conservation of the entirety of PUL33 (Supplementary Fig. 4B).
A Overview of the CRISPR-cas12-based genetic manipulation system, including the transformation of E. coli and conjugation to the recipient Bacteroides. Figure cartoon was created with BioRender.com [https://www.biorender.com/]. B PUL33 gene structure with GenBank locus tags, highlighting the regions of homology (LF: left flank, pink; RF: right flank, green), with colors matching to the plasmid in (A) and primer locations (P1–P4). The genes that are nonfunctional in the mutant are shaded in orange. SusC SusC-like TonB-dependent transporter, SusD SusD-like cell-surface glycan-binding protein, GH glycoside hydrolase, CE carbohydrate esterase, Est Sialate O-acetylesterase, Unk unknown function. C PCR results from wild-type (WT) colonies, representing failed Cas12a inductions in the attempt to create a mutant and a single mutant (KO) colony, with shorter PCR product (lacking the removed genes). D Growth curve plots of B. dorei, B. dorei Δ5k-PUL33 and B. dorei Δ5k-PUL33::p_compFull. Growth curves were measured in minimal media supplemented with sialylated HMOs (3’-SL, 6’-SL; dark and light pink, respectively), lactose (positive control; green) or no carbon media (negative control; gray), 0.5% weight/volume. Thick lines represent the average of two biological replicates, each consisting of three technical replicates, which are represented by thin lines (n = 6). p-values were calculated using a paired two-sided t-test at 24 h. E Structure overview of the plasmid p_compFull. Source data are provided as a Source Data file.
Next, we constructed a B. dorei mutant, lacking 5000 bp from PUL33, which renders the genes encoding GH20, GH33, SusD, and SusC unfunctional (Fig. 2B). We chose to delete this specific region as it contains the transport system and the unique GH33 sialidase which was the top upregulated gene in the transcriptional analysis (Fig. 1A). To validate the mutant, we performed two PCR reactions: First, using primers located before and after the deletion, showing that the PCR product is 5000 bp shorter in the mutant strain compared to the wild type (WT) strain (P1, P2 in Fig. 2C); Second, with one primer located within the deleted region, demonstrating a product at the predicted length for the WT and a lack of product for the mutant strain (P3, P4 in Fig. 2C). We also confirmed the B. dorei knock-out (KO) using sanger sequencing.
The KO mutant, designated B. dorei Δ5k-PUL33, was able to grow as well as the WT on the fucosylated and neutral HMOs (Supplementary Fig 5), however, it could not grow at all on any of the sialylated HMOs tested (p-values = 5.07 × 10−8 and 1.3 × 10−7 for 3’-SL and 6’-SL, respectively; Fig. 2D). To further validate our phenotype, we performed a complementation assay of the four deleted genes, using a previously described pNBU_erm integration vector75, containing the desired genes under the constitutively expressed P_BfP5E1 promoter67 (p_compFull; Fig. 2E). The complemented mutant, expressing the genes in a single copy from a heterologous genomic location, had restored the ability to grow on the sialylated HMOs (Fig. 2D, right). The complemented mutant reached a lower OD than the WT strain, possibly due to the expression of the complemented genes from the plasmid’s att1/2 integration site, rather than their endogenous genomic location. Taken together, these results indicated that the phenotype matches the genotype and further enhanced the essentiality of PUL33 in sialylated HMO utilization in B. dorei.
The core genes in PUL33 enable sialylated HMO utilization in additional Bacteroides species
Next, we sought to examine whether the mere expression of GH20, GH33, SusD, and SusC from PUL33 is sufficient to drive sialylated HMO utilization in additional Bacteroides species. We selected two Bacteroides strains: B. uniformis CL03T12C37 and a B. stercoris isolate, that were unable to grow on sialylated HMO (Fig. 3A, Supp. Fig 6A) and do not encode any PUL33 homologs (see the “Methods” section). We then conjugated them with an E. coli strain containing the four core genes on p_compFull. While both wild-type strains did not utilize sialylated HMOs, the transconjugants, which have integrated p_compFull into their genome, were able to grow on these sugars. The B. uniformis strain was able to fully grow on the sialylated glycans with the addition of the PUL33 core genes, to an OD600 comparable to that of lactose (Fig. 3A) and the B. stercoris isolate exhibited significant, yet modest growth on the sialylated glycans (Supplementary Fig 6A). Although the growth of B. stercoris::p_compFull on sialylated HMOs is not robust, it is noteworthy, especially in the context of B. stercoris’ growth on lactose, which is moderate as well (Supplementary Fig 6A). These results indicate that the four core genes (GH20, GH33, SusD, and SusC) are sufficient to cleave the sialic acid group from sialylated HMOs and thereby enable the growth of these sugars. Furthermore, this biochemical function is not entirely dependent on the specific cellular environment of B. dorei.
A Growth curve plots of B. uniformis CL03T12C37 in minimal media supplemented with sialylated HMOs (3’-SL, 6’-SL; dark and light pink, respectively), lactose (green), or water (no carbon; gray), 0.5% weight/volume. Thick lines represent the average of two biological replicates, each consisting of three technical replicates, which are represented by thin lines (n = 6). p-values were calculated using a paired two-sided t-test at 50 h. This strain gained the ability to grow on sialylated HMOs once conjugated with p_compFull. B Structure overview of the plasmids p_compPartial1/2/3/4. C Growth curve plots of B. dorei Δ5k-PUL33 and B. uniformis CL03T12C37 complemented with p_compPartial1/2/3/4, in minimal media supplemented with sialylated HMOs (3’-SL, 6’-SL), 0.5% weight/volume. The colors and replicates are the same as in (A). Source data are provided as a Source Data file.
GH33 in PUL33 is the key gene in the ability to utilize sialylated HMOs
Going back to B. dorei’s functionality and following the full complementation that restored the phenotype in B. dorei Δ5k-PUL33, we sought to explore whether all four complemented genes are necessary for PUL33’s function by partial complementation assays. We constructed three partial complementation plasmids (p_compPartial1/2/3) each complementing a different set of genes from the four core ones (Fig. 3B) and conjugated them into the mutant B. dorei strain (hereby annotated as B. dorei Δ5k-PUL33::p_compPartial1/2/3). We confirmed the expression of the complemented genes using reverse-transcription quantitative PCR (RT-qPCR; Supplementary Fig 7). Of the three complemented strains, B. dorei Δ5k-PUL33::p_compPartial1 and B. dorei Δ5k-PUL33::p_compPartial3 were able to grow well on sialylated HMOs, while B. dorei Δ5k-PUL33::p_compPartial2 was not able to grow at all on this media (Fig. 3C).
The plasmids that restored growth were only the ones that contained the GH33 gene (GKD17_15975), while the plasmid that did not restore growth was the one that lacked this gene (Fig. 3B, C). These results suggest that GH33 is critical for the utilization of sialylated HMOs in B. dorei, while the other genes of this cluster can be substituted by other enzymes encoded in B. dorei’s genome. We further validated this result by complementing B. dorei Δ5k-PUL33 with GH33 only (p_compPartial4), which enabled growth on sialylated HMOs (Fig. 3B, C). For B. uniformis CL03T12C37 as well, complementation with plasmids containing the GH33 gene (p_compPartial1/3/4) allowed modest growth on 3’-SL and 6’-SL, whereas complementation with the p_compPartial2 plasmid that lacks GH33 did not convey this ability (Fig. 3B, C). However, for the B. stercoris isolate, none of the partial complementation plasmids were sufficient to drive significant growth on sialylated HMOs (Supplementary Fig 6B, C), suggesting additional functions are missing in the B. stercoris cellular environment to enable even a modest growth.
Predicted structure of B. dorei’s GH33 suggests a broad substrate specificity
To have a broader understanding of the phylogenetic context of the GH33 gene, we next performed a phylogenetic analysis of a subset of GH33 protein sequences deposited in the Carbohydrate-Active EnZymes (CAZy) database76 (see the “Methods” section; Supplementary Fig 8). We found that most Bacteroides GH33 sequences cluster together, adjacent to other bacterial sialidases, with the GH33 proteins of B. dorei and B. vulgatus ATCC 8482 the most similar (Supplementary Fig 8), with a 97.44% identity (Supplementary Fig 9A). Additionally, some Bacteroides GH33 sequences were clustered closer to the eukaryotic branch as was previously described77.
When zooming into the Bacteroides branch, there is a single B. thetaiotaomicron VPI 5482 GH33 protein for which the structure has been characterized using crystallography (GenBank accession AAO75562.1; PDB code 4BBW)59. We compared the predicted structure of our B. dorei isolate’s GH33 (using AlphaFold;78 pLDDT = 95.1; Fig. 4A) to the known structure of B. thetaiotaomicron’s GH33. Other than the first 23 N-terminal residues of B. dorei’s GH33, for which AlphaFold could not predict structure, the overall structure of the proteins is very similar (RMSD = 0.444 Å for a 521 out of 546 residues alignment; Fig. 4B, Supplementary Fig. 9A, see the “Methods” section). The six-bladed β-propeller catalytic site of GH33 is conserved between B. thetaiotaomicron and B. dorei, with 100% identity for the catalytic residues and additional residues that were found to interact with the substrate in the active site (Fig. 4C, Supplementary Fig 9B). The wide-open groove around the active site in B. thetaiotaomicron GH33 sialidase was found to be responsible for the broad substrate specificity of this enzyme59. Of the three residues that contribute to this wide groove, two were conserved in B. dorei while one differed between the hydrophobic Alanine and Valine (residue 228; Fig. 4C). This high conservation of the catalytic site, its neighboring residues, and the wide topology around the active site suggests that like B. thetaiotaomicron VPI 5482, B. dorei’s GH33 could also break down various sialylated glycans found in the infant gut.
A 3D structure of B. dorei’s GH33 sialidase, as predicted by AlphaFold based on the protein sequence, excluding the first 23 residues for which the pLDDT values are low. The active site and carbohydrate-binding site are indicated by a green circle. B Alignment of GH33 protein structures of B. dorei (predicted, blue) and B. thetaiotaomicron VPI 5482 (crystallized, red; PDB code 4BBW). The first 23 residues from B. dorei’s enzyme were excluded similarly to (A). C The wide-open substrate-binding groove of the active site in B. dorei (left) and B. thetaiotaomicron VPI 5482 (right). Residues essential for activity are highlighted in red and residues responsible for laying out the wide-open groove are highlighted in magenta. Of the latter group of residues, although one of the three is not conserved between B. dorei and B. thetaiotaomicron (Ala228 and Val228, respectively), the wide topology of the groove is preserved.
Discussion
In this study, we explore the growth kinetics of B. dorei and B. vulgatus isolates on various HMOs and implement a CRISPR-Cas system to genetically manipulate a natural isolate of B. dorei from a breastfed-infant stool sample. Using this system, we find that a single polysaccharide utilization locus called PUL33 is necessary for sialylated-HMO utilization in B. dorei. We demonstrate that the first four genes encoded in PUL33—GH20, GH33, SusD and SusC—provide the ability to utilize sialylated HMOs in additional Bacteroides strains that are unable to grow on these sugars. Finally, utilizing partial complementation assays, we show that GH33 is the critical gene of PUL33 with respect to utilizing sialylated HMOs.
Bacteroides isolates exhibit robust growth on sialylated and neutral HMOs, whereas for most isolates the growth on fucosylated HMOs is partial, both in terms of a long lag phase and a low maximal OD600 value. Despite the fact that Bacteroides encode more fucosidase enzymes than sialidase enzymes in their genome, their ability to utilize fucosylated HMOs is poor. In Bifidobacterium species, well-known fucosylated-HMO utilizers79, it was shown that the mere presence of fucosidase genes does not necessarily translate phenotypically to better utilization, probably due to a strict substrate specificity80,81. Additional factors that can account for the inferior growth of Bacteroides on fucosylated HMOs are the lack of specific transporters for fucosylated HMOs, non-functional genes, or the inability to properly transcribe or translate genes in the pathway for the utilization of fucosylated HMOs.
From an evolutionary perspective, it seems like Bacteroides and Bifidobacterium species complement each other in their ability to utilize HMOs. Corresponding to the previous characterization of sialic acid consumption by Bacteroides species82, they are adaptive to the utilization of sialic acid-containing glycans, such as the glycoproteins in the mucus layer and sialylated HMOs. Bifidobacterium species, however, are superior in their fucosylated-HMO utilization79, perhaps at the expense of their sialylated-HMO utilization. This interesting dissimilarity in carbon preference could indicate that different species diversified to occupy various niches and utilize the variety of available carbon sources in the environment.
Focusing on our B. dorei isolate, we find that PUL33 is necessary for sialylated-HMO utilization. The conservation of PUL33 across all B. dorei and B. vulgatus genomes deposited in the PULDB database74 suggests that PUL33 is important for the breakdown of sialylated HMOs. For B. thetaiotaomicron VPI 5482, a commonly used type strain, the closest PUL33 homolog is PUL9, experimentally shown to be involved in degrading sialic acid-containing mucins and HMO utilization50,56. While the GH33 protein sequence in the B. dorei and B. thetaiotaomicron clusters is similar (67.65%), their SusC transporters are only 20.67% alike (Supp. Fig 9C), suggesting the transporters have evolved to recognize varying substrates, or that these two clusters evolved independently to perform similar functions in different ways.
Interestingly, in the complementation experiments of B. dorei, we observe two growth pattern types: one resembling the WT strain growth (Fig. 3C; p_compPartial 1 and 4) and the other more delayed, with diverging growth patterns for 3’-SL and 6’-SL (Figs. 2D, 3C; p_compFull and p_compPartial 3). These differences are intriguing and could represent additional underlying principles that can be further characterized in follow-up studies. Additionally, we demonstrate that B. dorei can grow on 3’-SL and 6’-SL even in the absence of the SusCD transport system. This phenotype could potentially be explained by alternative transporters or by non-specific entrance of the small molecules of 3’-SL and 6’-SL into the cell through pores. These hypotheses are supported by a study on B. thetaiotaomicron, showing delayed, yet successful, growth on ribose even in the absence of the ribose-specific transporters55, suggesting that some simple sugars can indeed enter the cell non-specifically. Moreover, sialic acid is known to enter microbial cells by diffusion through nonspecific porins (OmpF, OmpC) in addition to the active TonB-dependent transport system found in Bacteroides83,84, although this was not shown thus far for small molecules containing sialic acid.
Finally, when we zoom into GH33, the key gene in PUL33 responsible for the cleavage of the sialic acid residue from the HMO backbone, we find that B. dorei’s GH33 protein sequence is very similar to that of its close relative B. vulgatus and less so to B. thetaiotaomicron (97.44% vs. 67.65% similarity; Supplementary Fig. 9A). An interesting common feature of these GH33 proteins is a wide-open groove around the active site, which differs from other sialidases (e.g. leech) with a more closed active site85,86. The wide active site was previously linked to a wide substrate specificity, showing that GH33 can cleave terminal sialic acids bound to glycans in varying bonds (e.g., α-2,3, α-2,6, α-2,8). The ability to acquire sialic acid from a variety of sources is beneficial for survival in the competitive ecological environment of the infant gut and it was previously shown that sialic acid drives changes in the gut microbiota composition and pathogen colonization60,87.
Understanding how common infant gut microbes utilize sialylated HMOs and how the infant gut microbiome composition is modulated by these sugars will enable us to make informed decisions about formula supplementation, ultimately improving infant feeding and overall health. In the hypothetical matrix of which carbon sources can be utilized by which bacteria, there are still many unknowns. While these experiments are laborious, genetic manipulation tools tackling one pathway at a time are critical for gaining a mechanistic understanding of the biological pathways accounting for carbon utilization in bacteria.
Methods
In this study, we isolated strains from a total of 10 breastfed-infants stool samples. Mothers of all infants have agreed to participate in our study, which was approved by the Hebrew University’s Institutional Review Board (IRB) (approval number 20042021), and signed our consent forms.
Statistics and reproducibility
The previously published71 differential expression analysis results, re-analyzed here, include fold-change values and adjusted p-values that were used to generate Fig. 1A, B, and the statistical tests used are described in the relevant section below. For all growth curve experiments, the experiments were performed using two biological replicates, in triplicates (n = 6), and were repeated twice. The statistical tests used for comparison between growth curves are described in the relevant section below. For genetic manipulation experiments, multiple colonies (20–100) were screened for the desired deletion using PCR and three mutants were identified. Figure 2C (left panel) shows four representative colonies for which deletion failed, and one representative mutant colony for which genetic manipulation was successful. Stool samples collected from breastfed infants were only used for strain isolation, and no other characterization of the samples was performed. No comparisons were made across samples or between samples collected for this study and other cohorts, and no sex, gender, or age analysis is relevant. The Investigators were blinded with respect to what stool samples they were isolating strains from. Predetermined sample size calculations or randomization considerations were less relevant here for these reasons.
Microbial strains and growth conditions
The B. uniformis CL03T12C37 strain was kindly provided by the Geva-Zatorsky lab, Rappaport Technion Integrated Cancer Center, Israel. Anaerobes were cultured at 37 °C in an anaerobic chamber (COY) that was maintained at >2.5% H2 by flushing with a gas mix containing 20% H2, 5% CO2, and 75% N2. Bacteroides strains were routinely grown in supplemented Brain Heart Infusion (sBHI) media, supplemented with 50 ml/L fetal bovine serum (FBS; Sigma, F2442), 10 ml/L trace vitamins (ATCC® MD-VS™), 10 ml/L trace minerals (ATCC® MD-TMS™), 10 ml/L vitamin K1 and hemin (BBL, 212354), 1 g/L d-(+)-Cellobiose (Alfa Aesar, 528507), 1 g/L d-(+)-Maltose (Caisson, 6363537), 1 g/L d-(+)-Fructose (Sigma Aldrich, 1286504), and 0.5 g/L l-Cysteine (Acros Organics, 52904). For growth curve experiments, Bacteroides were grown in a Bacteroides-specific minimal media (MM) previously described50, supplemented with 0.5% (w/v) carbon source. E. coli S17-1 lambda pir, used for conjugation, was grown in Luria Bertoni (LB) media. When appropriate, antibiotics were used at the following concentrations: 100 µg/mL ampicillin (Amp; Enco, 14417), 25 µg/mL erythromycin (Erm; Enco, 16486) and 200 µg/mL gentamicin (Gent; Enco, 1405-41-0). We used 100 ng/mL anhydrotetracycline (aTc; Enco, 10009542) for the induction of CRISPR-FnCpf1.
Strain isolation
B. dorei, B. vulgatus and B. stercoris strains were isolated from breastfeeding infant stool samples as previously described71. Briefly, samples were transferred to −80 °C within 8 h. DNA was extracted using the DNeasy PowerSoil Pro Kit (QIAGEN) and used as a template for quantitative PCR (qPCR) reactions, with both control general 16S primers and Bacteroides-specific primers. Samples that passed the initial qPCR screening were plated on blood agar plates and then on selective bile esculin agar (BEA) plates and incubated anaerobically at 37 °C for 48 h. Single colonies were picked from the BEA plate, inoculated in liquid sBHI and the bacterial identification to the species level was performed using Sanger sequencing of the 16S region.
Growth curves
For each growth curve experiment, the strain of choice was plated on a sBHI agar plate that was incubated for 48 h at 37 °C. Two single colonies (biological replicates) were transferred to liquid sBHI media for overnight incubation. These cultures were then normalized to an optical density (OD600) of 1 and diluted 1:100 in MM (described above) with HMOs, lactose (positive control), or water (no carbon control) as the sole carbon source. The individual HMOs were received as a donation from DSM Nutritional Products Ltd (previously known as Glycom). All experiments were conducted with two biological replicates and three technical replicates per carbon source, under anaerobic conditions. Growth (OD600) was monitored every 30 min for 60 h, at 37 °C, using the Epoch2 microplate spectrometer (Agilent) with 96-well Costar culture plate (Corning, 3370). The optical density data was analyzed using R version 4.2.388. Comparisons between growth curve graphs were performed using a paired two-sided t-test at two timepoints: for B. dorei, at one and 24 hours; for B. uniformis and B. stercoris, at one and 50 h, as they grow relatively slower. The p-values for the latter timepoints are reported in the main figures (Figs. 2D, 3A) and the full statistical data at both timepoints, including the mean difference between the OD600 values of the growth curves, are reported in Supplementary Table 1. A mean difference >0.1 with a p-value < 0.01 was considered statistically significant.
CAZy enzymes per genome analysis
To calculate the average amount of sialidase and fucosidase enzymes encoded in the genome of B. dorei and B. vulgatus, we downloaded the CAZy database53 and counted the number of unique genes annotated as GH33 (sialidase) or GH29 and GH95 (fucosidases) in these species. We then divided this result by 28, the sum of 15 B. dorei and 13 B. vulgatus genomes deposited in CAZy.
RNAseq
Pure cultures of B. dorei growing on various carbon sources (in replicates) were harvested at log phase (OD600 ~ 0.7) using the Direct-zol™ RNA Miniprep Plus kit (Zymo Research, R2071). Agilent 2100 Bioanalyzer (Agilent Technologies) was employed for RNA quality control. Samples were processed according to a previously described protocol89 and sequenced in two separate pools: HMOs (2’-FL, DFL, 3’-SL, 6’-SL, LNT, LNnT) and glucose. Single-end 75 bp sequencing was performed on a NextSeq device and the data was deposited in the National Center for Biotechnology Information (NCBI) repository, accession numbers SRS11934615 (HMOs) and SRS11934616 (glucose)71.
The raw sequencing data were further filtered by trim_galore (https://github.com/FelixKrueger/TrimGalore) and classification of the B. dorei isolate to the strain level was achieved by alignment of reads using BLAST90. Genome mapping to B. dorei DSM 17855 genome (GenBank: CP046176.1) was performed by Bowtie291 and differential expression analysis between HMOs and glucose was carried out using featureCounts92 and DEseq293. DEseq2 uses the Wald test for differential expression analysis and the Benjamini and Hochberg method for multiple hypothesis correction. Significantly up-regulated genes were selected based on the p_adjusted <0.05 and log2 fold-change >1 parameters. Gene annotation of glycoside hydrolases (GHs) was downloaded from the CAZy (Carbohydrate Active enZYmes)53 database.
Conjugation
All plasmids (shuttle plasmids derivatives and pNBU2-ermGb derivatives) were introduced into Bacteroides via conjugation with E. coli S17-1, which harbors the conjugative machinery integrated onto chromosome94. To perform the conjugation, E. coli S17-1 and Bacteroides strains were grown until the early stationary phase, to an OD600 that approximately equals to 0.2–0.3. Next, cells were combined at a ratio of 1:5 (donor:recipient, v/v) and centrifuged at 5000×g for 15–20 min. The mixed cell pellet was resuspended in 100uL sBHI media and spotted onto a sBHI agar plate for a 20–24 h aerobic incubation at 37 °C. Next, we scraped the cell mass from the plate, resuspended it into 1 mL sBHI media, and performed 10-fold serial dilutions, up to ×105. 100 µL of each serially diluted culture were plated on sBHI plates containing Gent and Erm and incubated for 2 days anaerobically, to select for transconjugants, which are Bacteroides cells that have accepted the plasmid into the cell.
Plasmid construction
All strains and plasmids used in this study are listed in Supplementary Table 2 and the primers and genetic tools sequences that were used to construct them are listed in Supplementary Table 3. The shuttle plasmid for gene deletion, p_KO, is an all-in-one plasmid that encodes a Cas protein (FnCpf1), a custom single guide RNA (sgRNA), a repair template, and genetic elements for plasmid replication and conjugation69. The plasmid pB036 from Zheng et al. 2022 was kindly provided to us by the Lei Dai lab, Institute of Synthetic Biology, Chinese Academy of Sciences (Shenzhen). The plasmid pB036 was used as a template, which was linearized while replacing the gRNA, then assembled with the right and left flanking regions from B. dorei using Gibson Assembly (NEB, E2611) to create the circular p_KO. The gRNA was designed using CHOPCHOP95.
Complementation plasmids were based on the previously described pNBU2-ermGb plasmid family62. pNBU2_erm_P5E167 was built from pNBU2_erm_SIE1 (Addgene, #136356), which was linearized excluding the SIE1 cassette, then assembled with a synthetically synthesized sequence of the P_BfP5E1 promoter and ribosome binding site (IDT, Syntezza Bioscience) using Gibson Assembly. Genes were cloned by Gibson Assembly downstream to P_BfP5E1, in the SalI (NEB, R3138) and NcoI (NEB, M032) digested pNBU2_erm_P5E1. p_compFull, containing the four knocked-out genes under constitutive expression from P_BfP5E1, was used as a template for partial complementation experiments. Primers were designed to linearize p_compFull excluding portions of it, then the linearized PCR product was circularized by T4 DNA ligase (NEB, M0202). The ligated plasmids (p_compPartial1/2/3) were sequenced by Nanopore sequencing for verification and their structure is depicted in Fig. 3A.
Nanopore sequencing of isolates
The CRISPR-Cas12a genetic manipulation system69 requires the target strain to be sensitive to erythromycin, but the B. dorei isolate on which RNA-seq was performed was resistant. We isolated an additional B. dorei strain from a breastfeeding infant stool sample, which is erythromycin-susceptible. To validate the compatibility of the additional B. dorei isolate for genetic manipulation, we performed whole genome sequencing of both isolates. High-molecular-weight DNA was extracted from pure cultures of the isolates using the MagAttract HMW DNA Kit (Qiagen, 67563), and then processed using the Rapid Sequencing kit (ONT, SQK-RBK004) for library preparation. Whole genome nanopore sequencing was performed on a Minion device, using a R9.4.1 flow cell. Basecalling was performed using Guppy version 6.1.2 (ONT), quality assessment was performed using Nanoplot version 1.33.196, read filtering and trimming was performed using Porechop version 0.2.4 (https://github.com/rrwick/Porechop) and assembly was performed using Flye version 2.997. The final assembled genomes of both isolates were deposited in the NCBI database under GenBank assembly accessions GCA_030122685.1 (B. dorei 1) and GCA_030122665.1 (B. dorei 2). The average nucleotide identity (ANI) in Supplementary Fig. 4B was calculated using the OrthoANIu ANI calculator98. The synteny map in Supplementary Fig. 4B, comparing the genomes of the two B. dorei isolates to the type strain B. dorei DSM 17855, was constructed using the gggenomes package (https://github.com/thackl/gggenomes) under R version 4.2.388.
Markerless gene deletion Bacteroides
The plasmid for gene deletion, p_KO, was first transformed into E. coli S17-1 and then introduced into B. dorei using conjugation. Transconjugants were selected on sBHI + Gent + Erm as described above, then the Cas protein (FnCpf1) was induced to activate the gene editing process. Transconjugants were incubated overnight in sBHI + Gent + Erm + aTC, then plated on sBHI + aTC agar plates and incubated for two days until colonies were observed. Colonies were PCR-screened for the correct deletion using primers matching the genome of B. dorei, before and after the expected deletion site. The identified mutants were further verified using Sanger sequencing of the deletion region in B. dorei’s genome.
Plotting of PUL33 and the deleted area was performed using clinker99 version 0.0.27.
Complementation assays
The pNBU2-ermGb plasmid derivatives harbor the IntN2 tyrosine integrase, which mediates the recombination between the attN site on the plasmid and one of the two attBT sites at the 3’ end of the tRNASer genes (GKD17_04220 and GKD17_06425) on B. dorei’s chromosome100. Thus, the pNBU2-ermGb plasmid derivatives were integrated into the chromosome following conjugation.
Homology search of PUL33
Prior to the conjugation of B. uniformis CL03T12C37 and our B. stercoris isolate with genes from PUL33, in addition to phenotypically showing that they cannot grow on sialylated HMOs in vitro, we verified the absence of PUL33 homologs in their genome. For the type strain B. uniformis CL03T12C37, the sequence is publically available (GenBank accession number CP072255.1) and also deposited in the PULDB database74, with one GH33 gene encoded in PUL16, genes INE75_01450 to INE75_01456. The genomic context of GH33 and the gene structure of PUL16 in B. uniformis varies from that of B. dorei’s PUL33, with the B. uniformis cluster encoding SusC, SusD, GH33, an aldose 1-epimerase gene and three additional genes with unknown functions, lacking GH2 or GH20 genes present in B. dorei’s PUL33. Thus we concluded that the complementation of B. uniformis CL03T12C37 with genes from B. dorei’s PUL33 could provide capabilities which are non-overlapping with the existing repertoire of PULs encoded in B. uniformis. For B. stercoris, only one of six genomes deposited in PULDB encoded a GH33 gene74, so combined with the experiment inability of our isolate to utilize sialylated HMOs (Supplementary Fig. 6), we concluded that it doesn’t encode a cluster homologous to B. dorei’s PUL33.
Real-time qPCR (RT-qPCR)
Pure cultures of B. dorei and B. dorei Δ5k-PUL33::p_compPartial1/2/3 growing on lactose, 3’-SL and 6’-SL (in replicates) were harvested at log phase (OD ~ 0.7) using the Direct-zol™ RNA Miniprep Plus kit (Zymo Research, R2071). Reverse transcription of the total RNA samples to cDNA was performed using SuperScript™ III Reverse Transcriptase (Invitrogen, 18080093). RT-qPCR reaction volumes were 4 μL 0.1 ng/μL DNA, 0.5 μL 10 μM forward primer (500 nM), 0.5 μL 10 μM reverse primer (500 nM), 5 μL iTaq Universal SYBR Green Supermix (BioRad, 1725124). For SusC only, the primer concentration was found to be optimal at 100 nM. The amplification program consisted of (1) 95 °C for 30 s, (2) 95 °C for 10 s (3) 60 °C for 30 s (4) repeat 2–3 for 39 times. The fluorescent products were detected at the last step of each cycle using a CFX96 instrument (BioRad). Primer efficiencies were optimized and calculated using a standard curve and relative changes in gene expression were calculated compared to the housekeeping gene rpsL101, with the efficiency-corrected ΔCq method102.
Analysis of GH33 sequences
The phylogenetic tree of enzymes from the GH33 family in Supplementary Fig 8 was constructed using the SACCHARIS pipeline76, for randomly selected GH33 protein sequences deposited to the CAZy database in addition to B. dorei’s GH33 sialidase discussed in this manuscript and the B. thetaiotaomicron VPI 5482 crystallized GH33 protein59 (GenBank accession AAO75562.1; PDB code 4BBW). Alignment of protein sequences in Supp. Fig 9 was performed using Clustal Omega103, which utilizes the HHalign algorithm104 for calculating distances. The root mean square deviation (RMSD) values were calculated after superimposing the atomic coordinates of the proteins, using PyMOL’s (version 2.5.4105) ‘align’ function. Since the GH33 protein 3D structure of our B. dorei isolate and B. vulgatus ATCC 8482 (GenBank accession UBD82890.1) is predicted and not crystallized, we excluded residues 1–23 and 545–546, for which the predicted local distance difference test value (pLDDT) was lower than 70, indicating a low per-residue confidence score. On this account, the first 23 residues were also excluded from the crystallized B. thetaiotaomicron VPI 5482 GH33 protein sequence (its length is 544 so residues 545–546 were not relevant). In total, 521 residues from each of the three proteins were aligned for the RMSD calculation.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The RNA-seq data used in this study is deposited in the NCBI database under SRA accession codes SRS11934615 (HMOs) and SRS11934616 (glucose). The previously published differential expression analysis results, that were re-analyzed here, are available under DOI 10.3389/fcimb.2022.85412271. The Nanopore sequencing whole genome assembly data for B. dorei isolates generated in this study is deposited in the NCBI database under GenBank assembly accessions GCA_030122685.1 (B. dorei 1) and GCA_030122665.1 (B. dorei 2). Source data are provided with this paper.
References
Ferretti, P. et al. Mother-to-infant microbial transmission from different body sites shapes the developing infant gut microbiome. Cell Host Microbe 24, 133–145.e5 (2018).
Yassour, M. et al. Natural history of the infant gut microbiome and impact of antibiotic treatment on bacterial strain diversity and stability. Sci. Transl. Med. 8, 343ra81 (2016).
Kennedy, K. M. et al. Questioning the fetal microbiome illustrates pitfalls of low-biomass microbial studies. Nature 613, 639–649 (2023).
Lax, S. et al. Longitudinal analysis of microbial interaction between humans and the indoor environment. Science https://doi.org/10.1126/science.1254529 (2014)
Azad, M. B. et al. Gut microbiota of healthy Canadian infants: profiles by mode of delivery and infant diet at 4 months. CMAJ 185, 385–394 (2013).
Dominguez-Bello, M. G. et al. Delivery mode shapes the acquisition and structure of the initial microbiota across multiple body habitats in newborns. Proc. Natl Acad. Sci. USA 107, 11971–11975 (2010).
Madan, J. C. et al. Association of cesarean delivery and formula supplementation with the intestinal microbiome of 6-week-old infants. JAMA Pediatr. 170, 212–219 (2016).
Stewart, C. J. et al. Temporal development of the gut microbiome in early childhood from the TEDDY study. Nature 562, 583–588 (2018).
Bäckhed, F. et al. Dynamics and stabilization of the human gut microbiome during the first year of life. Cell Host Microbe 17, 690–703 (2015).
Engfer, M. B., 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. 71, 1589–1596 (2000).
Gnoth, M. J., Kunz, C., Kinne-Saffran, E. & Rudloff, S. Human milk oligosaccharides are minimally digested in vitro. J. Nutr. 130, 3014–3020 (2000).
Urashima, T., Fukuda, K. & Messer, M. Evolution of milk oligosaccharides and lactose: a hypothesis. Animal 6, 369–374 (2012).
Newburg, D. S. Innate immunity and human milk. J. Nutr. 135, 1308–1312 (2005).
Coppa, G. V. et al. Human milk oligosaccharides inhibit the adhesion to Caco-2 cells of diarrheal pathogens: Escherichia coli, Vibrio cholerae, and Salmonella fyris. Pediatr. Res. 59, 377–382 (2006).
Jantscher-Krenn, E. et al. Human milk oligosaccharides reduce Entamoeba histolytica attachment and cytotoxicity in vitro. Br. J. Nutr. 108, 1839–1846 (2012).
Guo, Y. et al. Structural basis for distinct ligand-binding and targeting properties of the receptors DC-SIGN and DC-SIGNR. Nat. Struct. Mol. Biol. 11, 591–598 (2004).
Charbonneau, M. R. et al. Sialylated milk oligosaccharides promote microbiota-dependent growth in models of infant undernutrition. Cell 164, 859–871 (2016).
Cowardin, C. A. et al. Mechanisms by which sialylated milk oligosaccharides impact bone biology in a gnotobiotic mouse model of infant undernutrition. Proc. Natl Acad. Sci. USA 116, 11988–11996 (2019).
Hauser, J. et al. Sialylated human milk oligosaccharides program cognitive development through a non-genomic transmission mode. Mol. Psychiatry 26, 2854–2871 (2021).
Buffie, C. G. & Pamer, E. G. Microbiota-mediated colonization resistance against intestinal pathogens. Nat. Rev. Immunol. 13, 790–801 (2013).
McLaren, M. R. & Callahan, B. J. Pathogen resistance may be the principal evolutionary advantage provided by the microbiome. Philos. Trans. R. Soc. Lond. B Biol. Sci. 375, 20190592 (2020).
Letten, A. D., Baumgartner, M., Pfrunder-Cardozo, K. R., Levine, J. M. & Hall, A. R. Human-associated microbiota suppress invading bacteria even under disruption by antibiotics. ISME J. 15, 2809–2812 (2021).
Letten, A. D., Hall, A. R. & Levine, J. M. Using ecological coexistence theory to understand antibiotic resistance and microbial competition. Nat. Ecol. Evol. 5, 431–441 (2021).
Walsh, C., Lane, J. A., van Sinderen, D. & Hickey, R. M. Human milk oligosaccharides: shaping the infant gut microbiota and supporting health. J. Funct. Foods 72, 104074 (2020).
Smith-Brown, P., Morrison, M., Krause, L. & Davies, P. S. W. Mothers secretor status affects development of childrens microbiota composition and function: a Pilot Study. PLoS ONE 11, e0161211 (2016).
Lawson, M. A. E. et al. Breast milk-derived human milk oligosaccharides promote Bifidobacterium interactions within a single ecosystem. ISME J. 14, 635–648 (2020).
Davis, J. C. C. et al. Identification of oligosaccharides in feces of breast-fed infants and their correlation with the gut microbial community. Mol. Cell. Proteom. 15, 2987–3002 (2016).
Bai, Y. et al. Fucosylated human milk oligosaccharides and N-glycans in the milk of Chinese mothers regulate the gut microbiome of their breast-fed infants during different lactation stages. mSystems 3, e00206–18 (2018).
Liu, F. et al. Maternal fucosyltransferase 2 status associates with the profiles of human milk oligosaccharides and the fecal microbiota composition of breastfed infants. J. Agric. Food Chem. 69, 3032–3043 (2021).
Laursen, M. F. et al. Maternal milk microbiota and oligosaccharides contribute to the infant gut microbiota assembly. ISME Commun. 1, 1–13 (2021).
Sela, D. A. & Mills, D. A. Nursing our microbiota: molecular linkages between bifidobacteria and milk oligosaccharides. Trends Microbiol. 18, 298–307 (2010).
Lewis, Z. T. et al. Maternal fucosyltransferase 2 status affects the gut bifidobacterial communities of breastfed infants. Microbiome 3, 13 (2015).
Sela, D. A. et al. The genome sequence of Bifidobacterium longum subsp. infantis reveals adaptations for milk utilization within the infant microbiome. Proc. Natl Acad. Sci. USA 105, 18964–18969 (2008).
Dethlefsen, L., McFall-Ngai, M. & Relman, D. A. An ecological and evolutionary perspective on human–microbe mutualism and disease. Nature 449, 811–818 (2007).
LoCascio, R. G., Desai, P., Sela, D. A., Weimer, B. & Mills, D. A. Broad conservation of milk utilization genes in Bifidobacterium longum subsp. infantis as revealed by comparative genomic hybridization. Appl. Environ. Microbiol. 76, 7373–7381 (2010).
Zabel, B. E. et al. Strain-specific strategies of 2′-fucosyllactose, 3-fucosyllactose, and difucosyllactose assimilation by Bifidobacterium longum subsp. infantis Bi-26 and ATCC 15697. Sci. Rep. 10, 15919 (2020).
Garrido, D. et al. Comparative transcriptomics reveals key differences in the response to milk oligosaccharides of infant gut-associated bifidobacteria. Sci. Rep. 5, 13517 (2015).
Garrido, D. et al. A novel gene cluster allows preferential utilization of fucosylated milk oligosaccharides in Bifidobacterium longum subsp. longum SC596. Sci. Rep. 6, 35045 (2016).
Ruiz-Moyano, S. et al. Variation in consumption of human milk oligosaccharides by infant gut-associated strains of Bifidobacterium breve. Appl. Environ. Microbiol. 79, 6040–6049 (2013).
Sela, D. A. et al. Bifidobacterium longum subsp. infantis ATCC 15697 α-fucosidases are active on fucosylated human milk oligosaccharides. Appl. Environ. Microbiol. 78, 795–803 (2012).
Kim, J.-H. et al. Proteomic analysis of Bifidobacterium longum subsp. infantis reveals the metabolic insight on consumption of prebiotics and host glycans. PLoS ONE 8, e57535 (2013).
Casaburi, G. et al. Metagenomic insights of the infant microbiome community structure and function across multiple sites in the United States. Sci. Rep. 11, 1472 (2021).
Vatanen, T. et al. Genomic variation and strain-specific functional adaptation in the human gut microbiome during early life. Nat. Microbiol. 4, 470–479 (2019).
Borewicz, K. et al. The association between breastmilk oligosaccharides and faecal microbiota in healthy breastfed infants at two, six, and twelve weeks of age. Sci. Rep. 10, 4270 (2020).
Wang, M. et al. Fecal microbiota composition of breast-fed infants is correlated with human milk oligosaccharides consumed. J. Pediatr. Gastroenterol. Nutr. 60, 825–833 (2015).
Borewicz, K., Gu, F., Saccenti, E. & Arts, I. C. W. Correlating infant fecal microbiota composition and human milk oligosaccharide consumption by microbiota of 1‐month‐old breastfed infants. Mol. Nutr. Food Res. 63, e1801214 (2019).
Xu, J. et al. A genomic view of the human–Bacteroides thetaiotaomicron symbiosis. Science 299, 2074–2076 (2003).
Bjursell, M. K., Martens, E. C. & Gordon, J. I. Functional genomic and metabolic studies of the adaptations of a prominent adult human gut symbiont, Bacteroides thetaiotaomicron, to the suckling period. J. Biol. Chem. 281, 36269–36279 (2006).
Hooper, L. V., Midtvedt, T. & Gordon, J. I. How host-microbial interactions shape the nutrient environment of the mammalian intestine. Annu. Rev. Nutr. 22, 283–307 (2002).
Martens, E. C., Chiang, H. C. & Gordon, J. I. Mucosal glycan foraging enhances fitness and transmission of a saccharolytic human gut bacterial symbiont. Cell Host Microbe 4, 447–457 (2008).
Porter, N. T. & Martens, E. C. The critical roles of polysaccharides in gut microbial ecology and physiology. Annu. Rev. Microbiol. 71, 349–369 (2017).
Lapébie, P., Lombard, V., Drula, E., Terrapon, N. & Henrissat, B. Bacteroidetes use thousands of enzyme combinations to break down glycans. Nat. Commun. 10, 2043 (2019).
Lombard, V., Ramulu, Golaconda, Drula, H., Coutinho, E. & Henrissat, P. M. B. The carbohydrate-active enzymes database (CAZy) in 2013. Nucleic Acids Res. 42, D490–D495 (2014).
Wexler, A. G. & Goodman, A. L. An insider’s perspective: bacteroides as a window into the microbiome. Nat. Microbiol. 2, 17026 (2017).
Glowacki, R. W. P. et al. A ribose-scavenging system confers colonization fitness on the human gut symbiont Bacteroides thetaiotaomicron in a diet-specific manner. Cell Host Microbe 27, 79–92.e9 (2020).
Marcobal, A. et al. Bacteroides in the infant gut consume milk oligosaccharides via mucus-utilization pathways. Cell Host Microbe 10, 507–514 (2011).
Salli, K. et al. Selective utilization of the human milk oligosaccharides 2’-fucosyllactose, 3-fucosyllactose, and difucosyllactose by various probiotic and pathogenic bacteria. J. Agric. Food Chem. 69, 170–182 (2021).
Moya-Gonzálvez, E. M. et al. Infant gut microbial metagenome mining of α-l-fucosidases with activity on fucosylated human milk oligosaccharides and glycoconjugates. Microbiol. Spectr. 10, e0177522 (2022).
Park, K.-H. et al. Structural and biochemical characterization of the broad substrate specificity of Bacteroides thetaiotaomicron commensal sialidase. Biochim. Biophys. Acta 1834, 1510–1519 (2013).
Bell, A., Severi, E., Owen, C. D., Latousakis, D. & Juge, N. Biochemical and structural basis of sialic acid utilization by gut microbes. J. Biol. Chem. 299, 102989 (2023).
Lim, B., Zimmermann, M., Barry, N. A. & Goodman, A. L. Engineered regulatory systems modulate gene expression of human commensals in the gut. Cell 169, 547–558.e15 (2017).
Koropatkin, N. M., Martens, E. C., Gordon, J. I. & Smith, T. J. Starch catabolism by a prominent human gut symbiont is directed by the recognition of amylose helices. Structure 16, 1105–1115 (2008).
Liu, H. et al. Functional genetics of human gut commensal Bacteroides thetaiotaomicron reveals metabolic requirements for growth across environments. Cell Rep. 34, 108789 (2021).
Béchon, N. et al. Bacteroides thetaiotaomicron uses a widespread extracellular DNase to promote bile-dependent biofilm formation. Proc. Natl. Acad. Sci. USA 119, e2111228119 (2022).
Hoces, D. et al. Fitness advantage of Bacteroides thetaiotaomicron capsular polysaccharide in the mouse gut depends on the resident microbiota. Elife 12, e81212 (2023).
Baughn, A. D. & Malamy, M. H. A mitochondrial-like aconitase in the bacterium Bacteroides fragilis: implications for the evolution of the mitochondrial Krebs cycle. Proc. Natl Acad. Sci. USA 99, 4662–4667 (2002).
Bencivenga-Barry, N. A., Lim, B., Herrera, C. M., Trent, M. S. & Goodman, A. L. Genetic manipulation of wild human gut bacteroides. J. Bacteriol. 202, e00544–19 (2020).
García-Bayona, L. & Comstock, L. E. Streamlined genetic manipulation of diverse bacteroides and parabacteroides isolates from the human gut microbiota. MBio 10, e01762–19 (2019).
Zheng, L. et al. CRISPR/Cas-based genome editing for human gut commensal bacteroides species. ACS Synth. Biol. 11, 464–472 (2022).
Feng, J. et al. Polysaccharide utilization loci in Bacteroides determine population fitness and community-level interactions. Cell Host Microbe 30, 200–215.e12 (2022).
Kijner, S., Cher, A. & Yassour, M. The infant gut commensal Bacteroides dorei presents a generalized transcriptional response to various human milk oligosaccharides. Front. Cell. Infect. Microbiol. 12, 854122 (2022).
García-López, M. et al. Analysis of 1,000 type-strain genomes improves taxonomic classification of bacteroidetes. Front. Microbiol. 10, 2083 (2019).
Ioannou, A., Knol, J. & Belzer, C. Microbial glycoside hydrolases in the first year of life: an analysis review on their presence and importance in infant gut. Front. Microbiol. 12, 631282 (2021).
Terrapon, N. et al. PULDB: the expanded database of Polysaccharide Utilization Loci. Nucleic Acids Res. 46, D677–D683 (2018).
Cullen, T. W. et al. Gut microbiota. Antimicrobial peptide resistance mediates resilience of prominent gut commensals during inflammation. Science 347, 170–175 (2015).
Jones, D. R. et al. SACCHARIS: an automated pipeline to streamline discovery of carbohydrate active enzyme activities within polyspecific families and de novo sequence datasets. Biotechnol. Biofuels 11, 27 (2018).
Almagro-Moreno, S. & Boyd, E. F. Insights into the evolution of sialic acid catabolism among bacteria. BMC Evol. Biol. 9, 118 (2009).
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Mills, D. A., German, J. B., Lebrilla, C. B. & Underwood, M. A. Translating neonatal microbiome science into commercial innovation: metabolism of human milk oligosaccharides as a basis for probiotic efficacy in breast-fed infants. Gut Microbes 15, 2192458 (2023).
Shani, G. et al. Fucosylated human milk oligosaccharide foraging within the species Bifidobacterium pseudocatenulatum is driven by glycosyl hydrolase content and specificity. Appl. Environ. Microbiol. 88, e0170721 (2022).
Kiyohara, M. et al. An exo-alpha-sialidase from bifidobacteria involved in the degradation of sialyloligosaccharides in human milk and intestinal glycoconjugates. Glycobiology 21, 437–447 (2011).
Juge, N., Tailford, L. & Owen, C. D. Sialidases from gut bacteria: a mini-review. Biochem. Soc. Trans. 44, 166–175 (2016).
Phansopa, C. et al. Structural and functional characterization of NanU, a novel high-affinity sialic acid-inducible binding protein of oral and gut-dwelling Bacteroidetes species. Biochem. J. 458, 499 (2014).
Condemine, G., Berrier, C., Plumbridge, J. & Ghazi, A. Function and expression of an N-acetylneuraminic acid-inducible outer membrane channel in Escherichia coli. J. Bacteriol. 187, 1959 (2005).
Luo, Y., Li, S. C., Chou, M. Y., Li, Y. T. & Luo, M. The crystal structure of an intramolecular trans-sialidase with a NeuAc alpha2->3Gal specificity. Structure 6, 521–530 (1998).
Buchini, S., Buschiazzo, A. & Withers, S. G. A new generation of specific Trypanosoma cruzi trans-sialidase inhibitors. Angew. Chem. Int. Ed. Engl. 47, 2700–2703 (2008).
Suzuki, Y. et al. Sialic acid species as a determinant of the host range of influenza A viruses. J. Virol. 74, 11825–11831 (2000).
R Foundation for Statistical Computing, Vienna, Austria. R Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, Austria, 2023)
Shishkin, A. A. et al. Simultaneous generation of many RNA-seq libraries in a single reaction. Nat. Methods 12, 323–325 (2015).
Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local alignment search tool. J. Mol. Biol. 215, 403–410 (1990).
Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359 (2012).
Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014).
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Simon, R., Priefer, U. & Pühler, A. A broad host range mobilization system for in vivo genetic engineering: transposon mutagenesis in Gram negative bacteria. Biotechnology 1, 784–791 (1983).
Labun, K. et al. CHOPCHOP v3: expanding the CRISPR web toolbox beyond genome editing. Nucleic Acids Res. 47, W171–W174 (2019).
De Coster, W., D’Hert, S., Schultz, D. T., Cruts, M. & Van Broeckhoven, C. NanoPack: visualizing and processing long-read sequencing data. Bioinformatics 34, 2666–2669 (2018).
Kolmogorov, M., Yuan, J., Lin, Y. & Pevzner, P. A. Assembly of long, error-prone reads using repeat graphs. Nat. Biotechnol. 37, 540–546 (2019).
Yoon, S.-H., Ha, S.-M., Lim, J., Kwon, S. & Chun, J. A large-scale evaluation of algorithms to calculate average nucleotide identity. Antonie Van Leeuwenhoek 110, 1281–1286 (2017).
Gilchrist, C. L. M. & Chooi, Y.-H. clinker & clustermap.js: automatic generation of gene cluster comparison figures. Bioinformatics 37, 2473–2475 (2021).
Wang, J., Shoemaker, N. B., Wang, G. R. & Salyers, A. A. Characterization of a Bacteroides mobilizable transposon, NBU2, which carries a functional lincomycin resistance gene. J. Bacteriol. 182, 3559–3571 (2000).
Ben-Assa, N. et al. Analysis of a phase-variable restriction modification system of the human gut symbiont Bacteroides fragilis. Nucleic Acids Res. 48, 11040–11053 (2020).
Bookout, A. L., Cummins, C. L., Mangelsdorf, D. J., Pesola, J. M. & Kramer, M. F. High-throughput real-time quantitative reverse transcription PCR. Curr. Protoc. Mol. Biol. Chapter 15, Unit 15.8 (2006).
Sievers, F. et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol. Syst. Biol. 7, 539 (2011).
Söding, J. Protein homology detection by HMM–HMM comparison. Bioinformatics 21, 951–960 (2005).
Schrödinger, L. L. C. The PyMOL Molecular Graphics System, Version 1.8. Schrödinger, LLC (2015).
Acknowledgements
We thank the Lei Dai lab and specifically Liggang Zheng, for plasmids and assistance with setting up the genetic manipulation system; The Goodman lab and Natasha A. Bencivenga-Barry, for the pNBU_erm plasmid; The DSM Nutritional Products Ltd (previously known as Glycom) for the HMO donation; Naama Geva-Zatorsky and her lab members for the B. uniformis CL03T12C37 strain; Illan Rosenshine’s lab for E. coli S17-1 lambda pir strain; Seth Rakoff-Nahoum for fruitful discussions and experimental feedback; Dina Schneidman for guidance regarding the protein structure analysis. This project, S.K. and M.Y. were supported by an Israeli Ministry of Health grant (3-18355), and an Israel Science Foundation grant (2660/18). A.F. is supported by The Abisch-Frenkel Faculty Development Lectureship. M.Y. is the Rosalind, Paul and Robin Berlin Faculty Development Chair in Perinatal Research, also supported by the Azrieli Foundation.
Author information
Authors and Affiliations
Contributions
S.K. established the experimental system & performed experiments and analysis. D.E. and S.S. assisted in establishing the experimental system. A.F. guided the genetic manipulation work. M.Y. guided the work and the computational analysis. S.K. and M.Y. wrote the manuscript. All authors contributed to the article and approved the submitted version.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Marco Jost, and the other, anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Kijner, S., Ennis, D., Shmorak, S. et al. CRISPR-Cas-based identification of a sialylated human milk oligosaccharides utilization cluster in the infant gut commensal Bacteroides dorei. Nat Commun 15, 105 (2024). https://doi.org/10.1038/s41467-023-44437-y
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-023-44437-y
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
