Cobamide Scavenging and Import via a Multiprotein Extracellular Relay in Bacteroides thetaiotaomicron

Abstract

Cobamides, the structurally diverse family of cobalt‐containing cofactors, are required by most bacteria yet synthesized by a subset of prokaryotes, making cobamide acquisition a major determinant of microbial fitness. Bacteroides thetaiotaomicron, an abundant commensal of the human gut, is auxotrophic for cobamides and relies on environmental scavenging to support its metabolism. To meet this demand in a competitive and diverse cobamide environment of the gut, B . thetaiotaomicron employs a B₁₂‐uptake (Btu) system that extends far beyond the well‐studied Escherichia coli Btu system. Rather than relying on a single outer membrane (OM) transporter, B . thetaiotaomicron deploys a multiprotein extracellular relay in which surface‐exposed BtuJ lipoproteins capture cobamides extracellularly and transfer them to BtuH and BtuG lipoproteins at the OM. BtuG forms a stable complex with the TonB‐dependent transporter BtuB, coupling cobamide capture to periplasmic translocation, and the inner membrane BtuFCD complex completes cytoplasmic delivery. BtuJ‐enriched OM vesicles further extend cobamide capture into the extracellular environment, and cobamide‐dependent riboswitches couple Btu protein expression to the availability and chemical identity of intracellular cobamides. We review the structural and mechanistic evidence for each stage of this relay, propose a model for cobamide import, and identify outstanding questions that will illuminate cobamide scavenging across gut bacteria.

Bacteroides thetaiotaomicron competes with gut microbes and the host for a limited cobamide pool. To meet this demand, it deploys surface associated lipoproteins (BtuJ, BtuH, BtuK, BtuG), three BtuB paralogs, and extracellular vesicles that extend cobamide capture beyond the cell surface. This review examines the structural and mechanistic evidence for each stage, proposes a model for cobamide import, and identifies outstanding questions.

1. Introduction

Access to nutrients is a primary determinant of microbial fitness, and microorganisms have evolved dedicated strategies to capture, sequester, and deliver scarce nutrients from competitive environments [1, 2, 3]. For nutrients that are both rare and chemically reactive, this machinery must accomplish more than simple transport: it must intercept the molecule before competitors do, protect it against adventitious reactions, and deliver it to the target enzymes that require it [4, 5, 6]. These demands have driven the evolution of multiprotein capture and relay systems in which dedicated proteins sequester the nutrient extracellularly, pass it directionally through successive protein partners of increasing affinity, and ultimately deliver it to its intracellular target [7, 8].

Cobamides, the structurally diverse family of cobalt‐containing cofactors that includes vitamin B₁₂, represent a class of nutrients whose acquisition shapes fundamental biological processes at the cellular, organismal, and community level [3]. Cobamides are synthesized de novo by a subset of bacteria and archaea, but required by organisms across all domains of life that possess cobamide‐dependent enzymes [1, 3]. In bacteria, B₁₂‐dependent enzymes catalyze chemically challenging reactions, and the genes encoding these enzymes are present in 86% of sequenced bacteria [3, 9]. Forty‐two percent of human gut microbiome (HGM) genomes encode the biosynthetic capacity to produce cobamides de novo, making cobamide availability a likely mediator of microbial community composition and function in the gut [1, 2, 3, 10, 11, 12].

Bacteria have evolved a dedicated multicomponent machinery to meet this cobamide demand. In Gram‐negative bacteria, cobamide import has long been understood through the Escherichia coli B₁₂ uptake (Btu) system [13, 14, 15]. The E. coli Btu system has been characterized in molecular detail and serves as the mechanistic reference point for cobamide import across Gram‐negative bacteria. What it does not explain is how cobamide acquisition occurs in the competitive, structurally heterogeneous cobamide environment of the gut. Bacteroides thetaiotaomicron, one of the most abundant commensals of the gut [16], illustrates the limitations of this paradigm most vividly. Unable to synthesize cobamides de novo, B. thetaiotaomicron relies on environmental scavenging to support its B₁₂ needs and cobamide acquisition by B. thetaiotaomicron is a direct determinant of competitive fitness in the gut [9, 17]. Rather than a single outer membrane (OM) transporter, B. thetaiotaomicron has evolved a multiprotein extracellular capture and relay system encoded across four operons, in which dedicated B₁₂‐binding lipoproteins capture cobamides before they reach the OM transporters, traffic them through successive protein partners, and deliver them to the cytoplasm [9, 18, 19, 20].

This review describes the structural and mechanistic evidence for each stage of this relay, compares the B. thetaiotaomicron system to the E. coli paradigm to identify what is conserved, what is distinctly elaborated, and proposes a model for cobamide import in gut Bacteroidota. Most gut bacteria require cobamides, yet the molecular mechanisms governing cobamide import and selectivity remain largely uncharacterized. Understanding these processes in the model gut microbe B. thetaiotaomicron will inform strategies to manipulate microbiome composition [2, 3].

2. Cobamide Structure and Diversity in the Gut

Cobamides are organometallic cofactors defined by the cobalt center of the corrin ring that vary in their lower axial ligand (the nucleotide base attached via the aminopropanol–phosphate‍–‍ribose linker) (Figure 1A) [21]. The cobalt atom is equatorially coordinated to the four pyrrole nitrogen atoms in the corrin ring and accepts upper (β) and lower (α) axial ligands. The β‐axial ligand varies and includes cyanocobalamin (CNCbl), aquocobalamin (OH₂Cbl), adenosylcobalamin (AdoCbl), and methylcobalamin (MeCbl) (Figure 1B). AdoCbl and MeCbl are catalytically active forms utilized for radical and methyl‐transfer reactions, respectively (Figure 1B). The identity of the lower axial ligand also varies, and multiple forms have been identified in nature (Figure 1C). In cobamides, the most common α ligands can be classified as benzimidazoles, purines, and phenolics (Figure 1C) [2]. Cobalamin (Cbl) is the specific cobamide bearing 5,6‐dimethylbenzimidazole (DMB) as its lower ligand (Figure 1A). Other naturally occurring cobamides include pseudocobalamin (adenine) or various phenolic/purine bases (Figure 1C). Corrinoids that lack the nucleotide loop, and thus also the lower axial ligand, are termed cobinamides (Cbi) (Figure 1A).

FIGURE 1.

(A) Structure of cobalamin. (B) Upper axial ligands found in cobamides (red). (C) Some lower axial ligands found in cobamides (blue). (D) Cobamides exist in the “base‐on,” “base‐off,” or “base‐off, His‐on conformation”.

In solution, the identity of the α ligand can determine whether a cobamide adopts the base‐on or base‐off coordination state. In the base‐on state, the lower axial ligand nitrogen coordinates with the cobalt atom, completing an octahedral 6‐c geometry (Figure 1D). In the base‐off state, this intramolecular coordination is absent, and the lower axial position is unoccupied or occupied by an exogenous water ligand. Benzimidazolyl and purinyl cobamides can adopt either base‐on or base‐off coordination state depending on pH and protein environment, while phenolyl cobamides are constitutively base‐off (Figure 1D). This distinction may have direct functional consequences for B₁₂‐dependent catalysis. In base‐on enzymes, including class II ribonucleotide reductases (NrdJ), the identity of the lower axial ligand may have a direct influence on the catalytic activity of the enzyme. Phenolyl cobamides do not support NrdJ‐dependent growth, consistent with the base‐on catalytic requirement [22]. In base‐off/His‐on enzymes, including methionine synthase (MetH) and methylmalonyl‐CoA mutase (MCM), a histidine from a conserved cobamide‐binding domain displaces the lower axial ligand upon cofactor binding (Figure 1D) [23, 24, 25]. MCM can accommodate a range of cobamide derivatives, but cofactor binding affinity and activity nonetheless exhibit lower ligand‐dependent selectivity, indicating that the lower axial ligand retains functionally significant interactions with the protein even in the base‐off binding mode [25, 26].

Notably, B. thetaiotaomicron encodes B₁₂‐dependent enzymes NrdJ, MetH, and MutAB (MCM), as well as, a methylamine/trimethylamine methyltransferase and HpnR/Hyp, a B₁₂‐binding domain/radical SAM domain protein [9, 16, 27, 28, 29, 30]. Furthermore, B₁₂‐dependent riboswitches couple Btu protein expression to intracellular cobamide availability (see Section 9) [9, 31, 32]. Thus, the identity of the lower axial ligand is likely to influence B. thetaiotaomicron physiology at multiple levels simultaneously. Determining which cobamides are competent cofactors for B₁₂‐dependent enzymes and governing whether intracellular cobamides trigger riboswitch‐mediated repression of the Btu system will enable linking the chemical identity of the environmental corrinoid pool directly to both metabolic capacity and cobamide acquisition.

Human fecal samples contain a structurally diverse cobamide pool. LC‐MS analysis of samples from 20 subjects identified Cbi and seven structurally distinct cobamides, with a mean total corrinoid content of approximately 1.3 μg per gram wet weight of feces (~1 µM, assuming fecal density ≈1 g/mL and an average cobamide MW ~1350 g/mol) [33, 34]. Cbl, despite its central role in human metabolism, constitutes only ~1.4% of this pool. Purinyl cobamides dominate, with 2‐methyladenylcobamide, adeninylcobamide, and 2‐methylthioadenylcobamide collectively accounting for the majority of detected corrinoids; p‐cresolylcobamide represents an additional ~16% of the total corrinoid pool, while benzimidazolyl species other than Cbl are present at less than 2% [33, 34]. This corrinoid landscape reflects the cumulative biosynthetic output of resident microbes, since de novo cobamide biosynthesis is restricted to an estimated 42% of HGM genomes, an asymmetry that makes cobamide sharing a predicted mediator of community dynamics [1, 2, 3, 10, 11]. For B. thetaiotaomicron, which cannot synthesize cobamides de novo, this structurally heterogeneous cobamide environment defines the selective pressure under which its expanded Btu system has evolved.

3. Outer Membrane Cobamide Import by the Btu System

Cobamide import in Gram‐negative bacteria is understood primarily through genetic, biochemical, and structural studies of the E. coli Btu system, which defines a conserved three‐component paradigm: BtuB, a TonB‐dependent OM transporter that captures environmental cobamide and uses proton motive force to drive periplasmic translocation; BtuF, a soluble periplasmic binding protein that captures cobamide and delivers it to the inner membrane (IM) ABC transporter by a docking‐dependent, allosterically driven mechanism; and BtuC₂D₂, which uses ATP hydrolysis to complete cytoplasmic delivery (Figure 2A) [13, 14, 35, 36].

FIGURE 2.

Btu operons in E. coli and B. thetaiotaomicron. (A) E. coli expresses a single copy of BtuB, BtuF, BtuC, and BtuD. (B) B. thetaiotaomicron encodes four operons with multiple paralogs of BtuB, BtuF, BtuC, and BtuD as well as additional surface‐associated lipoproteins BtuJ, BtuH, BtuG, BtuK, and BtuL that assist in cobamide capture and import. Operons 1–3 are regulated by a B₁₂‐dependent riboswitch‐encoded downstream of a promoter sequence (red triangle).

B. thetaiotaomicron inhabits the human gut in which competing organisms and host‐secreted B₁₂‐binding proteins contest the same corrinoid pool. In addition to the intense competition, the structural diversity of cobamides demands a capture system of broad specificity [33, 34]. The expanded Btu system that has evolved in response retains the E. coli paradigm but surrounds it with surface‐associated lipoproteins and extracellular capture proteins that have no counterpart in E. coli. Dedicated B₁₂‐binding proteins (BtuJ, BtuH, BtuK, and BtuG) capture cobamides before they reach the OM transporter BtuB (Figure 2B). In addition to this extracellular capture machinery, B. thetaiotaomicron encodes three BtuB paralogs that confer functionally distinct corrinoid preferences rather than a single promiscuous transporter, two IM BtuFCD‐class transporter complexes rather than one, and bacterial extracellular vesicles (BEVs) that extend BtuJ‐mediated cobamide capture beyond the immediate cell surface [9, 20]. The individual components of this relay are described below, with comparison to the E. coli system at each stage to highlight what is conserved, what is elaborated, and what remains mechanistically unresolved.

3.1. BtuB: A TonB‐Dependent OM Cobamide Transporter

The E. coli BtuB is a 22‐stranded transmembrane β‐barrel embedded in the OM, with a globular N‐terminal plug domain that occludes the barrel lumen in the resting state (Figure 3A) [39]. The extracellular face presents eleven loops of varying length; loops L2–L5 and L9–L11, together with residues from the plug domain apices, form a composite B₁₂‐binding site (Figure 3A) [39, 40]. B₁₂ binding by E. coli BtuB is calcium‐dependent with two calcium ions coordinated to loops L3 and L4, organizing the extracellular pocket and enabling CNCbl binding with an estimated K _D of ~0.3 nM [39]. B₁₂ binding induces conformational changes in loops L2–L5 that propagate through the plug domain to the TonB box, a conserved N‐terminal periplasmic sequence whose reorientation upon substrate binding is the critical transmembrane signal that enables TonB engagement [39]. The crystal structure of E. coli BtuB•TonB reveals that the TonB box adopts a β‐strand that intercalates into the TonB C‐terminal β‐sheet, forming an interprotein β‐sheet (Figure 3A). This structural observation directly supports a mechanical pulling model in which TonB‐driven tension on the TonB box propagates through the plug domain, disrupting its electrostatic contacts with the barrel wall and opening the translocation pathway [37]. Specifically, steered molecular dynamics (SMD) simulations [41] and single‐molecule atomic force microscopy (AFM) [42] data suggest that this mechanical force causes a partial unfolding of the plug domain to form a transient channel. Once displaced, cobamide passes through the barrel lumen into the periplasm for capture by BtuF [36].

FIGURE 3.

(A) Structure of the E. coli BtuB•TonB (PDB code: 2GSK) [37]. The OM β‐barrel (gray ribbon) in complex with TonB (blue ribbon). TonB Box (yellow ribbon) forms an interprotein β‐sheet. The plug domain (orange ribbon) and ordering of surface‐exposed loops upon calcium binding (green spheres) form the cobamide‐binding site (orange sticks). (B) Structure of B. thetaiotaomicron BtuB3•BtuG3•CNCbl (PDB code: 8P98) [38] shows BtuG3 (green ribbon) forms a lid for the β‐barrel with its N‐terminus anchored to the OM (green surface). CNCbl (orange sticks) binds to BtuG. An overlay of the EL8 loop in BtuB3•BtuG3 (gray) versus BtuB2•BtuG2 (PDB code: 8BMX) [19] shows that the EL8 loop undergoes a conformational change to accommodate cobamide binding.

3.2. BtuB•BtuG Form a Stable Complex

The most fundamental architectural departure of the B. thetaiotaomicron OM layer from E. coli is the mandatory involvement of a surface‐exposed lipoprotein BtuG in the OM transport step. Deletion of btuG2 from an operon 2‐only strain abolishes cobamide accumulation and causes rapid competitive exclusion from the guts of gnotobiotic mice, phenocopying deletion of btuB2 itself [18]. BtuG is therefore not an accessory protein but an obligate component of the OM cobamide capture and translocation machinery.

In B. thetaiotaomicron, each BtuB paralog forms a stable complex with its cognate BtuG lipoprotein [19]. Crystal structures of BtuB2G2 and cryo‐EM structures of BtuB1G1, together with the CNCbl‐bound structure of BtuB3G3, reveal a “pedal bin”‐like architecture in which BtuG caps the extracellular face of BtuB, with the β‐propeller lid anchored by the N‐terminal lipid modification to the OM and positioned directly over the extracellular loops of the BtuB barrel (Figure 3B) [19]. This arrangement is reminiscent of the SusCD systems that B. thetaiotaomicron uses for glycan uptake, in which a SusD caps a SusC TonB‐dependent transporter [19, 43, 44].

A conserved BtuB extracellular loop, EL8, is hypothesized to play a central mechanistic role in cobamide transfer from BtuG to BtuB (Figure 3B). In the apo BtuB2G2 crystal structure, the tip of the EL8 loop of BtuB2 occupies the B₁₂‐binding site of BtuG2, sterically blocking cobamide from binding (Figure 3B, blue ribbon) [19]. Molecular dynamics (MD) simulations suggest that BtuG rotates away from BtuB for cofactor loading, moving EL8 out of the binding site and exposing the negatively charged β‐propeller face to solvent [19]. Post cobamide capture by BtuG in the open state, lid closure acts as a spring‐loaded trigger: as BtuG closes over BtuB, EL8 re‐enters the BtuG binding pocket and, by steric clash, destabilizes the cofactor and drives its transfer to BtuB for TonB‐dependent periplasmic translocation [19]. The BtuB3G3•CNCbl structure directly captures two intermediate states of this transfer: in state 1, EL8 is displaced and CNCbl occupies the BtuG binding site (Figure 3B, dark gray ribbon); in state 2, EL8 occupies an intermediate position, consistent with the transfer mechanism proposed from the comparison between apo BtuB2G2 (blue ribbon) and B₁₂‐bound BtuB3G3 (gray ribbon) (Figure 3B) [19]. B₁₂ binding to BtuB is proposed to trigger conformational changes in the plug domain that expose the TonB box in the periplasm, enabling TonB engagement and proton motive force‐driven translocation, the same TonB‐dependent mechanism as in E. coli BtuB, but now initiated downstream of the BtuG capture and transfer step rather than by direct binding of cobamide to the E. coli BtuB extracellular loops [19, 36].

3.3. BtuB Paralogs Display Cobamide Specificity

B. thetaiotaomicron BtuB1 and BtuB3 confer distinct competitive advantages depending on the cobamide provided. BtuB1‐only strains outcompete BtuB3‐only strains when adeninylcobamide is the cobamide source, while BtuB3‐only strains dominate when 2‐methyladeninylcobamide is provided [9]. However, the relative abundance of BtuB1 and BtuB3 in these operon‐only backgrounds has not been reported, so it remains unclear whether this outcome reflects intrinsic cobamide selectivity or differences in the expression level between the two strains. If it does reflect intrinsic selectivity, this functional specialization among BtuB paralogs for cobamides differing by a single methyl group on the lower ligand would imply exquisite selectivity at some stage of the transport process. Whether this selectivity originates in the BtuB barrel itself, in BtuG, or in the upstream BtuJ/BtuH/BtuK relay proteins cannot be resolved from the current data. Critically, B. thetaiotaomicron BtuB’s have not yet been demonstrated to bind cobamides from solution, an important gap that prevents assignment of selectivity to the BtuB barrel itself [9]. The identity of the TonB homologs, among the eleven TonB proteins encoded by B. thetaiotaomicron, that energize each BtuB paralog also remains uncharacterized [45].

3.4. Structural Basis for Cobamide Binding to BtuG

BtuGH, BtuG2, and BtuG3, the three BtuG paralogs in B. thetaiotaomicron, adopt a seven‐bladed β‐propeller fold that constitutes the B₁₂‐binding scaffold (Figure 2B) [18, 19, 46]. Crystal structures of all three BtuG homologs have been determined in complex with Cbl or Cbi [19, 46]. In all BtuG structures, B₁₂ binds to one face of the β‐propeller (Figure 4) [19]. In BtuG2•AdoCbl, multiple hydrogen bonds between the corrin ring amide sidechains and the sidechains of Ser‐131, Asn‐107, Glu‐55, Gln‐93, Ser‐195, Gln‐219, Tyr‐196, and Tyr‐239 stabilize AdoCbl binding (Figure 4B). Tyr‐316 and Tyr‐335 form additional hydrogen bonds with the Ado moiety, and Phe‐58 forms π‐stacking interactions with the DMB moiety stabilizing the base‐on conformation [19, 46].

FIGURE 4.

(A) Structure of BtuG2•AdoCbl (PDB code: 8BMZ) [19] confirmed a seven‐bladed β‐propeller fold (green ribbon). AdoCbl (orange sticks) binds to one face of the β‐propeller, where multiple loops form the cofactor‐binding site. (B) A close‐up view of AdoCbl binding shows multiple hydrogen bonds between the amide sidechains of the corrin ring and the sidechains of Ser‐131, Asn‐107, Glu‐55, Gln‐93, Gln‐219, Ser‐195, Tyr‐196, and Tyr‐239, which stabilize AdoCbl binding.

BtuG binds B₁₂ with affinities in the femtomolar to picomolar range (Table 1). Various methods have been used to estimate the affinity for CNCbl binding to BtuG2. Grating‐coupled interferometry (GCI) yielded a K _D = 1.7 pM, surface plasmon resonance (SPR) yielded K _D = 187 fM to 42 pM, and free energy calculations estimate a K _D = 14 fM (Table 1) [18, 19, 20, 46]. Consistent with these exceptionally high affinities, MD simulations reveal that the highly negatively charged surface of BtuG2 efficiently attracts CNCbl over relatively large distances [19]. A critical differentiating feature among the three paralogs is their affinity for Cbi. BtuG2 binds Cbi with ~66‐fold higher affinity than BtuG3 and over 1000‐fold higher affinity than the β‐propeller domain in BtuGH (Table 1), a difference attributable to structural features of the binding pocket rather than differences in cobamide contact residues per se [46]. This paralog‐specific Cbi affinity suggests that BtuG2 may be specialized for acquiring all cobamides, while BtuGH and BtuG3 are more selective for intact cobamides with a nucleotide loop.

TABLE 1.

Summary of kinetics parameters for cobamide binding to BtuG.

Protein	Cobamide	K On , s −1 M −1	K Off , s −1	K D , M	Method
BtuGH	CNCbl	3.4 × 105 ± 0.2	4.8 × 10−3 ± 4.2	1.4 × 10−8 ± 1.2	SFS [20]
BtuGH	CNCbl	6.40 × 105 ± 0.03	2.3 × 10−5 ± 0.2	3.70 × 10−11 ± 0.01	SPR [20]
BtuGH (G domain)	CNCbl	8.3 × 107 ± 1.1	1.1 × 10−4 ± 0.3	1.3 × 10−12 ± 0.4	GCI [46]
BtuGH (G domain)	OH2Cbl	5.1 × 107	1.6 × 10−4	3 × 10−12	GCI [46]
BtuGH (G domain)	Cbi	3.1 × 102 ± 0.0	3.6 × 10−3 ± 1.1	6.9 × 10−5 ± 0.2	GCI [46]
BtuG2	CNCbl	6.2 × 107 ± 0.1	9.3 × 10−1 ± 16	1.5 × 10−8 ± 2.6	SFS [20]
BtuG2	CNCbl	5.20 × 105 ± 0.02	2.2 × 10−5 ± 0.1	4.2 × 10−11 ± 0.3	SPR [20]
BtuG2	CNCbl	—	—	1.4 × 10–14	Free Energy Calculations [19]
BtuG2	CNCbl	1.40 × 109 ± 0.05	2.59 × 10−4 ± 0.96	1.87 × 10−13 ± 0.76	SPR [18]
BtuG2	DicyanoCbi	2.61 × 109 ± 1.56	4.54 × 10−4 ± 1.38	1.93 × 10−13 ± 0.63	SPR [18]
BtuG2	CNCbl	1.4 × 108 ± 0.8	1.5 × 10−4 ± 1.1	1.7 × 10−12 ± 1.4	GCI [46]
BtuG2	OH2Cbl	2.6 × 107	5.5 × 10−5	2.2 × 10−12	GCI [46]
BtuG2	Cbi	1.4 × 104 ± 0.2	7.8 × 10−4 ± 4.2	5.9 × 10−8 ± 2.9	GCI [46]
BtuG3	CNCbl	5.80 × 107 ± 0.08	—	—	SFS [20]
BtuG3	CNCbl	2.00 × 105 ± 0.01	4.0 × 10−5 ± 0.1	2.00 × 10−10 ± 0.05	SPR [20]
BtuG3	CNCbl	3.8 × 107 ± 1.5	8.6 × 10−4 ± 7.7	1.7 × 10−11 ± 1.5	GCI [46]
BtuG3	OH2Cbl	2.8 × 108 ± 2.6	7.4 × 10−4 ± 4.7	2.8 × 10−11 ± 3.9	GCI [46]
BtuG3	Cbi	4.2 × 102 ± 0.6	1.7 × 10−3 ± 0.4	3.9 × 10−6 ± 1.0	GCI [46]

Note: Dashes indicate values not reported.

Abbreviations: GCI, grating‐coupled interferometry; SSFS, stopped‐flow spectroscopy; SRP, surface plasmon resonance.

3.5. BtuG as an Active Cobamide Scavenger

Beyond its role in the BtuB•BtuG complex, BtuG2 functions as an effective extracellular cobamide scavenger whose B₁₂‐binding affinity is sufficient to out compete the binding affinity of host‐secreted B₁₂‐binding proteins. Human Intrinsic Factor (IF) binds CNCbl with K _D = 5 fM and is secreted into the small intestine to facilitate dietary cobamide absorption prior to receptor‐mediated uptake in the terminal ileum [47, 48, 49]. B. thetaiotaomicron strains expressing only operon 2 can grow using IF‐bound CNCbl as the sole cobamide source [18]. Size‐exclusion chromatography analysis demonstrated transfer of CNCbl from IF to BtuG2, establishing that BtuG2 strips cobamide from this host‐secreted binding protein, likely because BtuG2’s femtomolar binding affinity approaches or matches that of IF, making the transfer thermodynamically favorable or driven by the BtuG2 concentration at the cell surface [18]. The ecological implications of this activity extend to the competitive exclusion of other gut organisms. Purified BtuG2 sequesters cobamides produced by Sinorhizobium meliloti in coculture, and this sequestration suppresses the growth of B₁₂‐dependent competitors including E. coli, demonstrating that BtuG2’s high‐affinity scavenging is sufficient to reshape the corrinoid landscape available to surrounding microbes [50]. These data suggest that BtuGs might function as active cobamide scavengers from both host‐secreted proteins and the shared extracellular corrinoid pool. The physiological relevance of this independent scavenging role remains to be elucidated, as it is not yet clear whether BtuG is present in stoichiometric excess over BtuB in the OM. If BtuG is expressed in excess, it would enable a prominent scavenger role for BtuG given its high affinity for cobamides, besides its ability to form a stable complex with BtuB.

4. Extracellular Cobamide Capture by BtuJ

B. thetaiotaomicron encodes two paralogs of the DUF4465 (Pfam PF14717) protein family, BtuJ1 (BT1491) and BtuJ2 (BT1957), at operons 1 and 2, respectively (Figure 2B) [20, 51]. Both BtuJ1 and BtuJ2 carry predicted Sec/SPII signal sequences and Bacteroidota lipoprotein export signals, directing their export to the OM surface [51]. Surface localization of BtuJ1 was directly confirmed by proteinase K accessibility experiments on intact B. thetaiotaomicron cells. BtuJ1 was degraded by exogenous proteinase K over time, while a periplasmic control protein (SusA) was protected, establishing that BtuJ1 is accessible at the extracellular face of the OM [51].

Under B₁₂‐limiting conditions, BtuJ1 is the most abundant Btu protein across all Btu operons in B. thetaiotaomicron [19, 20, 51]. Comparative proteomic analysis of cells and BEVs shows an approximately 30‐fold decrease in BtuJ1 accumulation following B₁₂ addition, consistent with riboswitch‐mediated repression at the same operon [20]. The functional importance of BtuJ was demonstrated in vivo. Deletion of btuJ1 from a simplified genetic background retaining only operon 1 components abolishes growth in B₁₂‐limiting conditions (0.4 nM CNCbl), while growth at replete concentrations (40 nM CNCbl) is unaffected [51]. Complementation with btuJ1 expressed from a heterologous chromosomal site partially restores growth under B₁₂‐limiting conditions, and competition assays between wild‐type and ΔbtuJ1 strains show that the wild‐type strain rapidly outcompetes the deletion mutant within the first 24 h under B₁₂‐limiting conditions [51]. BtuJ1 therefore provides a significant fitness advantage specifically under B₁₂ limitation and is not merely a redundant cobamide import protein.

4.1. Structural Basis for Cobamide Binding to BtuJ

BtuJ adopts an augmented β‐jellyroll fold, consisting of eight β‐strands arranged into two antiparallel four‐stranded β‍‐‍sheets extended by five additional N‐terminal β‐strands and a loop‐associated α‐helix (Figure 5A) [51, 52]. Crystal structures of BtuJ1•CNCbl and BtuJ2•CNCbl have been reported [20, 51]. The B₁₂‐binding site is formed by a loop‐rich distal region of the protein that creates a negatively charged clamp‐like pocket. Multiple hydrogen bonds between the sidechains of Asn‐208, Asn‐291, Tyr‐167, Tyr‐295, Tyr‐73, and Tyr‐75 on BtuJ2 and the amide functional groups of the corrin ring indicate that the primary binding contacts are made with the corrin ring side chains (Figure 5B) [20, 51]. A ConSurf analysis using 250 BtuJ homologs revealed that residues making hydrogen‐bonding contacts with CNCbl are among the most conserved positions across the family, confirming that this binding mode is functionally maintained across DUF4465 homologs [51].

FIGURE 5.

(A) Structure of BtuJ2•CNCbl (PDB code: 9I2L [20]), revealed a β‐jelly roll fold (purple ribbon). CNCbl binds in a loop rich region and is sequestered away from the solvent. (B) A zoomed‐in view of the cobamide‐binding site showing hydrogen bonds between the cobamide and the side chains of Asn‐208, Asn‐291, Tyr‐167, Tyr‐295, Tyr‐73, and Tyr‐75.

Within the BtuJ2 binding pocket, the lower axial ligand projects outward from the cleft, making hydrophobic or π‐stacking interactions with Tyr‐135 and Phe‐210, while the upper axial ligand faces inward toward an enclosed cavity [51]. This orientation explains the structural promiscuity of BtuJ because neither the lower nor upper axial ligand makes significant stabilizing contacts with the protein. BtuJ binds to CNCbl, AdoCbl, and Cbi, establishing that neither axial ligand is required for productive binding [20, 51]. This promiscuous binding mode is functionally well‐suited to the structurally heterogeneous cobamide pool of the human gut [33, 34].

GCI estimated a K _D = 3.6 nM for CNCbl and K _D = 11.4 nM for Cbi binding to BtuJ1 (Table 2) [51]. Stopped‐flow spectroscopy measurements for CNCbl binding to BtuJ1 reported a K _D = 150 pM, while SPR from the same study estimated a K _D = 846 pM (Table 2) [20]. The substantial variation across methods spanning approximately 20‐fold reflects methodological sensitivities. Importantly, BtuJ1 bind cobamides in the low nanomolar to subnanomolar range, sufficient to compete effectively for cobamides in the gut milieu. The key question is how cobamides are transferred out of this high‐affinity site on BtuJ to their client proteins.

TABLE 2.

Summary of kinetics parameters for cobamide binding to BtuJ, BtuH, BtuK, and BtuF.

Protein	Cobamide	K On , s −1 M −1	K Off , s −1	K D , M	Method
BtuJ1	CNCbl	2.10 × 108 ± 0.07	3.2 × 10−2 ± 0.2	1.5 × 10−10 ± 0.1	SFS [20]
BtuJ1	CNCbl	1.3 × 107 ± 0.01	1.10 × 10−2 ± 0.01	8.46 × 10−10 ± 0.10	SPR [20]
BtuJ1	CNCbl	2.33 × 106 ± 0.28	8.26 × 10−3 ± 1.27	3.6 × 10−9	GCI [51]
BtuJ1	Cbi	1.38 × 106 ± 0.14	1.57 × 10−2 ± 0.16	1.14 × 10−8	GCI [51]
BtuJ2	CNCbl	2.20 × 108 ± 0.03	—	—	SFS [20]
BtuH1	CNCbl	9.5 × 106 ± 0.2	9.0 × 10−2 ± 10	9.5 × 10−9 ± 10.5	SFS [20]
BtuH1	CNCbl	5.10 × 105 ± 0.05	5.60 × 10−3 ± 0.03	1.1 × 10−8 ± 0.01	SPR [20]
BtuH2	CNCbl	8.9 × 106 ± 0.2	2.0 × 10−1 ± 7.2	2.3 × 10−8 ± 8.1	SFS [20]
BtuH2	CNCbl	1.2 × 109	8.61 × 10−5	6.41 × 10−12	SPR [53]
BtuK1	CNCbl	1.80 × 107 ± 0.02	2.9 × 10−1 ± 5.0	1.6 × 10−8 ± 2.8	SFS [20]
BtuK1	CNCbl	1.80 × 105 ± 0.01	1.60 × 10−5 ± 0.08	8.9 × 10−11 ± 0.5	SPR [20]
BtuF	CNCbl	8.8 × 107 ± 0.2	3.3 × 10−1 ± 9.8	3.8 × 10−9 ± 11.1	SFS [20]
BtuF2	CNCbl	6.20 × 107 ± 0.05	—	—	SFS [20]
BtuF2	CNCbl	6.60 × 105 ± 0.07	5.7 × 10−3 ± 0.04	8.6 × 10−9 ± 0.1	SPR [20]

Note: Dashes indicate values not reported.

Abbreviations: GCI, grating‐coupled interferometry; SFS, stopped‐flow spectroscopy; SPR, surface plasmon resonance.

4.2. Phylogenetic Distribution of BtuJ

BtuJ homologs (Pfam PF14717) are distributed across 948 bacterial genomes, with Bacteroidota accounting for 84% of hits [51]. At the family level, Bacteroidaceae comprise ~20% of all PF14717‐containing genomes. Gene duplication is common: 63% of genomes carry a single copy, while 26% carry two copies, consistent with the BtuJ1/BtuJ2 duplication observed in B. thetaiotaomicron [51]. PF14717 homologs are also present in non‐Bacteroidota bacteria from gut microbiomes, hydrothermal vents, and soil environments, and characterization of a DUF4465 protein from Coraliomargarita akajimensis demonstrated that the β‐jellyroll B₁₂‐binding function is conserved across this evolutionarily divergent set of homologs [52]. No PF14717 homologs have been detected in archaeal genomes [51], indicating that this is a bacterial‐specific protein family. The broad distribution of BtuJ‐like proteins across gut Bacteroidota, combined with the fitness data from B. thetaiotaomicron, suggests that extracellular cobamide capture by surface‐exposed jellyroll‐fold proteins is a broadly conserved and ecologically significant adaptation for cobamide scavenging in the competitive corrinoid environment of the gut.

5. BtuH: A Structurally Novel Class of Cobamide‐Binding Lipoproteins

B. thetaiotaomicron BtuH was discovered through a vitamin B₁₂ affinity‐based probe (B₁₂‐ABP) screen [53]. The functional importance of BtuH2 is demonstrated by competitive colonization experiments in gnotobiotic mice. Mice colonized with a mixture of wild‐type B. thetaiotaomicron and a ΔbtuH2 deletion strain show rapid competitive exclusion of the ΔbtuH2 strain by the wild type, with the deletion strain reaching undetectable levels within the duration of the experiment [53]. This fitness defect establishes BtuH2 as required for competitive cobamide acquisition in the native gut environment. The fitness defect is diet‐dependent, consistent with BtuH2 playing a role particularly when cobamide is scarce [53].

5.1. BtuH Domain Architecture

Three B. thetaiotaomicron proteins share the BtuH B₁₂‐binding domain: BtuH1 (BT1488) consists of the B₁₂‐binding domain only; BtuH2 (BT1956) carries three N‐terminal PKD/Ig‐like domains followed by the C‐terminal B₁₂‐binding domain (Figure 6A); and BtuGH (BT1490) in which the B₁₂‐binding domain is fused at the N terminus of a seven‐bladed β‐propeller domain characteristic of BtuG proteins [20, 53]. All three BtuH proteins carry lipoprotein signal sequences and are proposed to be surface‐exposed OM lipoproteins, consistent with their roles in extracellular cobamide capture [20, 53].

FIGURE 6.

(A) Domain architecture of BtuH2. (B) Structure of BtuH2•CNCbl (PDB code: 7BIZ [53]) with PKD domains 2 and 3 (dark blue, ribbon) and a globular B₁₂‐binding domain (blue ribbon). (C) The residues Tyr‐510, His‐476, and Gln‐333 in BtuH2 hydrogen bond with the amide side chains of the corrin ring. The upper axial face of CNCbl (orange sticks) is solvent exposed. The PO₄ group on the nucleotide loop is further stabilized by a Na (purple sphere) and Ca (green sphere) metal‐binding site.

5.2. Structural Basis of Cobamide Binding to BtuH

The X‐ray crystal structure of BtuH2•CNCbl confirmed a multidomain architecture in which two of the three N‐terminal PKD/Ig‐like domains that extend the protein away from the OM surface were resolved, followed by the C‐terminal B₁₂‐binding domain (Figure 6B) [53]. The C‐terminal domain adopts a compact globular fold with no recognizable structural homologs, forming a B₁₂‐binding pocket whose contacts are primarily with the corrin ring sidechains and the nucleotide loop (Figure 6B,C) [53]. BtuH2 makes no contacts with the upper axial ligand, leaving the β‐face of the corrin ring largely solvent‐exposed [53]. The upper axial face of this redox‐active cofactor is typically protected or buried away from the solvent to prevent side reactions as seen in most cobamide‐binding proteins, suggesting that BtuH might serve an intermediate step in cobamide import [7]. His‐476, Gln‐333, and Tyr‐510 make hydrogen bonds with the side chains of the corrin ring, and multiple aromatic residues (Tyr‐475, Trp‐577, Phe‐401, and Phe‐334) stabilize the corrin ring [53]. Interestingly, the phosphate group oxygens on the nucleotide loop are coordinated to a two‐metal binding site. Residues (Glu‐406, Glu‐584, Asp‐522, Asn‐399, and Ser‐582) coordinating to Na and Ca in this site are highly conserved in BtuH, which suggests an important role in nucleotide loop binding [53]. The functional consequences of lower axial ligand identity on BtuH2 affinity have not been characterized, specifically, whether BtuH2 binds purinyl or phenolyl cobamides with similar affinities compared to CNCbl. Characterizing this selectivity represents a critical knowledge gap, which will provide insight into the functional role of BtuH in cobamide import.

BtuH1 binds CNCbl with nanomolar affinity (K _D = 9.5 nM by stopped‐flow spectroscopy and K _D = 11 nM by SPR) (Table 2) [20]. BtuH2 exhibits a wider range of measured affinities depending on methodology: stopped‐flow spectroscopy yielded a K _D = 23 nM, while SPR estimated a K _D = 6.41 pM (Table 2) [20, 53]. The large variation in BtuH2 affinity measurements underscores the need for caution in comparing kinetic values across the Btu protein family.

5.3. BtuGH Fusion

BtuGH (BT1490) at operon 1 is a chimeric protein in which the B₁₂‐binding domain of BtuH is fused to a seven‐bladed β‍‐‍propeller domain characteristic of BtuG [19, 20]. Both domains are independently capable of binding CNCbl [20]. The β‐propeller domain of BtuGH is structurally homologous to those of BtuG2 and BtuG3 and is proposed to contribute to the OM‐associated capture function characteristic of BtuG, while the BtuH domain may provide an additional B₁₂‐binding reservoir or relay function [19]. The role of the BtuH‐BtuG domain fusion has, not been investigated, but suggests that BtuH and BtuG might function together for cobamide import.

5.4. Phylogenetic Distribution of BtuH

BtuH homologs were identified in 37 gut‐associated Bacteroidota, frequently coencoded with BtuB homologs in the same operon [53]. BtuGH fusion homologs have been identified in several gut bacteria beyond B. thetaiotaomicron, including B. fragilis [20]. The distribution of BtuH homologs across gut Bacteroidota, combined with the strict fitness requirement for BtuH2 demonstrated in gnotobiotic mice, argues that this structurally novel B₁₂‐binding domain represents a broadly conserved and functionally nonredundant component of the cobamide‐scavenging machinery in gut Bacteroidota.

6. BtuK

B. thetaiotaomicron encodes two tandem lipoproteins, BtuK1 (BT1486) and BtuK2 (BT1487), each bearing an N‐terminal PKD (polycystic kidney disease‐like Ig‐fold) domain and a C‍‐‍terminal seven‐bladed β‐propeller domain (Figure 2B) [20]. BtuK1 binds CNCbl with nanomolar to picomolar affinity as measured by stopped‐flow spectroscopy (K _D = 16 nM) and SPR (K _D = 89 pM) (Table 2), while BtuK2 did not bind cobamides when loaded from solution [20]. Crystal structures of BtuK1•CNCbl (PDB: 9S6C) [54] and apo‐BtuK2 (PDB: 4R5O) [55] are available, and structural alignment of the two proteins provides a possible explanation for their differential B₁₂‐binding ability despite their overall structural similarity (Figure 7A,B). The structural basis for this binding lies in a loop (residues 22–32) of BtuK1, where Glu‐23 and Asn‐62 form hydrogen bonds with the corrin ring side chains and Trp‐25 and His‐28 hydrogen bond with the nucleotide loop of CNCbl (Figure 7B). In BtuK2, the analogous loop is truncated to four residues (129–132), and the equivalent side chains are not within hydrogen‐bonding distance of the cobamide, which could prevent cofactor binding. Additional hydrogen bonds between Lys‐76, Thr‐60, Asn‐102, and Glu‐77 and the sidechains of the corrin ring stabilize cobamide binding. Additionally, loop 151–163 on BtuK2 occupies the B₁₂‐binding site in BtuK1. Notably, the structure of BtuK1•CNCbl reveals that the cofactor is primarily solvent‐exposed in the bound state, consistent with a role in cobamide hand‐off rather than in sequestration. The identity of BtuK1 client proteins remains unknown, and the inability of BtuK2 to bind cobalamin makes it difficult to assign a role in cobamide import by analogy, leaving the function of both proteins within the operon 1 machinery unresolved.

FIGURE 7.

(A) An overlay of BtuK1•CNCbl (yellow ribbon) (PDB code: 9S6C) [54] and BtuK2 (gray ribbon) (PDB code: 4R5O) [55] β‐propeller domains. (B) A close‐up view of the B₁₂‐binding in BtuK1 showing hydrogen bonds between the CNCbl and His‐28, Trp‐25, Glu‐77, Asn‐102, Thr‐60, Lys‐76, Asn‐62, and Glu‐23. Loop 129–132 in BtuK2 is shorter than the analogous B₁₂‐binding loop 22–32 in BtuK1. Loop 151–163 on BtuK2 occupies the B₁₂‐binding site in BtuK1.

7. Bacterial Extracellular Vesicles as Extracellular Corrinoid Reservoirs

Like other Gram‐negative bacteria, B. thetaiotaomicron releases BEVs into the extracellular milieu, and several Btu proteins are packaged into these BEVs [20, 56, 57]. Comparative proteomics of cells and BEVs under cobamide starvation revealed selective enrichment of BtuJ1, BtuJ2, BtuK1, BtuK2, BtuH1, and BtuL on BEVs relative to the OM, with BtuK1 and BtuK2 being selectivey enriched at distinct time points during early growth phases [20]. Despite this broad Btu protein enrichment on BEVs, B₁₂‐binding on the BEV surface is attributable to BtuJ1 and BtuJ2. Expression of BtuK1 or BtuK2 in isolation in a strain lacking all B₁₂ uptake operons did not restore BEV‐associated B₁₂‐binding activity, confirming that neither BtuK contribute to cobamide capture on BEVs [20]. This positions BEVs as extracellular cobamide supplies whose B₁₂‐capturing function is mediated specifically by BtuJ, extending BtuJ‐dependent capture beyond the immediate cell surface. The presence of BtuK1 and BtuK2 on BEVs without a demonstrable B₁₂‐binding role raises the unresolved question of whether these proteins serve a structural, regulatory, or as‐yet‐unidentified function in the BEV context.

7.1. BtuL and the Regulation of BEV Release

BtuL is an operon 2‐encoded lipoprotein that does not bind cobamide yet is one of the most selectively enriched proteins on B. thetaiotaomicron BEVs under cobamide starvation [20]. Functional assays established that BtuL promotes early‐phase BEV release: strains expressing BtuL in isolation show elevated BEV production during early growth, while BtuL deletion reduces but does not abolish BEV formation [20]. BtuL is therefore a modulator of the BEV release efficiency rather than a structural requirement for BEV biogenesis. The mechanism by which a noncobamide‐binding lipoprotein encoded within a cobamide transport operon influences BEV formation is entirely uncharacterized. Whether BtuL acts through direct interactions with the OM, effects on OM protein composition, or indirect regulation of operon 2 expression during BEV formation cannot be resolved from the current data. BtuL is coregulated with cobamide transport genes under riboswitch control, raising the possibility that BtuL‐mediated BEV release is itself part of the cobamide acquisition response: specifically, that BtuL increases BEV release under cobamide limitation to expand the extracellular BtuJ‐decorated corrinoid reservoir precisely when cobamide is most scarce. This hypothesis is consistent with the cobamide‐starvation‐dependent enrichment of BtuL in BEVs but has not been directly tested [20]. BtuL is predicted to be a cell wall biogenesis lipoprotein bearing an N‐terminal lipobox followed by a large, repetitive extracellular domain assigned to the receptor L‐domain superfamily (InterPro: IPR036941), based on structural modeling in the AlphaFold Protein Structure Database (Figure 8).

FIGURE 8.

AlphaFold predicted structure of BtuL bearing an N‍‐‍terminal lipobox followed by a large, repetitive extracellular domain assigned to the Receptor L‐domain superfamily (gray cartoon).

7.2. B. thetaiotaomicron BEVs Bind Structurally Diverse Cobamides

B. thetaiotaomicron BEVs incubated with cobamides bearing five structurally distinct lower axial ligands (5‐methylbenzimidazole, 5‐hydroxybenzimidazole, 5‐methoxybenzimidazole, adenine, and 2‐methyladenine) supported B. thetaiotaomicron growth, and each was more effective than Cbl at rescuing growth when delivered via BEVs [58]. This structural breadth is consistent with the corrin‐ring‐centered binding mode of BtuJ1 and BtuJ2 and with the broader observation that B. thetaiotaomicron can grow on a structurally diverse range of cobamides [22, 58]. Deletion of all Btu operons abolished the ability of B. thetaiotaomicron to utilize BEV‐delivered cobamides, confirming that import of BEV‐bound cobamides still requires the cellular Btu uptake machinery and that BEVs function as an upstream delivery vehicle rather than a bypass of the core transport system [20].

Beyond microbial communities, B. thetaiotaomicron BEVs interact directly with host cells [59]. Caco‐2 intestinal epithelial cells internalize corrinoid‐loaded BEVs and traffic them to lysosomes, suggesting a potential route by which host cells acquire cobamides of microbial origin [58, 60]. B. thetaiotaomicron BEVs can also cross the intestinal epithelium via paracellular passage and have been detected in distal tissues including the liver; whether B₁₂‐loaded BEVs specifically contribute to the host cobamide status at distal sites remains to be demonstrated directly, but the biodistribution data are consistent with this possibility [60]. An important caveat is that host intestinal cobamide absorption is itself highly selective for cobalamin: IF and the downstream cubilin receptor complex discriminate Cbl from cobamide analogs with substantially reduced affinity for purinylcobamides and Cbi [47, 49]. BEV‐delivered cobamides that bypass the canonical IF‐cubilin pathway via lysosomal trafficking or paracellular passage may therefore access host cells with different selectivity constraints than dietary cobalamin absorption, raising the possibility that the structural diversity of BEV‐bound cobamides in the gut could influence which cobamides reach host tissues via this route. Together, the BEV‐associated cobamide reservoir of B. thetaiotaomicron represents an extracellular extension of the Btu relay system. Its ecological and physiological reach spans microbial competitors, Bacteroidota community members, and host epithelial cells, substantially exceeding what the cell‐surface transport machinery could achieve alone.

8. Cobamide Trafficking Across the Inner Membrane

Cobamide import across the IM in E. coli is accomplished by a single BtuFCD complex operating through a well‐defined allosteric mechanism [15, 35, 61, 62, 63, 64, 65]. BtuF, a soluble periplasmic binding protein, captures cobamide in the base‐on conformation and delivers it to BtuC₂D₂ by docking onto the periplasmic loops of BtuC [15, 66]. BtuC₂D₂ lacks a cobamide‐binding site; substrate identity is therefore determined exclusively by BtuF [15]. ATP binding and hydrolysis by BtuD drive conformational transitions between outward‐ and inward‐facing states of BtuC₂D₂, and cobamide release from BtuF into the translocation channel is driven allosterically by these transitions rather than by spontaneous dissociation [63, 64, 65]. BtuF also accommodates Cbi, which lacks the nucleotide loop, through repositioning of a conserved tryptophan residue (Trp‐66), demonstrating that the periplasmic binding step is not strictly dependent on a complete lower ligand [67].

B. thetaiotaomicron retains this core mechanism but departs from the E. coli template. It encodes two IM transporter complexes, BtuFCD at operon 2 and BtuF2C2D2 at operon 4 (Figure 2B), and at least one is necessary for B₁₂ uptake, with cobamide import through these systems constituting the only known route for cytoplasmic cobamide delivery [9, 17]. Neither complex is dedicated to a particular OM transporter, and corrinoids imported by any of the three BtuB paralogs can be delivered to the cytoplasm via either BtuFCD or BtuF2C2D2, indicating that OM paralog identity does not determine IM routing [9]. Most strikingly, both B. thetaiotaomicron periplasmic binding proteins (BtuF and BtuF2) carry lipoprotein signal sequences that anchor them to the IM, in direct contrast to the freely soluble E. coli BtuF [20, 68]. Since the substrate‐identity gating function in E. coli depends on BtuF docking onto BtuC through mobile periplasmic contacts, the functional consequences of membrane anchoring for cobamide capture, BtuC docking geometry, and the allosteric release mechanism in B. thetaiotaomicron are unknown. BtuF and BtuF2 share only 21.4% and 20.4% sequence identity with E. coli BtuF, respectively, though both bind CNCbl with nanomolar affinity (Table 2), comparable to the affinity of the E. coli BtuF for CNCbl (15 nM measured by ITC) [20, 68].

Pairwise alignment shows that the BtuC and BtuD subunits are conserved. The E. coli BtuC shares 35.6% and 34.1% identity with B. thetaiotaomicron BtuC and BtuC2, respectively, and the E. coli BtuD shares 26.1% and 27.4% identity with BtuD and BtuD2. Walker A and B motifs in BtuD and BtuD2 are intact, and the transmembrane topology of BtuC and BtuC2 are predicted to match the E. coli template, supporting conservation of the overall alternating‐access mechanism [9]. However, B. thetaiotaomicron BtuF and BtuCD have not been characterized structurally and biochemically, and whether BtuFCD and BtuF2C2D2 differ in cobamide selectivity remains unknown.

9. Riboswitch‐Mediated Regulation of Btu Gene Expression

Cobamide riboswitches repress translation upon B₁₂ binding [31], coupling Btu expression in B. thetaiotaomicron to intracellular cobamide availability across operons 1, 2, and 3, while operon 4 lacks an easily identifiable canonical riboswitch sequence (Figure 2B) [9]. However, a recent study provides evidence consistent with riboswitch‐mediated regulation of operon 4 despite the absence of a sequence that can readily be identified bioinformatically [20]. Btu expression is thus inversely correlated with the intracellular cobamide concentration through riboswitch‐mediated repression. The response saturates at physiologically relevant cobamide concentrations, giving a switch‐like rather than an indefinitely graded regulatory output [9, 18, 32]. B₁₂‐dependent riboswitches in B. thetaiotaomicron display cobamide selectivity. CNCbl and adeninylcobamide each fully repress riboswitch‐controlled Btu protein expression, while 5‐methylbenzimidazole yields only partial repression [20]. This intermediate repression indicates that riboswitches might discriminate among cobamides and that the identity of the lower axial ligand may have direct physiological consequences for Btu protein expression and, therefore, cobamide acquisition. The structural basis for this discrimination is not yet determined. This would mean that B. thetaiotaomicron can sense not just the quantity but the precise chemical identity of its intracellular cobamide pool, adjusting the transport capacity in response to the availability of preferred cofactor forms.

The architecture of operon 1 riboswitch regulation adds a further layer of sophistication (Figure 2B). Two B₁₂‐dependent riboswitches are encoded within the operon 1 transcript: the first lies upstream of btuK1, btuK2, and btuH1, controlling the expression of this gene cluster; the second is positioned immediately downstream of btuH1 and represses the translation of btuB1, btuGH, and btuJ1 upon B₁₂ binding [9, 20]. Each riboswitch is paired with its own promoter; the btuK1 promoter is approximately an order of magnitude weaker than the btuB1 promoter [20], consistent with at least partially independent transcriptional control of the two gene clusters rather than a single cotranscribed unit. No intervening transcriptional terminator has been identified between the two clusters, so read‐through transcription from the btuK1 promoter into the btuB1 region remains possible, a caveat tempered by the general difficulty of predicting terminators in Bacteroides. This dual‐promoter architecture, whether transcriptionally independent or subject to partial read‐through, provides a mechanism to tune the relative abundance of upstream capture proteins (BtuK, BtuH1) against the core OM transporter components (BtuB1, BtuGH, BtuJ1) according to intracellular cobamide status. Whether the two riboswitches are differentially saturated at physiologically relevant cobamide concentrations, and what consequence this would have for the stoichiometry of operon 1 relay proteins, remains unexplored.

10. Summary and Outlook

10.1. Model for Cobamide Import in Gut Bacteria

The B. thetaiotaomicron Btu system represents a multistage relay whose architecture is shaped by the competitive environment of the gut. The literature reviewed here allows us to hypothesize a model for cobamide import in B. thetaiotaomicron (Figure 9). Cobamide capture may initiate in the extracellular space by BtuJ. BtuJ proteins are constitutively expressed at high levels and are the dominant B₁₂‐binding proteins on BEVs, extending their capture function into the extracellular milieu beyond the cell surface [20, 51]. Cobamides are then transferred from BtuJ to BtuK or BtuH or BtuG (Figure 9). BtuH and BtuK are proposed to function as intermediate relay components between BtuJ and BtuG, though whether BtuH and BtuK transfer cobamides to the BtuB•BtuG complex remains unresolved. The existence of the BtuGH fusion across Bacteroides suggests that BtuH might facilitate the efficiency of cobamide transfer to the BtuB•BtuG complex, though this has not been reported. Cobamide is captured by BtuG in the open conformation of the BtuB•BtuG complex. Lid closure sterically displaces cobamide from the BtuG binding site via the BtuB EL8 loop, transferring it to BtuB for TonB‐dependent, proton motive force‐driven translocation into the periplasm [19]. In the periplasm, cobamide is captured by membrane‐anchored BtuF and delivered to BtuC₂D₂ for ATP hydrolysis‐driven cytoplasmic import.

FIGURE 9.

Model for cobamide import in B. thetaiotaomicron. Cobamide (red rectangle) capture may initiate in the extracellular space by BtuJ (a, d). BtuJ proteins (red circle) are the dominant B₁₂‐binding proteins on BEVs, extending their capture function into the extracellular milieu. Cobamides are then transferred from BtuJ to BtuK (a) or BtuH (b) or BtuG (c). BtuH and BtuK are proposed to function as intermediate relay components between BtuJ and BtuG. Cobamide is captured by BtuG in the open conformation of the BtuB•BtuG complex. Lid closure sterically displaces cobamide from the BtuG binding site via the BtuB EL8 loop, transferring it to BtuB for TonB‐dependent, proton motive force‐driven translocation into the periplasm. In the periplasm, cobamide is captured by membrane‐anchored BtuF and delivered to BtuC₂D₂ for ATP‐driven cytoplasmic import.

A key unresolved question is how cobamides are transferred from one high‐affinity binding site to downstream client proteins. One explanation is that transfer of the cofactor is thermodynamically driven in which the affinity gradient across the extracellular relay from picomolar affinity at BtuJ to femtomolar affinity at BtuG provides the driving force for unidirectional cobamide transfer (Figure 9, Tables 1 and 2). Whether transfer requires conformational changes, direct protein–protein interactions with the acceptor, or proceeds through simple kinetic off‐rate dissociation remains uncharacterized. In vitro assays demonstrated that BtuJ1 loads cobalamin onto both BtuG2 and BtuH2 in solution, providing direct evidence for BtuJ1 as an upstream donor in the relay [51]. However, under physiological conditions, BtuG and BtuB form a stable complex in which the BtuG B₁₂‐binding site is sterically occluded by the BtuB EL8 loop [19]. Whether BtuJ can transfer cobamide to the BtuB•BtuG complex has not been tested. Evaluating transfer to the physiological BtuB•BtuG target rather than isolated BtuG is therefore critical and may also clarify the roles of BtuH and BtuK, which are coencoded with BtuJ at operon1 and may function precisely to bridge cobamide delivery to a sterically restricted BtuB•BtuG complex that BtuJ alone cannot efficiently load.

10.2. B. thetaiotaomicron Btu System as a Model for Cobamide Import in Gut Bacteria

The widespread distribution of BtuB, BtuG, BtuH, BtuK, and BtuJ homologs across gut Bacteroidota [9, 20, 51, 53], each contributing a distinct functional layer to extracellular cobamide capture, suggests that multistage, lipoprotein‐assisted relay systems represent a broadly conserved strategy for cobamide scavenging in the gut. The distribution of individual relay components beyond Bacteroidota raises further questions. BtuH homologs occur predominantly in Bacteroidota, but three Firmicutes (or Bacillota) species that are human gut commensals also carry predicted BtuH homologs, at least one of which (Eubacterium siraeum EUBSIR_01285) is encoded in proximity to a cobamide transporter [53]. BtuJ (DUF4465) homologs are distributed even more broadly, spanning eight bacterial clades including Bacillota, Pseudomonadota, Cyanobacteriota, and organisms from hydrothermal vent and soil environments, with B₁₂ binding confirmed by structural and biochemical characterization across seven sequence‐diverse family members sharing as little as 16% pairwise identity [51, 52]. The presence of BtuH and BtuJ homologs in Gram‐positive Firmicutes, which lack an OM, raises a mechanistic question: in the absence of the OM lipoprotein‐sorting machinery that anchors BtuJ and BtuH at the cell surface in B. thetaiotaomicron, these proteins would presumably function as secreted or cell wall‐associated cobamide binders. Whether they serve analogous extracellular capture roles in a different cellular context, or have been recruited for distinct functions, remains entirely uncharacterized. The possibility that a structurally conserved B₁₂‐binding fold participates in cobamide scavenging across the Gram‐positive/Gram‐negative divide would have significant implications for understanding how cobamide competition and sharing operate at the community level in the gut, where Bacteroidota and Firmicutes codominate and compete for the same structurally heterogeneous corrinoid pool. This relay does not terminate at the cell surface: as described above, B. thetaiotaomicron extends its cobamide‐capturing reach further still by enriching BtuJ proteins onto BEVs that function as mobile, extracellular cobamide reservoirs.

10.3. Key Gaps

Several mechanistic questions remain unanswered.

• •The affinity gradient model for relay directionality, while parsimonious, is inferential: transfer rates between BtuJ, BtuH, BtuG, and BtuB have not been measured, the affinity data span multiple methodologies with substantial variation, and the conformational coupling between BtuG lid closure and BtuB channel opening has not been validated.
• •The functional significance of multiple paralogs of Btu proteins is unknown.
• •IM cobamide import is understood entirely by homology with E. coli.
• •The functional roles of BtuH, BtuK1, and BtuK2 are unknown.
• •At the system level, which of the eleven B. thetaiotaomicron TonB homologs energize each BtuB paralog has not been established.
• •The molecular basis of cobamide selectivity in B. thetaiotaomicron has not been determined.

Addressing these gaps will require a combination of structural and biochemical approaches with in vivo experiments in physiologically relevant conditions.

Author Contributions

Alexander Drake: formal analysis, writing – original draft, writing – review and editing. Romila Mascarenhas: conceptualization, writing – original draft, writing – review and editing.

Funding

This study was supported by the National Institutes of Health (GM143482).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

All data are available in the manuscript.