Xiaofang Jiang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Xiaofang Jiang has a total of 225 co-authors. Here are the top co-authors by number of shared publications:
- Brantley Hall
Department of Cell Biology and Molecular Genetics, University of Maryland, College Park, College Park, MD, USA
14 shared publications Keith Dufault-Thompson
National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, 8600 Rockville Pike, Bethesda, MD 20894, USA
13 shared publications- Keith Dufault-Thompson
National Library of Medicine, National Institutes of Health, Bethesda, MD, USA
9 shared publications Yiyan Yang
National Library of Medicine, National Institutes of Health
9 shared publications- Angela K. Jiang
National Library of Medicine, National Institutes of Health, Bethesda, MD, USA
7 shared publications - Gabriela Arp
Department of Cell Biology and Molecular Genetics, University of Maryland, College Park, College Park, MD, USA
6 shared publications
Latest Publications
- Open Access
- 11 November 2026
- Published by American Society for Microbiology in mBio
- Vol. 17 (11) , e0218026
The core interaction network of transcription elongation factors is broadly maintained across diverse bacteria
Bacterial transcription elongation has been investigated most thoroughly for the model bacterium Escherichia coli. The E. coli elongation factor NusG associates through its N-terminal domain with the transcription elongation complex (TEC), while its C-terminal domain forms mutually exclusive interactions with either ribosomal protein S10 or Rho termination factor. The E. coli NusG:S10 complex bridges the TEC with a pioneering ribosome for coupling of transcription and translation. NusG also participates in antitermination complexes (ATC) that inhibit Rho-dependent termination and increase transcription of rRNA operons. A characteristic sequence in the nascent rrn RNA, called boxA , loads E. coli NusB:S10 onto the NusG-bound TEC, leading to the formation of the rrn ATC. However, these findings may not extend to all bacteria. Rho-dependent termination is less important in gram-positive bacteria, including some that do not encode Rho, and transcription and translation remain uncoupled in Bacillus subtilis and many other gram-positive bacteria. It is therefore unclear whether the interaction network for E. coli elongation factors is conserved in these organisms. We identified boxA -like sequences upstream of 16S and 23S genes in Bacillus species and found that they associate directly with NusB. We also found that B. subtilis S10 associates with both NusB and NusG, and that Rho can also associate with NusG. Bioinformatic analyses revealed conservation of interaction interface residues across multiple phyla. Together, these data demonstrate that the core interaction network of bacterial elongation factors is broadly conserved across diverse bacteria, despite differences in how these interactions function in individual species. IMPORTANCE: Escherichia coli is known to use several key elongation factors that promote direct coupling of transcription with translation. These factors also trigger the formation of an antitermination complex that protects the synthesis of ribosomal RNA and CRISPR transcripts from interference by the Rho termination factor. However, Bacillus subtilis is not thought to couple transcription with translation. It is also not known whether B. subtilis needs to suppress Rho, given that its Rho is dispensable. Therefore, we purified B. subtilis elongation factors and mapped their interaction network. Surprisingly, we find that the B. subtilis elongation factors form complexes that closely resemble E. coli counterparts, despite the significant mechanistic differences in how these bacteria control transcription elongation.
- 11 November 2026
- Published by American Society for Microbiology in mBio
- Vol. 17 (11) , e0218026
Open AccessThe core interaction network of transcription elongation factors is broadly maintained across diverse bacteria
Bacterial transcription elongation has been investigated most thoroughly for the model bacterium Escherichia coli. The E. coli elongation factor NusG associates through its N-terminal domain with the transcription elongation complex (TEC), while its C-terminal domain forms mutually exclusive interactions with either ribosomal protein S10 or Rho termination factor. The E. coli NusG:S10 complex bridges the TEC with a pioneering ribosome for coupling of transcription and translation. NusG also participates in antitermination complexes (ATC) that inhibit Rho-dependent termination and increase transcription of rRNA operons. A characteristic sequence in the nascent rrn RNA, called boxA , loads E. coli NusB:S10 onto the NusG-bound TEC, leading to the formation of the rrn ATC. However, these findings may not extend to all bacteria. Rho-dependent termination is less important in gram-positive bacteria, including some that do not encode Rho, and transcription and translation remain uncoupled in Bacillus subtilis and many other gram-positive bacteria. It is therefore unclear whether the interaction network for E. coli elongation factors is conserved in these organisms. We identified boxA -like sequences upstream of 16S and 23S genes in Bacillus species and found that they associate directly with NusB. We also found that B. subtilis S10 associates with both NusB and NusG, and that Rho can also associate with NusG. Bioinformatic analyses revealed conservation of interaction interface residues across multiple phyla. Together, these data demonstrate that the core interaction network of bacterial elongation factors is broadly conserved across diverse bacteria, despite differences in how these interactions function in individual species. IMPORTANCE: Escherichia coli is known to use several key elongation factors that promote direct coupling of transcription with translation. These factors also trigger the formation of an antitermination complex that protects the synthesis of ribosomal RNA and CRISPR transcripts from interference by the Rho termination factor. However, Bacillus subtilis is not thought to couple transcription with translation. It is also not known whether B. subtilis needs to suppress Rho, given that its Rho is dispensable. Therefore, we purified B. subtilis elongation factors and mapped their interaction network. Surprisingly, we find that the B. subtilis elongation factors form complexes that closely resemble E. coli counterparts, despite the significant mechanistic differences in how these bacteria control transcription elongation.
- Open Access
- 4 November 2026
- Published by American Society for Microbiology in Journal of Bacteriology
Identification of Vibrio cholerae proteins that bind 3′,3′-cyclic guanosine adenine dinucleotide
Cyclic dinucleotides are used as signaling molecules in both antiphage defense and normal cellular functions. Horizontal gene transfer can introduce antiphage defense systems that encode specific signaling molecules into a new host. Whether these new signaling molecules interact with existing host signaling proteins and whether such interactions have evolutionary implications remain unknown. Vibrio cholerae represents an organism in which a new cyclic dinucleotide signaling molecule, 3′,3′-cyclic guanosine adenine dinucleotide (3′,3′-cGAMP) produced by DncV, is introduced by horizontal gene transfer of an antiphage system. 3′,3′-cGAMP activates CapV phospholipase activity, leading to reduced growth and swim motility. To identify additional 3′,3′-cGAMP-binding proteins, we performed a 3′,3′-cGAMP-binding screen of the V. cholerae El Tor ORFeome. In addition to the expected 3′,3′-cGAMP-binding proteins CapV and three V-cGAP proteins, this screen identified seven additional 3′,3′-cGAMP-binding proteins that were previously found to bind c-di-GMP. These seven all bind to c-di-GMP with greater affinity, suggesting that these proteins do not bind 3′,3′-cGAMP specifically. We assessed the function of 3′,3′-cGAMP-binding proteins by testing the expression of DncV, CapV, or both in V. cholerae . Expression of DncV alone did not affect growth or swimming motility, in contrast to co-expression with CapV. In Escherichia coli, which lacks native 3′,3′-cGAMP-signaling pathways, co-expression of CapV and DncV also led to reduced growth and swimming motility. In contrast, Pseudomonas aeruginosa, which also lacks endogenous 3′,3′-cGAMP, is less susceptible to CapV and DncV co-expression. Signaling nucleotides from antiphage systems primarily function in phage defense but have the potential to regulate other processes when the phage pressure subsides. IMPORTANCE: In this study, we systematically identified 3′,3′-cGAMP-binding proteins in V. cholerae , which also uses the similar cyclic dinucleotide c-di-GMP for signaling. In addition to previously observed 3′,3′-cGAMP-binding proteins, including 3′,3′-cGAMP-activated phospholipase CapV and V-cGAP proteins, this screen identified seven new 3′,3′-cGAMP-binding proteins that bound c-di-GMP as well. We found that both 3′,3′-cGAMP production and CapV phospholipase activity were necessary and sufficient to inhibit growth and swimming motility as previously observed. In heterologous bacteria that use c-di-GMP but lack 3′,3′-cGAMP-signaling pathways, 3′,3′-cGAMP activation of CapV phospholipase was similarly required to inhibit growth and swimming motility as observed in V. cholerae . These results suggest that the acquisition of signaling nucleotides from antiphage systems may evolve to regulate cellular processes.
- 4 November 2026
- Published by American Society for Microbiology in Journal of Bacteriology
Open AccessIdentification of Vibrio cholerae proteins that bind 3′,3′-cyclic guanosine adenine dinucleotide
Cyclic dinucleotides are used as signaling molecules in both antiphage defense and normal cellular functions. Horizontal gene transfer can introduce antiphage defense systems that encode specific signaling molecules into a new host. Whether these new signaling molecules interact with existing host signaling proteins and whether such interactions have evolutionary implications remain unknown. Vibrio cholerae represents an organism in which a new cyclic dinucleotide signaling molecule, 3′,3′-cyclic guanosine adenine dinucleotide (3′,3′-cGAMP) produced by DncV, is introduced by horizontal gene transfer of an antiphage system. 3′,3′-cGAMP activates CapV phospholipase activity, leading to reduced growth and swim motility. To identify additional 3′,3′-cGAMP-binding proteins, we performed a 3′,3′-cGAMP-binding screen of the V. cholerae El Tor ORFeome. In addition to the expected 3′,3′-cGAMP-binding proteins CapV and three V-cGAP proteins, this screen identified seven additional 3′,3′-cGAMP-binding proteins that were previously found to bind c-di-GMP. These seven all bind to c-di-GMP with greater affinity, suggesting that these proteins do not bind 3′,3′-cGAMP specifically. We assessed the function of 3′,3′-cGAMP-binding proteins by testing the expression of DncV, CapV, or both in V. cholerae . Expression of DncV alone did not affect growth or swimming motility, in contrast to co-expression with CapV. In Escherichia coli, which lacks native 3′,3′-cGAMP-signaling pathways, co-expression of CapV and DncV also led to reduced growth and swimming motility. In contrast, Pseudomonas aeruginosa, which also lacks endogenous 3′,3′-cGAMP, is less susceptible to CapV and DncV co-expression. Signaling nucleotides from antiphage systems primarily function in phage defense but have the potential to regulate other processes when the phage pressure subsides. IMPORTANCE: In this study, we systematically identified 3′,3′-cGAMP-binding proteins in V. cholerae , which also uses the similar cyclic dinucleotide c-di-GMP for signaling. In addition to previously observed 3′,3′-cGAMP-binding proteins, including 3′,3′-cGAMP-activated phospholipase CapV and V-cGAP proteins, this screen identified seven new 3′,3′-cGAMP-binding proteins that bound c-di-GMP as well. We found that both 3′,3′-cGAMP production and CapV phospholipase activity were necessary and sufficient to inhibit growth and swimming motility as previously observed. In heterologous bacteria that use c-di-GMP but lack 3′,3′-cGAMP-signaling pathways, 3′,3′-cGAMP activation of CapV phospholipase was similarly required to inhibit growth and swimming motility as observed in V. cholerae . These results suggest that the acquisition of signaling nucleotides from antiphage systems may evolve to regulate cellular processes.
- Open Access
SpiE is a distinct gut microbial 4-cholesten-3-one 5β-reductase that acts with SpiR to convert cholesterol to coprostanol
Gut microbes convert cholesterol into coprostanol, a poorly absorbed sterol that is largely excreted in feces and is linked to lower serum cholesterol levels. However, the enzymes responsible for this process remain poorly characterized. The second step of the pathway, cholestenone reduction into coprostanone, has long been established, yet the enzyme(s) responsible were unknown. Here, we use comparative genomics to identify SpiE, a cholestenone (4-cholesten-3-one) reductase that is part of the steroid processing ( spi ) pathway with SpiR. We confirm that SpiE binds to and reduces both cholestenone and progesterone into 5β-reduced forms. SpiR and SpiE together convert cholesterol into coprostanol, indicating that both enzymes are sufficient for coprostanol formation. SpiE belongs to the short-chain dehydrogenase (SDR) family and represents an evolutionarily distinct lineage of SDR ene-reductases. Metagenomic analyses reveal that spiE relative abundance is strongly correlated with stool coprostanol (Spearman’s ρ=0.81, p =5.99e-65). In longitudinal patient cohorts, antibiotic treatment coincided with loss of SpiE and coprostanol production. Post antibiotics, the emergence of distinct SpiE-encoding genera rescued coprostanol production in some individuals. Our findings provide the biochemical basis of the cholesterol to coprostanol pathway and lay the groundwork for future studies linking it to host physiology.
Open AccessSpiE is a distinct gut microbial 4-cholesten-3-one 5β-reductase that acts with SpiR to convert cholesterol to coprostanol
Gut microbes convert cholesterol into coprostanol, a poorly absorbed sterol that is largely excreted in feces and is linked to lower serum cholesterol levels. However, the enzymes responsible for this process remain poorly characterized. The second step of the pathway, cholestenone reduction into coprostanone, has long been established, yet the enzyme(s) responsible were unknown. Here, we use comparative genomics to identify SpiE, a cholestenone (4-cholesten-3-one) reductase that is part of the steroid processing ( spi ) pathway with SpiR. We confirm that SpiE binds to and reduces both cholestenone and progesterone into 5β-reduced forms. SpiR and SpiE together convert cholesterol into coprostanol, indicating that both enzymes are sufficient for coprostanol formation. SpiE belongs to the short-chain dehydrogenase (SDR) family and represents an evolutionarily distinct lineage of SDR ene-reductases. Metagenomic analyses reveal that spiE relative abundance is strongly correlated with stool coprostanol (Spearman’s ρ=0.81, p =5.99e-65). In longitudinal patient cohorts, antibiotic treatment coincided with loss of SpiE and coprostanol production. Post antibiotics, the emergence of distinct SpiE-encoding genera rescued coprostanol production in some individuals. Our findings provide the biochemical basis of the cholesterol to coprostanol pathway and lay the groundwork for future studies linking it to host physiology.
- 14 August 2026
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 12 (33)
Ancestral proteins trace the emergence of substrate specificity and oligomerization within bacterial DEDDy dinucleases
Nucleases are crucial for various bacterial processes, including genome maintenance and host defense. Deoxydinucleases (diDNases), a class of Gram-positive bacteria–specific nucleases associated with mobile genetic elements, are homologous to nanoRNase C (NrnC) in Gram-negative bacteria but exhibit notable differences: diDNases form dimers and cleave DNA dinucleotides, whereas NrnC forms octamers that process both RNA and DNA dinucleotides. The mechanism by which substrate specificity emerged, and whether it is linked to oligomerization, remained unknown. Here, we reconstructed a common ancestor of diDNases and NrnC orthologs that forms a dimer with intermediate preference for DNA. Structures of ancestral and extant dinucleases reveal gradual changes in conformation that gave rise to substrate preference, oligomeric state, and catalytic efficiency. These findings highlight how subtle, concerted structural modifications enable large-scale changes in molecular assembly and functional specialization, harnessing a conserved protein fold. DNA dinucleotide preference in the early ancestor and preservation of DNase activity in all extant enzymes strongly argue for a biological function of DNA dinucleotides.
- 14 August 2026
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 12 (33)
Ancestral proteins trace the emergence of substrate specificity and oligomerization within bacterial DEDDy dinucleases
Nucleases are crucial for various bacterial processes, including genome maintenance and host defense. Deoxydinucleases (diDNases), a class of Gram-positive bacteria–specific nucleases associated with mobile genetic elements, are homologous to nanoRNase C (NrnC) in Gram-negative bacteria but exhibit notable differences: diDNases form dimers and cleave DNA dinucleotides, whereas NrnC forms octamers that process both RNA and DNA dinucleotides. The mechanism by which substrate specificity emerged, and whether it is linked to oligomerization, remained unknown. Here, we reconstructed a common ancestor of diDNases and NrnC orthologs that forms a dimer with intermediate preference for DNA. Structures of ancestral and extant dinucleases reveal gradual changes in conformation that gave rise to substrate preference, oligomeric state, and catalytic efficiency. These findings highlight how subtle, concerted structural modifications enable large-scale changes in molecular assembly and functional specialization, harnessing a conserved protein fold. DNA dinucleotide preference in the early ancestor and preservation of DNase activity in all extant enzymes strongly argue for a biological function of DNA dinucleotides.
- Open Access
Structure of a key adhesin-exopolysaccharide interaction provides insights into matrix assembly in Vibrio cholerae biofilms
Biofilms serve as a protective mechanism for bacteria, including many pathogens. To form such communities, bacteria secrete macromolecules that form an extracellular matrix serving as a barrier against environmental threats, such as predation, antibiotics, and the host immune system. To be effective, a biofilm must anchor to foreign surfaces and retain sufficient stiffness, in the environment or in a host. However, how this matrix self-organizes to support biofilm formation remains a mystery at the molecular level. The human pathogen Vibrio cholerae produces biofilms primarily composed of an exopolysaccharide called VPS (Vibrio polysaccharide), consisting of an unusually-modified repeating tetrasaccharide core unit. VPS engages with two secreted adhesion proteins, Bap1 and RbmC, which adhere the biofilm to abiotic and biotic surfaces, and serve to strengthen the biofilm by interacting with VPS using a conserved β-propeller. To pinpoint the interaction between the adhesins and purified segments of VPS, we determined the ~1.6 Å X-ray crystal structure of Bap1 bound to fragmented VPS and used the structure to carry out molecular dynamics simulations. The structure revealed a single binding site consisting of one tetrasaccharide unit involving an induced magnesium binding site. Unexpectedly, the tetrasaccharide adopted a bent state caused by a rotation of the glycosidic bond between the central two monosaccharide units. Using a combination of mutagenesis, light scattering, and in situ fluorescent microscopy, we demonstrate that Bap1 not only facilitates biofilm adhesion, but is also required for proper VPS organization, through the identified binding pocket. Our structure reveals for the first time the interaction between a biofilm exopolysaccharide and matrix protein, as well as insights into conformational changes of exopolysaccharide induced by this binding. Our findings provide a generalizable approach for studying the biophysical and biochemical properties of carbohydrate-dependent biofilm assembly, which may lead to new ways to treat disease caused by biofilm-forming bacterial pathogens by disrupting the exopolysaccharide-protein interactions.
Open AccessStructure of a key adhesin-exopolysaccharide interaction provides insights into matrix assembly in Vibrio cholerae biofilms
Biofilms serve as a protective mechanism for bacteria, including many pathogens. To form such communities, bacteria secrete macromolecules that form an extracellular matrix serving as a barrier against environmental threats, such as predation, antibiotics, and the host immune system. To be effective, a biofilm must anchor to foreign surfaces and retain sufficient stiffness, in the environment or in a host. However, how this matrix self-organizes to support biofilm formation remains a mystery at the molecular level. The human pathogen Vibrio cholerae produces biofilms primarily composed of an exopolysaccharide called VPS (Vibrio polysaccharide), consisting of an unusually-modified repeating tetrasaccharide core unit. VPS engages with two secreted adhesion proteins, Bap1 and RbmC, which adhere the biofilm to abiotic and biotic surfaces, and serve to strengthen the biofilm by interacting with VPS using a conserved β-propeller. To pinpoint the interaction between the adhesins and purified segments of VPS, we determined the ~1.6 Å X-ray crystal structure of Bap1 bound to fragmented VPS and used the structure to carry out molecular dynamics simulations. The structure revealed a single binding site consisting of one tetrasaccharide unit involving an induced magnesium binding site. Unexpectedly, the tetrasaccharide adopted a bent state caused by a rotation of the glycosidic bond between the central two monosaccharide units. Using a combination of mutagenesis, light scattering, and in situ fluorescent microscopy, we demonstrate that Bap1 not only facilitates biofilm adhesion, but is also required for proper VPS organization, through the identified binding pocket. Our structure reveals for the first time the interaction between a biofilm exopolysaccharide and matrix protein, as well as insights into conformational changes of exopolysaccharide induced by this binding. Our findings provide a generalizable approach for studying the biophysical and biochemical properties of carbohydrate-dependent biofilm assembly, which may lead to new ways to treat disease caused by biofilm-forming bacterial pathogens by disrupting the exopolysaccharide-protein interactions.
- 27 June 2026
- Published by Oxford University Press (OUP) in NAR Genomics and Bioinformatics
- Vol. 8 (3)
LAMBDA: a prophage detection benchmark for genomic language models
Transformer-based genomic sequence models represent an emerging frontier in computational biology. Yet, their embeddings have not yet shown the same level of predictive power as natural and protein language models, highlighting a gap between current implementations and theoretical promise. Existing benchmarks for DNA language models primarily focus on classifying regulatory elements in eukaryotic genomes, leaving open the fundamental question of whether these models learn sequence-level features across whole genomes. We introduce LAMBDA, a benchmark designed to rigorously evaluate genome language model embeddings through phage–bacteria sequence discrimination across four categories of increasing complexity: probing tasks, fine-tuning assessments, diagnostic tests, and genome-wide prophage detection. Our comprehensive analysis of current genomic language models provides insight into the importance of training data selection relative to model size, the need for domain-specific training, and the capabilities and limitations of genomic language models for detecting prophage sequences. This benchmark represents a challenging genomic annotation task in the bacterial domain and addresses a key computational problem with direct relevance to microbiology and medicine.
- 27 June 2026
- Published by Oxford University Press (OUP) in NAR Genomics and Bioinformatics
- Vol. 8 (3)
LAMBDA: a prophage detection benchmark for genomic language models
Transformer-based genomic sequence models represent an emerging frontier in computational biology. Yet, their embeddings have not yet shown the same level of predictive power as natural and protein language models, highlighting a gap between current implementations and theoretical promise. Existing benchmarks for DNA language models primarily focus on classifying regulatory elements in eukaryotic genomes, leaving open the fundamental question of whether these models learn sequence-level features across whole genomes. We introduce LAMBDA, a benchmark designed to rigorously evaluate genome language model embeddings through phage–bacteria sequence discrimination across four categories of increasing complexity: probing tasks, fine-tuning assessments, diagnostic tests, and genome-wide prophage detection. Our comprehensive analysis of current genomic language models provides insight into the importance of training data selection relative to model size, the need for domain-specific training, and the capabilities and limitations of genomic language models for detecting prophage sequences. This benchmark represents a challenging genomic annotation task in the bacterial domain and addresses a key computational problem with direct relevance to microbiology and medicine.
- Open Access
Discovery of BilV reveals a multienzymatic basis for bilirubin reduction across vertebrate gut microbiomes
Gut bacteria reduce bilirubin to urobilinogen, allowing it to be excreted through feces and urine, but studies have long noted a heterogeneous mixture of partially reduced bilirubin-derived intermediates, suggesting that multiple enzymes are involved. Here we identify bilirubin vinyl reductase (BilV), a novel Old Yellow Enzyme family reductase encoded in the genomic neighborhood of the known bilirubin reductase ( bilR ). Using heterologous expression and LC-MS/MS, we show that BilR acts on the methine bridges in the bilirubin reduction pathway; co-expression with BilV enables vinyl-group reduction and complete conversion to urobilinogen. In bacterial genomes, bilV co-occurs primarily with the bilR -insertion subtype and is largely absent alongside bilR -short. Analysis of 1,197 gut metagenomes across 14 vertebrate species reveals that this differential co-occurrence shapes pathway availability across hosts: carnivores and omnivores carry balanced bilR and bilV , whereas avian microbiomes, dominated by bilR -short, are depleted for bilV . These findings establish that bilirubin reduction to urobilinogen involves two enzymes with complementary regioselectivity, and that their distribution across vertebrate gut microbiomes varies in concert with host bile pigment chemistry.
Open AccessDiscovery of BilV reveals a multienzymatic basis for bilirubin reduction across vertebrate gut microbiomes
Gut bacteria reduce bilirubin to urobilinogen, allowing it to be excreted through feces and urine, but studies have long noted a heterogeneous mixture of partially reduced bilirubin-derived intermediates, suggesting that multiple enzymes are involved. Here we identify bilirubin vinyl reductase (BilV), a novel Old Yellow Enzyme family reductase encoded in the genomic neighborhood of the known bilirubin reductase ( bilR ). Using heterologous expression and LC-MS/MS, we show that BilR acts on the methine bridges in the bilirubin reduction pathway; co-expression with BilV enables vinyl-group reduction and complete conversion to urobilinogen. In bacterial genomes, bilV co-occurs primarily with the bilR -insertion subtype and is largely absent alongside bilR -short. Analysis of 1,197 gut metagenomes across 14 vertebrate species reveals that this differential co-occurrence shapes pathway availability across hosts: carnivores and omnivores carry balanced bilR and bilV , whereas avian microbiomes, dominated by bilR -short, are depleted for bilV . These findings establish that bilirubin reduction to urobilinogen involves two enzymes with complementary regioselectivity, and that their distribution across vertebrate gut microbiomes varies in concert with host bile pigment chemistry.
- Open Access
- 14 April 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
SpiR is a gut microbial enzyme that drives cholesterol conversion
The gut microbiota contributes to cholesterol homeostasis by converting cholesterol into coprostanol, a non-absorbable sterol excreted in the feces. However, the enzymes mediating this process remain poorly defined. Here, we identify spiR , a steroid Δ 5-4 isomerase/3-keto reductase from Eubacterium coprostanoligenes that catalyzes the initial oxidation of cholesterol to cholestenone, a requisite step in coprostanol production. We confirm that SpiR oxidizes both cholesterol and pregnenolone, and stereospecifically reduces 3-keto-steroids to 3β-hydroxylated forms. We show that SpiR preferentially binds to cholesterol over related steroids and functions as an NAD(H)-dependent homodimer. Through phylogenetic analysis, we show that spiR clusters with known Δ 5-4 isomerases and is restricted to an uncultured clade within Acutalibacteraceae, where it frequently co-occurs with species encoding ismA , a gene previously implicated in cholesterol conversion. We analyze a multi-omic dataset from three human cohorts and find that spiR homologs were strongly enriched in individuals exhibiting cholesterol conversion. We also show that spiR homologs have a greater predictive power for cholesterol conversion than ismA homologs, establishing them as superior markers of microbial cholesterol metabolism. Our findings refine the enzymatic model of cholesterol metabolism in the gut and establish spiR as a critical biomarker and mechanistic driver for microbiome-mediated cholesterol reduction.
- 14 April 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessSpiR is a gut microbial enzyme that drives cholesterol conversion
The gut microbiota contributes to cholesterol homeostasis by converting cholesterol into coprostanol, a non-absorbable sterol excreted in the feces. However, the enzymes mediating this process remain poorly defined. Here, we identify spiR , a steroid Δ 5-4 isomerase/3-keto reductase from Eubacterium coprostanoligenes that catalyzes the initial oxidation of cholesterol to cholestenone, a requisite step in coprostanol production. We confirm that SpiR oxidizes both cholesterol and pregnenolone, and stereospecifically reduces 3-keto-steroids to 3β-hydroxylated forms. We show that SpiR preferentially binds to cholesterol over related steroids and functions as an NAD(H)-dependent homodimer. Through phylogenetic analysis, we show that spiR clusters with known Δ 5-4 isomerases and is restricted to an uncultured clade within Acutalibacteraceae, where it frequently co-occurs with species encoding ismA , a gene previously implicated in cholesterol conversion. We analyze a multi-omic dataset from three human cohorts and find that spiR homologs were strongly enriched in individuals exhibiting cholesterol conversion. We also show that spiR homologs have a greater predictive power for cholesterol conversion than ismA homologs, establishing them as superior markers of microbial cholesterol metabolism. Our findings refine the enzymatic model of cholesterol metabolism in the gut and establish spiR as a critical biomarker and mechanistic driver for microbiome-mediated cholesterol reduction.
- Open Access
LAMBDA: A Prophage Detection Benchmark for Genomic Language Models
Transformer-based genomic sequence models represent an emerging frontier in computational biology. Yet, their embeddings have not yet shown the same level of predictive power as natural and protein language models, indicating a gap between current implementations and theoretical promise. Existing benchmarks for DNA language models primarily focus on classifying regulatory elements in eukaryotic genomes, leaving open the fundamental question of whether these models learn sequence-level features across whole genomes. We introduce LAMBDA, a benchmark designed to rigorously evaluate genome language model embeddings through phage-bacteria sequence discrimination across four categories of increasing complexity: probing tasks, fine-tuning assessments, diagnostic tests, and genome-wide prophage detection. Our comprehensive analysis of current genomic language models provides novel insights into the importance of training data quality relative to model size, the need for domain-specific training, and the application of genomic language models for detecting prophage sequences. This benchmark represents a challenging genomic annotation task in the bacterial domain and addresses a key computational problem with direct relevance to microbiology and medicine.
Open AccessLAMBDA: A Prophage Detection Benchmark for Genomic Language Models
Transformer-based genomic sequence models represent an emerging frontier in computational biology. Yet, their embeddings have not yet shown the same level of predictive power as natural and protein language models, indicating a gap between current implementations and theoretical promise. Existing benchmarks for DNA language models primarily focus on classifying regulatory elements in eukaryotic genomes, leaving open the fundamental question of whether these models learn sequence-level features across whole genomes. We introduce LAMBDA, a benchmark designed to rigorously evaluate genome language model embeddings through phage-bacteria sequence discrimination across four categories of increasing complexity: probing tasks, fine-tuning assessments, diagnostic tests, and genome-wide prophage detection. Our comprehensive analysis of current genomic language models provides novel insights into the importance of training data quality relative to model size, the need for domain-specific training, and the application of genomic language models for detecting prophage sequences. This benchmark represents a challenging genomic annotation task in the bacterial domain and addresses a key computational problem with direct relevance to microbiology and medicine.
- Open Access
- 26 November 2025
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 53 (22)
Functional analyses of bacterial NanoRNase B proteins reveals defining features of this enzyme family
A combination of exoribonucleases and endoribonucleases degrades RNA polymers to recycle nucleoside monophosphates. A byproduct of these reactions is the accumulation of short RNAs, 2–5 nucleotides in length. Characteristic enzymes, generally referred to as nanoRNases, specifically process short RNAs. Genes encoding nanoRNases are essential in some bacteria; therefore, it is assumed that the accumulation of short RNAs is detrimental to cells. However, the substrate preferences and enzymatic mechanisms of the known categories of nanoRNase enzymes have not been equally investigated. The NrnB category of nanoRNases has been particularly understudied. In this study, we identified bacterial NrnB homologs and discovered they can be grouped into three classes of proteins, which can be identified by their characteristic sequence features. Purified representatives of these classes of proteins revealed that they all process RNA substrates from the 3′-terminus. The presence of sequence features at the C-terminus was shown to be diagnostic for general exoribonuclease activity against long RNA substrates, whereas the absence of these C-terminal elements was correlated with proteins that preferentially acted against shorter RNA substrates. Together, these data define members of the overall NrnB family of nanoRNase proteins and identify some of their key features.
- 26 November 2025
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 53 (22)
Open AccessFunctional analyses of bacterial NanoRNase B proteins reveals defining features of this enzyme family
A combination of exoribonucleases and endoribonucleases degrades RNA polymers to recycle nucleoside monophosphates. A byproduct of these reactions is the accumulation of short RNAs, 2–5 nucleotides in length. Characteristic enzymes, generally referred to as nanoRNases, specifically process short RNAs. Genes encoding nanoRNases are essential in some bacteria; therefore, it is assumed that the accumulation of short RNAs is detrimental to cells. However, the substrate preferences and enzymatic mechanisms of the known categories of nanoRNase enzymes have not been equally investigated. The NrnB category of nanoRNases has been particularly understudied. In this study, we identified bacterial NrnB homologs and discovered they can be grouped into three classes of proteins, which can be identified by their characteristic sequence features. Purified representatives of these classes of proteins revealed that they all process RNA substrates from the 3′-terminus. The presence of sequence features at the C-terminus was shown to be diagnostic for general exoribonuclease activity against long RNA substrates, whereas the absence of these C-terminal elements was correlated with proteins that preferentially acted against shorter RNA substrates. Together, these data define members of the overall NrnB family of nanoRNase proteins and identify some of their key features.