Clostridium scindens ATCC 35704: Integration of Nutritional … · Clostridium scindens ATCC 35704: Integration of Nutritional Requirements, the Complete Genome Sequence, and Global

Clostridium scindens ATCC 35704: Integration of NutritionalRequirements, the Complete Genome Sequence, and GlobalTranscriptional Responses to Bile Acids

Saravanan Devendran,a,b Rachana Shrestha,c João M. P. Alves,d Patricia G. Wolf,b Lindsey Ly,a,e Alvaro G. Hernandez,f

Celia Méndez-García,g Ashley Inboden,c J’nai Wiley,c Oindrila Paul,c Avery Allen,c Emily Springer,c Chris L. Wright,e

Christopher J. Fields,e Steven L. Daniel,c Jason M. Ridlona,b,e,h,i

aMicrobiome Metabolic Engineering Theme, Carl R. Woese Institute for Genomic Biology, Urbana, Illinois, USAbDepartment of Animal Sciences, University of Illinois at Urbana-Champaign, Urbana, Illinois, USAcDepartment of Biological Sciences, Eastern Illinois University, Charleston, Illinois, USAdDepartment of Parasitology, Institute of Biomedical Sciences, University of São Paulo, São Paulo, BrazileDivision of Nutritional Sciences, University of Illinois at Urbana-Champaign, Urbana, Illinois, USAfKeck Center for Biotechnology, University of Illinois at Urbana-Champaign, Urbana, Illinois, USAgIndependent Researcher, Boston, Massachusetts, USAhCancer Center of Illinois, University of Illinois at Urbana-Champaign, Urbana, Illinois, USAiDepartment of Microbiology and Immunology, School of Medicine, Virginia Commonwealth University, Richmond, Virginia, USA

ABSTRACT In the human gut, Clostridium scindens ATCC 35704 is a predominantbacterium and one of the major bile acid 7�-dehydroxylating anaerobes. While thisorganism is well-studied relative to bile acid metabolism, little is known about thebasic nutrition and physiology of C. scindens ATCC 35704. To determine the aminoacid and vitamin requirements of C. scindens, the leave-one-out (one amino acidgroup or vitamin) technique was used to eliminate the nonessential amino acidsand vitamins. With this approach, the amino acid tryptophan and three vitamins (ri-boflavin, pantothenate, and pyridoxal) were found to be required for the growth ofC. scindens. In the newly developed defined medium, C. scindens fermented glucosemainly to ethanol, acetate, formate, and H2. The genome of C. scindens ATCC 35704was completed through PacBio sequencing. Pathway analysis of the genome se-quence coupled with transcriptome sequencing (RNA-Seq) under defined cultureconditions revealed consistency with the growth requirements and end products ofglucose metabolism. Induction with bile acids revealed complex and differential re-sponses to cholic acid and deoxycholic acid, including the expression of potentiallynovel bile acid-inducible genes involved in cholic acid metabolism. Responses totoxic deoxycholic acid included expression of genes predicted to be involved inDNA repair, oxidative stress, cell wall maintenance/metabolism, chaperone synthesis,and downregulation of one-third of the genome. These analyses provide valuable in-sight into the overall biology of C. scindens which may be important in treatment ofdisease associated with increased colonic secondary bile acids.

IMPORTANCE C. scindens is one of a few identified gut bacterial species capable ofconverting host cholic acid into disease-associated secondary bile acids such as de-oxycholic acid. The current work represents an important advance in understandingthe nutritional requirements and response to bile acids of the medically importanthuman gut bacterium, C. scindens ATCC 35704. A defined medium has been devel-oped which will further the understanding of bile acid metabolism in the context ofgrowth substrates, cofactors, and other metabolites in the vertebrate gut. Analysis ofthe complete genome supports the nutritional requirements reported here. Genome-wide transcriptomic analysis of gene expression in the presence of cholic acid and de-

Citation Devendran S, Shrestha R, Alves JMP,Wolf PG, Ly L, Hernandez AG, Méndez-García C,Inboden A, Wiley J, Paul O, Allen A, Springer E,Wright CL, Fields CJ, Daniel SL, Ridlon JM. 2019.Clostridium scindens ATCC 35704: integration ofnutritional requirements, the completegenome sequence, and global transcriptionalresponses to bile acids. Appl Environ Microbiol85:e00052-19. https://doi.org/10.1128/AEM.00052-19.

Editor Volker Müller, Goethe University,Frankfurt am Main

Copyright © 2019 American Society forMicrobiology. All Rights Reserved.

Address correspondence to Jason M. Ridlon,[email protected].

S.D. and R.S. contributed equally to this work.

Received 8 January 2019Accepted 30 January 2019

Accepted manuscript posted online 8February 2019Published

PHYSIOLOGY

crossm

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 1Applied and Environmental Microbiology

22 March 2019

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

oxycholic acid provides a unique insight into the complex response of C. scindens ATCC35704 to primary and secondary bile acids. Also revealed are genes with the potential tofunction in bile acid transport and metabolism.

KEYWORDS Clostridium scindens, RNA-Seq, bile acid, defined medium, deoxycholicacid, growth factor requirements

Humans synthesize primary bile acids, cholic acid (CA; 3�,7�,12�-trihydroxy-5�-cholan-24-oic acid) and chenodeoxycholic acid (CDCA; 3�,7�-dihydroxy-5�-

cholan-24-oic acid), from cholesterol in the liver (1, 2). These two primary bile acids,conjugated to taurine or glycine, are secreted into the small intestine and play vitalroles in digestion of dietary lipids (1, 2). However, several hundred milligrams ofunconjugated CA and CDCA are converted to the toxic secondary bile acids deoxy-cholic acid (DCA; 3�,12�-dihydroxy-5�-cholan-24-oic acid) and lithocholic acid (LCA;3�-monohydroxy-5�-cholan-24-oic acid), respectively, in the human colon each day (3).Removal of the 7�-hydroxyl group requires a multienzyme pathway known as the bileacid 7�-dehydroxylation pathway (4).

The bile acid 7�-dehydroxylation pathway is the result of a bile acid-inducible (bai)regulon in a small number of identified Firmicutes in the genus Clostridium, includingClostridium scindens and Clostridium hylemonae in cluster XIVa, Clostridium hiranonis,Clostridium sordellii, and Clostridium bifermentans in cluster XI, and Clostridium leptum incluster IV (5). C. scindens is considered a high-activity bile acid 7�-dehydroxylatingbacterium, with reported activity �100-fold higher than that of cluster XI and IVmembers (6). The first C. scindens isolate was reported by White et al. (1980) and namedEubacterium sp. strain VPI 12708 (7). It was in this study that the first indication that thegenes encoding enzymes that convert CA to DCA and CDCA to LCA, respectively, wereinduced by CA but not DCA (7). Decades later, Eubacterium VPI 12708 was reclassifiedto C. scindens VPI 12708 (8). Shortly after the initial first report of Eubacterium VPI 12708,Clostridium strain 19, a human fecal strain capable of the side chain cleavage of cortisolto 11�-hydroxyandrostendione, was isolated (9). Clostridium strain 19 was renamedClostridium scindens; the specific epithet “scindens” means “to cut,” owing to thesteroid-17,20-desmolase activity observed following incubation with cortisol (10). ThisC. scindens isolate became the type strain (ATCC 35704) and was also shown to convertCA to DCA (10). C. scindens ATCC 35704 is a chemoheterotrophic, endospore-formingobligate anaerobe (11). Cells are nonmotile, occurring singly or in chains, and spores areterminal (11). Previous studies have identified carbohydrates such as D-glucose,D-lactose, D-fructose, D-mannose, D-ribose, and D-xylose as carbon and energy sourcesfor C. scindens (11). C. scindens does not produce enzymes such as lecithinase, lipase, orcatalase and is unable to digest gelatin, milk, and meat (11). This bacterium is alsoincapable of reducing nitrate or hydrolyzing starch and esculin; however, hydrogensulfide is produced in sulfide-indole motility medium (11).

Molecular cloning and enzymology of bile acid-inducible (bai) genes (3) revealed acomplex, multistep biochemical pathway distinct from the two-step mechanism orig-inally proposed by Bergstrom et al. (12). Recently, crystal structures and catalyticmechanisms for the NAD(H)-dependent 3�-hydroxysteroid dehydrogenase (BaiA) andthe rate-limiting bile acid 7�-dehydratase (BaiE) have been reported (13, 14). Surpris-ingly, very little is known about nutritional requirements and end products of glucosemetabolism by C. scindens ATCC 35704. Bile acid 7�-dehydroxylation is a redox process,resulting in a net 2-electron reduction involving flavin and pyridine nucleotides (3, 4).Global transcriptional responses to CA and DCA are predicted to enhance our under-standing of the bai regulon, overall physiological changes to bile acid substrates andproducts, and microbial adaptation to the toxic and detergent nature of bile acids.

Global transcriptional analyses of bile acid 7�-dehydroxylating bacteria after bileacid induction have yet to be reported. C. scindens ATCC 35704 has traditionally beencultivated in complex, undefined medium (UM) for bile acid studies (6–11, 15, 16).Problematically for bile acid transformation and transcriptional studies, complex media

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 2

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

do not allow exchange or limitation of nutrients, and, being composed partially ofanimal by-products (e.g., peptone and brain heart infusion [BHI]), contain bile acids andother lipids (17). The purpose of this study was to define the nutritional requirementsof C. scindens ATCC 35704, to identify major end products of glucose fermentationduring growth in a defined medium, and to determine transcriptional responses to bileacids under defined culture conditions. Here, we report development of a definedmedium (DM) that supports the growth of C. scindens ATCC 35704. The genome of C.scindens ATCC 35704 has been closed through PacBio sequencing. The genomic dataprovide insight into the nutritional requirements of C. scindens and allowed compre-hensive scaffolding of transcriptome sequencing (RNA-Seq) data during growth in DM.Our results describe the transcriptional response of C. scindens ATCC 35704 to thenewly formulated DM, as well as to primary and secondary bile acids.

RESULTSDevelopment of defined and minimal media for C. scindens ATCC 35704. To

determine amino acid and vitamin requirements of C. scindens, a leave-one-out tech-nique was utilized to eliminate nonessential amino acids and vitamins (data notshown). Tryptophan was the only amino acid that when omitted precluded growth ofC. scindens ATCC 35704. When tryptophan was the sole amino acid provided in DM,growth was supported, signifying that tryptophan was the only amino acid required forgrowth by C. scindens ATCC 35704 (see Fig. S1 in the supplemental material). C. scindensATCC 35704 thus displays auxotrophy for tryptophan. When vitamins were omitted oneat a time, C. scindens ATCC 35704 was found to clearly require riboflavin and panto-thenic acid, and the reduced growth observed after three sequential transfers in theabsence of pyridoxal HCl suggested that this vitamin might also be a growth factor forC. scindens ATCC 35704 (Fig. S2A). Indeed, only when the combination of riboflavin,pantothenic acid, and pyridoxal HCl were present together was growth maintainable(Fig. S2B). Based on these findings, a minimal medium (MM) was developed; MM wasDM modified to contain riboflavin, pantothenic acid, and pyridoxal HCl as sole vitaminsand tryptophan as the sole amino acid.

In MM and PO4-buffered MM cultures, glucose-dependent growth by C. scindensATCC 35704 was slow, and cell yields were reduced compared to levels with DM andPO4-buffered DM cultures (Fig. 1A). These findings suggest that the sparse nutrientsfound in minimal medium were a metabolic (e.g., anabolic) challenge to C. scindensATCC 35704. Doubling times for cells grown in BHI, UM, DM, and PO4-buffered DMapproximated 2.0, 2.3, 3.3, and 7.0 h, respectively. While previous work has shown thatCA is 7�-dehydroxylated to DCA by cells grown in BHI medium (15), we report here CAconversion to DCA by C. scindens ATCC 35704 grown in a defined or minimal medium(Fig. 1B). Furthermore, growth in PO4-buffered DM was not altered by addition of a finalconcentration of 100 �M CA or DCA (Fig. 1C). The bile acid concentration chosenreflects both in vivo physiologically relevant concentrations of bile acids in fecal water(18, 19) and concentrations used previously in numerous in vitro studies of C. scindensstrains (6, 10, 15). Separate additions of CA in early and mid-log phase have beenpreviously shown to result in robust induction of the bai regulon (7).

To obtain cells for RNA isolation, C. scindens ATCC 35704 was acclimated to PO4-buffered DM (control) or to PO4-buffered DM with 100 �M CA or DCA by two consec-utive 24-h transfers. On the third transfer, 50 �M CA, 50 �M DCA, or methanol vehiclecontrol (�5%, vol/vol) was introduced at the time of inoculation and again at an opticaldensity at 600 nm (OD600) of 0.25 for a final concentration of 100 �M. At an OD600 of0.45, the rate of conversion of [24-14C]CA to [24-14C]DCA over a 20-min interval wasdetermined from 1 ml of culture to be 0.013 � 6.5E�4 nmol min�1 mg�1 for controlwhole cells, 0.09 � 0.025 nmol min�1 mg�1 for CA-induced whole cells, and 0.017 �

4.1E�3 nmol min�1 mg�1 for DCA-induced whole cells (Fig. 1D). After it was confirmedthat CA and DCA at 100 �M did not significantly alter growth patterns for C. scindensATCC 35704 and that CA was converted to DCA in CA-induced, but not DCA or vehicle

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 3

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

control whole cells, RNA was isolated from RNAprotect-treated cells for genome-widetranscriptomics.

Complete genome of Clostridium scindens ATCC 35704 and RNA-Seq analyses.We next wanted to complete the genome of C. scindens ATCC 35704 in order to assessthe genomic basis for the determined nutritional requirements, as well as to obtain acomplete scaffold for mapping mRNA sequencing reads. The closed genome of C.scindens ATCC 35704 was obtained by PacBio sequencing using SMRT (single-moleculereal-time) technology. The genome consists of a single circular chromosome of3,658,040 bp, containing 3,657 coding sequences (CDS), 12 rRNA genes, 4 rRNA cistrons,

FIG 1 Growth and bile acid metabolism by Clostridium scindens ATCC 35704. (A) Growth profiles after a minimum of three sequential transfersunder each culture condition: �, brain heart infusion (BHI) broth; Œ, undefined medium (UM; DM with 0.1% yeast extract); Δ, defined medium(DM); Œ, PO4-buffered DM; �, minimal medium (MM); and e, PO4-buffered MM. (B) Bile acid metabolism in DM and MM (left to right): lane 1,cholate (CA) standard; lane 2, deoxycholate (DCA) standard; lane 3, DM and C. scindens 35704; lane 4, DM with CA, and C. scindens 35704; lane5, DM modified to contain tryptophan as the sole amino acid plus CA and C. scindens 35704; lane 6, DM modified to contain riboflavin, pantothenicacid, and pyridoxal HCl as sole vitamins plus CA and C. scindens 35704; and lane 7, MM with CA and C. scindens 35704. The initial concentrationof CA and DCA in standards and CA in cultures was 100 �M. (C) Growth profiles in PO4-buffered DM in the absence (control, Œ) and presenceof cholic acid (CA, e) or deoxycholic acid (DCA, �). Arrows indicate addition of 50 �M bile acid, and the star indicates addition of RNAprotectfor RNA-Seq analysis as well as removal of a 1-ml aliquot for the [24-14C]CA conversion assay. (D) Determination of the rate of conversion of[24-14C]CA to [24-14C]DCA. The inset is the autoradiograph of TLC-separated [24-14C]CA metabolites. CA and 7-oxo-DCA were scraped andcounted, and DCA and 3-oxo-DCA were scraped and counted.

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 4

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

each containing the three expected genes (5S, 16S, and 23S), and 58 tRNA genes (Fig.2). Annotation of the genome reveals that most of the biosynthetic pathways consid-ered essential for growth and viability appear to be present (e.g., biosynthesis ofvitamins, amino acids, purines, and pyrimidines). Importantly, and consistent withresults from our nutritional requirement studies, genes involved in the biosynthesis oftryptophan, riboflavin, pyridoxal phosphate, and pantothenic acid appear to be absent.Tryptophan was the only amino acid determined necessary for growth of C. scindensATCC 35704 (Fig. S1). C. scindens ATCC 35704 has the genes for the complete shikimatepathway leading to chorismate and then to prephenate, whose bifurcation results inthe formation of phenylalanine and tyrosine. Routes into the shikimate pathwayinclude the pentose-phosphate pathway (erythrose-4-phosphate) and quinic acid.However, C. scindens ATCC 35704 lacks genes involved in anthranilate synthesis andmetabolism (trpEGDFC). Tryptophan synthetase (trpAB) genes appear to be present,suggesting both the production of indole from tryptophan and potentially the synthe-sis of L-tryptophan through the acquisition of indole (0.25 to 1.1 mM in feces) producedby other gut bacteria (20, 21). Tryptophan auxotrophy may suggest a strategy to limitgrowth of C. scindens in vivo by inhibiting tryptophanase (22).

To determine pathways that were expressed during growth in PO4-buffered DM,which may indicate the synthesis or transport of vitamins and other nutrients, weperformed RNA-Seq analysis. After removal of rRNA and TruSeq mRNAseq libraryconstruction, Illumina HiSeq 4000 sequencing resulted in �15 to 18 million reads persample. The average number of transcripts per million (TPM) among annotated genes(minus tRNA and rRNA) was 273.5 for control samples. Of the 3,656 annotated genes,1,867 genes were at or below 50 TPM, and 3,486 genes were below 1,000 TPM. Only adozen open reading frames (ORFs) were above 10,000 TPM (Data Set S1).

Riboflavin was determined to be a necessary vitamin (Fig. S2), and the organismlacks the genes involved in biosynthesis of riboflavin (ribABCGOT) except for twoputative riboflavin transporters, ribU1 (4,020 � 1,257 TPM) and ribU2 (HDCHBGLK_02697; 1,151 � 411 TPM), whose predicted genes were highly expressed. C. scindensATCC 35704 does have the gene for riboflavin kinase (ribF; HDCHBLK_03126) required

FIG 2 The complete genome of Clostridium scindens ATCC 35704. The genome map describing in order,outside to inside, coding sequences (CDS) in the plus strand, CDS in the minus strand, rRNA genes inthe plus strand, rRNA genes in the minus strand, tRNA genes in the plus strand, tRNA genes in theminus strand, CG content plot (windows of 10 kbp, steps of 200 bp), and CG skew plot [(G �C)/(G � C)].

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 5

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

for the conversion of riboflavin to the enzyme cofactors flavin mononucleotide (FMN)and flavin adenine dinucleotide (FAD�), respectively (Data Set S1). Relative to itspantothenic acid requirement, C. scindens ATCC 35704 harbors genes involved in theconversion of pyruvate to �-ketoisovalerate, as well as a gene predicted to encodebranched-chain amino acid transferase (llvE), generating L-valine. However, the genesinvolved in conversion of �-ketoisovalerate to pantoate (panB and panE) and then topantothenate (panC) are absent in the genome. A gene predicted to encode extracy-toplasmic function (ECF)-dependent pantothenate transporter (panT) (HDCHB-GLK_02231) is expressed (2,039 � 166.7 TPM). The complete biosynthetic pathwayleading to the conversion of pantothenate to coenzyme A is, however, present in C.scindens ATCC 35704.

C. scindens synthesizes pyridoxal 5=-phosphate synthase and pyridoxal kinase; how-ever, genes involved in de novo synthesis of B6 derivatives are not present in thegenome, supporting a nutritional requirement for B6. The lipoic acid salvage pathwaygenes (lae, lipT1, and lplA) are present but not the biosynthesis genes (lipA and lipB).While biotin was not determined to be necessary, the majority of genes were identified(fabF, fabG, fabZ, fabI, bioH, and bioB), but we were not able to locate dethiobiotinsynthase (bioD) or DAPA (7,8-diaminopelargonic acid) aminotransferase (bioA). Thus,the combination of leave-one-nutrient-out analysis with genomics and transcriptomicsexplains and supports the nutrient requirements of C. scindens ATCC 35704.

Transcriptional response of bile acid-metabolizing genes to CA and DCA. C.

scindens ATCC 35704 is a prominent human gut bile acid 7�-dehydroxylating bacte-rium; however, the global transcriptional response of this bacterium to bile acids hasnot been reported. Illumina HiSeq sequencing of cDNA derived from total RNA isolatedfrom CA-induced (n � 2) and DCA-induced (n � 2) whole cells of C. scindens ATCC35704 was performed and compared to results with vehicle control cells (n � 2). Weidentified 1,430 genes significantly differentially regulated by CA (�1.5-fold change;false discovery rate [FDR], �0.05) with 697 upregulated and 733 downregulated genes(Fig. 3 to 5 and Data Sets S1 and S2). Growth in the presence of DCA resulted in 684genes significantly upregulated and 1,033 genes downregulated (Fig. 3 to 5 and DataSets S1 and S3). Multiple-dimensional scaling of control, CA, and DCA transcriptomesshowed adequate separation of control and treatment groups (Fig. S3). CA and DCAshare 897 differentially expressed genes, while 278 and 207 are unique to CA and DCA,respectively (Fig. S3). Since induction with CA leads to DCA formation during growth,shared genes likely reflect some combination of DCA-induced genes or genes differ-entially regulated by bile acids generally. Distinguishing between these possibilities willhave to await the genetic manipulation of C. scindens.

Genes that were significantly upregulated and downregulated were placed intoclusters of orthologous groups (COGs) using eggNOG (62). DCA significantly alteredtranscriptional responses from group C (energy production and conservation;P � 0.001) and group S (unknown function; P � 0.004). CA also significantly alteredgroup C (P � 0.0004) but led to a significant downregulation from group L (replicationand repair; P � 0.001) (Fig. 3). As expected, genes encoding bai enzymes were amongthe most highly expressed genes in the genome in the presence of CA (Fig. 4A and 5A)but were significantly downregulated by DCA, with the exception of the baiA1 (Fig. 4Band 5B). This observation was confirmed by quantitation of baiE transcript by quanti-tative reverse transcription-PCR (qRT-PCR), which demonstrated significant inductionby CA treatment (5.79 � 0.56 log2 fold change [log2FC]; P � 0.01) but not by DCAtreatment (1.13 � 0.56 log2FC; P � 0.4) relative to the level of the control (Fig. 5C).Figure 6A shows sequencing coverage across the bai operon as well as the biochem-istry of the CA 7�-dehydroxylation pathway. In contrast, the baiN gene (HDCHB-GLK_03018), previously shown to encode a flavoprotein involved in the reductive armof the bile acid 7�-dehydroxylation pathway, was constitutively expressed at anaverage of 232 � 63.6 TPM (Data Set S1) (24). The recently reported NADP-dependent12�-hydroxysteroid dehydrogenase (HSDH) (HDCHBGLK_00743) characterized from C.

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 6

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

scindens ATCC 35704 preferentially converted 12-oxo-LCA to DCA relative to CA me-tabolites (15). This gene was not differentially regulated by CA or DCA (15). Similarly,a gene encoding NADP-dependent 7�-HSDH, sharing 98% amino acid identity withNADP-dependent 7�-HSDH (HDCHBGLK_01258) characterized from C. scindens VPI12708 (25), was also constitutively expressed at an average of 640 � 121.5 TPM(Fig. 6B).

The expression of a gene predicted to encode tryptophan-rich protein TspO wassignificantly upregulated by CA (3.56 log2FC; P � 6.8E�04; FDR � 0.003) but down-regulated, although not statistically significantly, by DCA (0.7 log2FC; P � 0.53;FDR � 0.69) (Fig. 5A and Data Sets S2 and S3). We previously reported a gene encoding

FIG 3 eggNOG category analysis of RNA-Seq data sets after induction with cholic acid (CA) or deoxycholic acid (DCA) in PO4-buffered defined medium. (A)Orthologous group category of CA-downregulated and CA-upregulated genes relative to number of genes per group in the C. scindens ATCC 35704 genome,as indicated. (B) Orthologous group category of DCA-downregulated and DCA-upregulated genes relative to number of genes per group in the C. scindens ATCC35704 genome, as indicated. Orthologous group categories displayed on the x axis are as follows: A, RNA processing and modification; B, chromatin structureand dynamics; C, energy conversion; D, cell cycle control and mitosis; E, amino acid metabolism and transport; F, nucleotide metabolism and transport; G,carbohydrate metabolism and transport; H, coenzyme metabolism; I, lipid metabolism; J, translation; K, transcription; L, replication and repair; M, cellwall/membrane/envelope biogenesis; N, cell motility; NH, no hits/unknown function; O, posttranslational modification, protein turnover, chaperone functions;P, inorganic ion transport and metabolism; Q, secondary structure; S, function unknown; T, signal transduction; U, intracellular trafficking and secretion; andV, defense mechanisms. Only genes with a P value of �0.05 and a log2FC value greater than �0.58 were included. Statistical significance (*; Bonferroni-correctedP � 0.005) was determined by Fisher’s exact test.

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 7

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

a TspO/mitochondrial benzodiazepine receptor (MBR) family protein in C. scindens VPI12708 downstream and on the opposite strand of the monocistronic baiA2 andupstream of a bai promoter region and the baiJKL operon (26). While the role of theTspO/MBR family in bile acid metabolism (if any) is currently unknown, this protein has

FIG 4 Log2 fold change versus log counts per million (CPM) scatterplot of Clostridium scindens ATCC 35704transcriptome data after cultivation in PO4-buffered defined medium with cholic acid (CA) or deoxycholicacid (DCA). (A) Scatterplot of CA data versus the control levels showing upregulated and downregulatedgenes. (see Data Set S2 in the supplemental material). (B) Scatterplot of DCA data versus control levels (DataSet S3). Labeled arrows represent genes in the bile acid-inducible (bai) operon, which is induced by CA.Upregulated and downregulated genes were defined at a threshold of P � 0.05 and a log2FC value greaterthan �0.58. non-DE, not differentially expressed.

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 8

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

been shown previously to function in the import and metabolism of steroids and,potentially, in the prevention of oxidative stress (27). The position of the gene encodingthe TspO/MBR family protein, downstream of baiA, and induction by CA but not DCAmay indicate that this gene is involved in bile acid metabolism or has an importantfunction during high intracellular concentrations of bile acid intermediates and endproducts (Fig. 5A and 6B).

Similarly, a gene encoding a predicted pyridine nucleotide-dependent flavoprotein(urocanate reductase, urdA; HDCHBGLK_03451) is upregulated by CA (3.82 log2FC;P � 1.61E�28; FDR � 5.35E�26) but not by DCA (Fig. 5A). This observation wasconfirmed by qRT-PCR demonstrating significant induction by CA treatment(7.03 � 2.15 log2FC; P � 0.002) but not by DCA treatment (0.9 � 0.38 log2FC; P � 0.58)

FIG 5 Differential expression of genes from Clostridium scindens ATCC 35704 grown in PO4-buffered defined medium with cholic acid (CA) or deoxycholic acid(DCA). (A) Genes upregulated by CA and DCA relative to the level with DM. (B) Genes downregulated by CA and DCA. False discovery rate (Bonferroni correction)is displayed along with predicted gene annotation. (C) qRT-PCR of the bile acid 7�-dehydratase (baiE gene) and the “urocanate reductase” gene predicted toencode a bile acid 5�-reductase. Results are based on four biological replicates using the 2�ΔΔCT method (reported as standard error of the mean) andnormalized relative to the level of the housekeeping gene recA. See Materials and Methods for primer sequences. *, P � 0.05.

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 9

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

(Fig. 5C). The predicted urdA gene product shares 47.5% amino acid identity with thebaiJ gene product (62-kDa gene product) encoding a predicted flavoprotein in C.scindens VPI 12708, which is part of a polycistronic operon that includes the baiK gene.The baiK gene encodes a bile acid coenzyme A (CoA) transferase; however, thefunctions of the remaining genes are currently unknown (26). Previous reports byHylemon et al. (1991) identified allo-deoxycholic acid (5�-reduced) as a cholic acid-inducible side product of bile acid 7�-dehydroxylation (28). The finding that allo-bileacids are formed leads to the prediction that at least two genes encoding stereospecificflavin-dependent oxidoreductases encoding bile acid 5�-reductase and bile acid 5�-reductase. We recently reported an NAD(H)-dependent flavoprotein, baiN, which cat-alyzes two steps in the reductive arm of the bile acid 7�-dehydroxylation pathway(3-dehydro-4,6-DCA ¡ 3-dehydro-4-DCA ¡ 3-dehydro-DCA) (24). The baiN gene,encoding a bile acid 5�-reductase, is not differentially regulated by bile acids but isinstead constitutively expressed (�200 TPM). It is intriguing to identify another genethat is predicted to catalyze C�C bond formation (Fig. 6C) but that is differentiallyregulated by primary bile acids. Further research is needed to determine the role ofurdA, particularly as a bile acid 5�-reductase.

FIG 5 (Continued)

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 10

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

Classical bile acid efflux permeases were studied in Escherichia coli, encoded by theacrAB genes (29) whose homologs are absent in the genome of C. scindens ATCC 35704.There are currently no candidates for bile acid efflux permeases in bile acid 7�-dehydroxylating bacteria, which may represent a potential drug target to inhibitgrowth of these bacteria during chronic DCA excess. Growth in the presence of bileacids resulted in significant expression of genes predicted to encode membrane-spanning efflux permease proteins (Table S2). A few candidates include the putativemultidrug export permease ygaD (HDCHBGLK_00878; 2.14 log2FC; FDR � 5.0E�04),putative ABC transporter yxlF (HDCHBGLK_01721; 1.83 log2FC, FDR � 2.65E�08), andmultidrug-resistance protein 3 (HDCHBGLK_02921; 1.80 log2FC; FDR � 2.58E�09).Table S2 lists expression levels of numerous genes (�0.58 log2FC; �0.05 FDR) predictedto play a role in response to bile acid stress, consistent with previous studies (30–38).Future studies should address the identity and number of bile acid export proteins andthe importance that these and other proteins have in the growth and bile acidmetabolism of C. scindens.

Carbohydrate screening and end product analysis of glucose metabolism by C.scindens. Conversion of primary to secondary bile acids by C. scindens results in a net2-electron reduction which is flavin and pyridine nucleotide dependent (39). Thus,primary bile acid metabolism provides electron acceptors, which are coupled to thefermentation of sugars and potentially other substrates as well in vivo. However, a basicunderstanding of fermentable sugars and the types and amounts of end productsformed during fermentation by C. scindens is lacking. We thus, examined the ability ofC. scindens to metabolize different carbohydrates as well as their genomic basis andexpression under chemically defined culture conditions, providing key insights into thephysiology of this bacterium. Replacement of glucose in DM with various simple andcomplex carbohydrates was performed to determine growth by C. scindens ATCC35704. Out of 40 carbohydrates examined, only 6 monosaccharides (glucose, fructose,mannose, galactose, ribose, and xylose), 2 sugar alcohols (dulcitol and sorbitol), and 1disaccharide (lactose) were growth supportive (Table S1). Carbohydrate metabolismtherefore appears to be largely restricted to a few monosaccharides and the disaccha-ride lactose.

Analysis of the annotated genome provided confirmation of observed sugar fer-mentation profiles. Genes involved in lactose metabolism were located in the genome

FIG 5 (Continued)

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 11

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

sequence. Lactose is cleaved by �-galactosidase (bglY; HDCHBGLK_02011), converted to1-phospho-galactose by galactokinase (HDCHBGLK_01716), and then transferred toUDP-�-glucose by UDP-glucose-1-phosphate uridylyltransferase (HDCHBGLK_01717), forming UDP-�-D-galactose. The last is then converted to UDP-�-D-glucoseby UDP-glucose 4-epimerase (galE; HDCHBGLK_02011) and aldose-1-epimerase(galM; HDCHBGLK_03553), while lacF (HDCHBGLK_02016) and lacG (HDCHBGLK_02017) are located downstream of lacZ and galE. Upstream of galE is a cryptichigh-molecular-weight �-galactosidase (HDCHBGLK_02009). As expected, the ex-pression levels (as TPM) of genes predicted to hydrolyze lactose and metabolizegalactose were low, given that glucose, rather than lactose, was present in DM (DataSet S1).

L-Xylose is also fermented by C. scindens ATCC 35704 (Table S1). Genes predictedto encode a xylose-importing ATP-binding protein (xylG; HDCHBGLK_01674) andxylulose kinase (HDCHBGLK_00721) were located in the genome, as was a gene for

FIG 6 Mapped RNA-Seq counts against genes involved in bile acid metabolism by Clostridium scindens ATCC 35704. (A) Reads mapped againstthe polycistronic bai operon, along with biochemistry of cholic acid 7�-dehydroxylation. Metabolites are as follows: I, cholic acid; II, cholyl�CoA;III, 3-dehydrocholyl�CoA; IV, 3-dehydro-4-cholyl�CoA; V, 3-dehydro-4,6-deoxycholyl�CoA; VI, 3-dehydro-deoxycholyl�CoA; VII, deoxycholyl-�CoA; VIII, deoxycholic acid. (B) Reads mapped to bile acid 3�-, 7�-, and 12�-hydroxysteroid dehydrogenase genes. The CA-induced TspO genepredicted to be involved in bile acid metabolism is located downstream of the baiA gene. (C) Reads mapped to a CA-inducible gene encodinga predicted flavoprotein hypothesized to generate allo-(5�)-deoxycholic acid. Conversion from compound V to allodeoxycholic acid predicted tobe in two steps, with the BaiA1 catalyzing the final NADH-dependent reductive step.

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 12

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

a predicted mannose-6-phosphate isomerase (HDCHBGLK_01361). The sugar alco-hol sorbitol is also metabolized (Table S1), and we located two genes encodingputative sorbitol dehydrogenases (HDCHBGLK_00843; HDCHBGLK_01676). Whilethe complex carbohydrates tested in our study were not metabolized, C. scindensATCC 35704 does have the genes for a predicted neopullulanase 1 (HDCHB-GLK_01304) as well as an acetylxylan esterase (HDCHBGLK_03158), suggesting thatmetabolism of certain complex dietary or host-derived carbohydrates cannot beruled out.

Next, we determined the end products of glucose metabolism in PO4-buffered DMand in PO4-buffered MM (Table 1). After 24 h of growth in PO4-buffered DM, glucosewas not detected, indicating that C. scindens ATCC 35704 completely consumed theglucose (19.4 mM). The major end products formed were ethanol (24.6 mM), acetate(10.7 mM), formate (6.3 mM), and H2 (16.9 mM), whereas other products were producedin very negligible (� 1 mM) amounts. Based on substrate-product profiles of glucose-grown cells, carbon recovery was 72% and may be attributed to the fact that CO2 levelsand biomass carbon production were not determined. In PO4-buffered MM, C. scindensATCC 35704 converted glucose essentially to the same products as cells grown inPO4-buffered DM. However, the concentrations of most end products in PO4-bufferedMM were lower due to the decreased glucose consumption. Therefore, to compare cellsgrown under these different conditions, the ratio of product formed (mM) per glucose(mM) consumed was used. On this basis, in PO4-buffered DM and in PO4-buffered MM,cells exhibited similar substrate-product profiles (Table 1). It is worth noting that as the

FIG 6 (Continued)

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 13

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

culture conditions became more nutritionally demanding (DM ¡ MM), ethanol pro-duction increased slightly while formate levels decreased.

Relative to ethanol formation by C. scindens ATCC 35704, pyruvate is converted toacetyl-CoA by the enzyme pyruvate:ferredoxin oxidoreductase (PFOR), which also yieldsCO2 and reduced ferredoxin (FDred). The reoxidation of FDred is then achieved by oneof two ways, reduction of NAD� via ferredoxin:NAD� oxidoreductase or conversion toH2 by hydrogenase. H2 was detected in the headspace gas of cultures (Table 1) and ispredicted to be generated by a highly expressed ferredoxin-dependent group B [FeFe]hydrogenase (HDCHBGLK_03096; 1,013.1 � 308.2 TPM) (40). Another gene encodingpredicted ferredoxin-dependent H2-evolving [FeFe] hydrogenase (group A) was ex-

FIG 6 (Continued)

TABLE 1 Substrate-end product profiles of C. scindens ATCC 35704 grown in PO4-buffereddefined medium and PO4-buffered minimal mediuma

Substrate consumed orproduct formedc

PO4-buffered definedmedium (mM)b

PO4-buffered minimalmedium (mM)b

Substrate consumedGlucose 19.4 � 2.7 10.9 � 5.5

Product formedEthanol 24.6 � 6.6 (1.27) 22.0 � 8.4 (2.02)Succinate 0.8 � 0.0 (0.04) 0.3 � 0.0 (0.03)Lactate 0.4 � 0.0 (0.02) 0.4 � 0.0 (0.04)Formate 6.3 � 0.4 (0.32) 0.5 � 0.5 (0.05)Acetate 10.7 � 0.9 (0.55) 4.2 � 1.5 (0.39)Isobutyrate 0.1 � 0.2 (0.01) 0.8 � 0.6 (0.07)Butyrate ND 0.1 � 0.1 (0.01)Isovalerate 0.4 � 0.1 (0.02) 0.3 � 0.1 (0.03)Valerate ND NDH2 16.9 � 2.0 (0.87) 7.2 � 1.7 (0.37)

aPO4-buffered defined medium and PO4-buffered minimal medium were DM and MM, respectively, with thefollowing modifications: NaHCO3 was replaced by a phosphate buffer (KH2PO4 [2.6 g/liter] and K2HPO4 [5.4g/liter]); CO2 was replaced by dinitrogen (N2) or argon (Ar) gas as the initial gas phase; and glucose,routinely added prior to autoclaving, was replaced with glucose (22 mM) added to tubes of sterile mediafrom an anoxic, filter-sterilized glucose stock solution.

bValues represent the differences between values at time zero (at the time of inoculation) and values at thetime when maximum growth was achieved and are the means of triplicate tubes � standard deviations.Parenthetical values are the amount of product formed (mM) per amount of glucose (mM) consumed. ND,not detected.

cCarbon recovery was 72% and 93% in PO4-buffered defined medium and PO4-buffered minimal medium,respectively.

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 14

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

pressed at a much lower level (HDCHBGLK_00634; 72.6 � 11.6 TPM). RNA-Seq analysisdetermined that a gene encoding predicted bifunctional aldehyde-alcohol dehydroge-nase (adhE; HDCHBGLK_00499) was highly expressed (2,246.8 � 434.5 TPM). Regardingacetate formation, genes involved in the conversion of acetyl-CoA to acetyl phosphate(phosphotransacetylase; HDCHBGLK_02911; 842.2 � 140.9 TPM) and to acetate (acetatekinase; HDCHBGLK_02912; 221.0 � 83.7 TPM) were expressed. Formate was also gen-erated, and genes encoding putative formate acetyltransferase (HDCHBGLK_02831;496.8 � 72.6 TPM) and pyruvate formate-lyase-activating enzyme (HDCHBGLK_02832;803.1 � 308.4 TPM) are expressed in DM. Although glucose is the only carbohydratepresent, the presence of bile acids led to a noticeable increase in gene expression levelsof predicted transporters of lactose, trehalose, and arabinose, indicating that with C.scindens ATCC 35704 sugar preference may be altered in the presence of CA and DCA(Fig. 5A).

Genomic potential for amino acid fermentation. As a final note, Clostridium spp.are also recognized as unique for their ability to generate ATP through substrate-levelphosphorylation via oxidation of one amino acid followed by the reduction of anotherin what is known as Stickland fermentation (41, 42). A Western diet high in animalprotein and fat increases both the concentration of bile acids and the concentration ofamino acids available for fermentation (43). Amino acids as electron donors may becoupled to bile acids as electron acceptors (e.g., 3-dehydro-4,6-DCA) via NADH. Giventhe expression levels in PO4-buffered DM, the genomic potential is present for aminoacid fermentations in C. scindens ATCC 35704, including genes for proline reductaseand glycine reductase. The glycine cleavage pathway is also present, suggesting thatL-glycine oxidation may be coupled to L-proline or L-glycine reduction (Table S3). Whilethese results indicate that C. scindens ATCC 35704 may utilize Stickland fermentation,further work will be needed both in vitro and in vivo to determine the contribution ofamino acids as electron donors and acceptors.

DISCUSSION

Several decades of research have focused on the molecular biology and enzy-mology of the bai regulon in Clostridium scindens from a reductionist point of view(3, 4); however, global responses to bile acids by strains of this species have notbeen determined. Secondary bile acid metabolites from C. scindens ATCC 35704were recently shown to regulate hepatic natural killer (NKT) cell accumulation,affecting growth of liver tumors (44). Another study showed that DCA promotes thesenescence-associated secretory phenotype in obesity-associated hepatocellularcarcinoma, with bacterial operational taxonomic unit 154 (OTU-154) closely relatedto C. hylemonae or C. scindens in Clostridium cluster XIVa representing 0.5% of thetotal fecal microbiome (45). A recent analysis of fecal microbiota of patients with(n � 233) and without (n � 547) adenomas identified an increased shift in second-ary bile acid biosynthesis in adenoma patients (46). This is consistent with decadesof literature implicating secondary bile acids in colorectal cancer (47). Gallstonepatients have been shown to harbor �42-fold-higher levels of bile acid 7�-dehydroxylating bacteria than control patients (48). Antibiotic treatment reducedcholesterol saturation in gallstone patients by targeting 7�-dehydroxylating bac-teria and formation of DCA (49). Understanding how C. scindens responds to CA andDCA may indicate potential intervention strategies to reduce toxic and cancer-promoting secondary bile acids such as DCA (4, 43, 50).

On the other hand, formation of DCA may be important in precluding the growthof pathogens such as Clostridium difficile and may indicate acute uses of bile acid7�-dehydroxylating bacteria as probiotics (51–54). Indeed, recent work by Kang etal. (2018) demonstrated that the interaction between C. scindens and C. difficileinvolves not only secondary bile acids but also tryptophan-derived antibiotics (54).Surprisingly, very little is still known about the biology of the few gut bacteriacapable of generating secondary bile acids. The completion of the genome of C.scindens ATCC 35704 and transcriptomic analysis in defined and minimal media are

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 15

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

expected to hasten identification of genes involved in formation of peptide anti-biotics and further integrate bile acid metabolism with fermentative pathways andadaptations to bile salt stress.

An important contribution of the present work is both the determination of thenutritional requirements for C. scindens ATCC 35704 and the integration of genomicand transcriptomic data with the physiology of this bacterium. In the present study,we report genome-wide transcriptomic analysis of bile acid induction by a bile acid7�-dehydroxylating bacterium. Our data indicate that complex and bile acid-specific transcriptional responses are determined by the presence or absence of the7�-hydroxyl group. Addition of CA to the growth medium resulted in a several-logfold change in the genes of the bai regulon, which were among the most highlyupregulated genes. DCA, which differs from CA only in the absence of the 7�-hydroxyl group, did not induce the bai regulon. This is consistent with previousreports of bile acid 7�-dehydroxylating activity and induction of the baiB gene (55).

The first and last oxidative steps in the bile acid 7�-dehydroxylation pathway involveoxidation/reduction of the C-3 oxygen in an NAD(H)-dependent reaction (3, 4, 56).Previous work before genome sequences were available suggested that there werethree copies of the baiA gene in C. scindens: the baiA2 is encoded within the baiB-CDEAFGHI operon and there are two additional mono-cistronic copies, baiA1 and baiA3(56). Completion of the genome of C. scindens ATCC 35705 reveals that there are onlytwo copies of the baiA gene encoding bile acid 3�-HSDH, the baiA2 gene and what isnow referred to as the baiA1, distant from the bai operon, with its own conserved baipromoter region (56). Kinetic analysis suggests that BaiA2 and BaiA1 may function inboth the oxidative and reductive arms of the bile acid 7�-dehydroxylating pathway,converting CA�CoA to 3-dehydro-CA�CoA and 3-dehydro-DCA�CoA to DCA�CoA,respectively (14). Interestingly, our transcriptomic analysis indicated that while thebaiA2 gene, coexpressed on the polycistronic bai operon, was downregulated by DCA,baiA1 was significantly increased (Fig. 4 and 5). This may suggest that the baiA1 geneproduct is involved in the final reductive step in the pathway; however, it is still unclearif there are additional genes encoding enzymes capable of converting 3-dehydro-DCA(�SCoA) to DCA(�SCoA). Development of a genetic system for C. scindens andtargeting of the baiA genes with addition of 3-dehydro-CA to whole cells will allow thisquestion to be properly addressed.

Intriguingly, genes encoding putative TspO and urocanate reductase were signifi-cantly upregulated by CA but not by DCA (Fig. 6C). While their roles in bile acidmetabolism are unclear, it is interesting that TspO homologs are membrane-spanningproteins which are involved in enzymatic metabolism or transport of steroids andbenzodiazopene drugs (27). We recently reported a gene encoding a flavin-dependentbile acid 5�-4,6-reductase (baiN) involved in converting the stable 3-dehydro-4,6-DCA/LCA intermediate (following 7�/�-dehydroxylation) to 3-oxo-DCA (5�-A/B ring orien-tation) (24). However, previous reports established the CA-inducible expression of a5�-4,6-reductase which results in formation of allo-DCA (5�-A/B ring orientation) (28,55). Future research will be required to elucidate the function of the proteins encodedby these newly identified CA-inducible genes.

The bile acid 7�-dehydroxylation process is a series of oxidation and reductionreactions which require coenzymes and cofactors (3). Transcriptomic analysis of CA-induced whole cells of C. scindens suggests an intimate linkage between the growth-limiting nutritional requirements and bile acid 7�-dehydroxylation process in C. scin-dens ATCC 35704 (Fig. 7). Indeed, tryptophan is a precursor for cofactor NAD� requiredin both oxidative and reductive steps, pantothenate is a core precursor for cofactorcoenzyme A (CoA) added by the BaiB and transferred by BaiF/BaiK, and pyridoxalphosphate is a vital cofactor in amino acid and carbohydrate metabolism which linkselectron donor and bile acid acceptor. Equally important is the identification ofnumerous CA- and DCA-induced transmembrane efflux pumps, some of which may beinvolved in transporting DCA out of the cell (see Table S2 in the supplemental material).The current study thus represents an important advance in understanding the biology

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 16

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

of C. scindens ATCC 35704 and the response of this important bile acid 7�-dehydroxylating bacterium to bile acids.

Conclusion. The current study defined the nutritional requirements of C. scindensATCC 35704 and integrated genomic and transcriptomic data sets that support thefindings that tryptophan, pyridoxal phosphate, pantothenic acid, and riboflavin arerequired nutrients. The bai regulon was highly expressed in the presence of CA but notDCA, consistent with previous studies on bile acid induction. Novel candidates for bileacid-metabolizing enzymes and efflux pumps were identified. Transcriptome analysissuggests that C. scindens ATCC 35704 upregulates genes involved in stress, includingcell wall metabolism, DNA repair, expression of chaperones, and multidrug effluxpumps. The current work will facilitate future understanding of the linkage betweensugar or amino acid fermentation, nutritional requirements, and bile acid metabolismby C. scindens, which will be important in understanding the role of diet and microbialmetabolic interactions in health and disease.

MATERIALS AND METHODSChemicals. CA and DCA were purchased from Sigma-Aldrich (St. Louis, MO, USA). Carbohydrates

were purchased from Sigma-Aldrich, MP Biomedicals (Solon, OH, USA), Acros Organics (Geel, Belgium),and Fisher Scientific (Pittsburgh, PA, USA). All other chemicals were of the highest possible purity andwere purchased from Fisher Scientific.

Bacterial strain and growth conditions. Clostridium scindens ATCC 35704, obtained from theAmerican Type Culture Collection (ATCC; Manassas, VA, USA), was maintained in butyl rubber-stoppered,aluminum crimp-sealed culture tubes (18 by 150 mm; series 2048 [Bellco Glass, Inc., Vineland, NJ, USA];�27.2-ml stoppered volume at 1 atm [101.29 kPa]) containing anaerobic brain heart infusion (BHI) broth(15).

Unless noted otherwise, the defined medium (DM) contained the following (in milligrams per liter):glucose, 4,500; NaCl, 500; (NH4)2SO4, 500; KCl, 250; KH2PO4, 250; MgSO4·7H2O, 25; sodium nitrilotriacetate,3; MnSO4·H2O, 1; FeSO4·7H2O, 0.2; Co(NO3)2·6H2O, 0.2; ZnCl2, 0.2; NiCl2·6H2O, 0.1; H2SeO3, 0.1;CuSO4·5H2O, 0.02; AIK(SO4)2·12H2O, 0.02; H3BO3, 0.02; Na2MoO4·2H2O, 0.02; Na2WO4·2H2O, 0.02; resazurin,1; vitamins (biotin, 0.2; folic acid, 0.2; pyridoxal HCl, 0.2; lipoic acid, 0.5; nicotinic acid, 0.5; D-pantothenicacid, 0.5; p-aminobenzoic acid, 0.5; riboflavin, 0.5; thiamine, 0.5; cyanocobalamin, 0.5); and amino acids(L-alanine, L-arginine·HCl, L-asparagine·H2O, L-aspartic acid, L-cystine, L-glutamic acid, L-glutamine,L-glycine, L-histidine, L-isoleucine, L-leucine, L-lysine, L-phenylalanine, L-proline, L-methionine, L-serine,

FIG 7 A proposed model for the coupling of glucose metabolism and bile acid 7�-dehydroxylation by Clostridiumscindens ATCC 35704. Embden-Meyerhof-Parnas (EMP) pathway.

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 17

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

L-threonine, L-tryptophan, L-tyrosine, and L-valine; 40 mg each). The medium pH was adjusted to 7, andNaHCO3 (7.5 g/liter) was added. DM was prepared anaerobically by boiling the medium under 100% CO2

and, after cooling, by adding Na2S·9H2O (0.5 g/liter) and dispensing the medium (10 ml) under 100% CO2

into culture tubes. Tubes were then crimp sealed and autoclaved. UM was DM with 0.1% yeast extractadded. The initial pH (before inoculation) of DM and UM approximated 6.8.

Unless noted otherwise, the minimal medium (MM) contained the following (in milligrams per liter):glucose, 4,500; NaCl, 500; (NH4)2SO4, 500; KCl, 250; KH2PO4, 250; MgSO4·7H2O, 25; sodium nitrilotriace-tate, 3; MnSO4·H2O, 1; FeSO4·7H2O, 0.2; Co(NO3)2·6H2O, 0.2; ZnCl2, 0.2; NiCl2·6H2O, 0.1; H2SeO3, 0.1;CuSO4·5H2O, 0.02; AIK(SO4)2·12H2O, 0.02; H3BO3, 0.02; Na2MoO4·2H2O, 0.02; Na2WO4·2H2O, 0.02; resazurin,1; vitamins (pyridoxal HCl, 0.2; D-pantothenic acid, 0.5; riboflavin, 0.5); and an amino acid (L-tryptophan,40 mg). The medium pH was adjusted to 7, and NaHCO3 (7.5 g/liter) was added. MM was preparedanaerobically by boiling the medium under 100% CO2 and, after cooling, by adding Na2S·9H2O (0.5 g/liter) and dispensing the medium (10 ml) under 100% CO2 into culture tubes. Tubes were then crimpsealed and autoclaved. The initial pH (before inoculation) of MM approximated 6.8.

PO4-buffered DM and PO4-buffered MM were DM and MM, respectively, with the following modifi-cations for both: NaHCO3 was replaced by a phosphate buffer (KH2PO4 [2.6 g/liter] and K2HPO4 [5.4g/liter]), CO2 was replaced by dinitrogen (N2) or argon (Ar) gas as the initial gas phase, and glucose,routinely added prior to autoclaving, was replaced with glucose added to tubes of sterile medium froman anoxic, filter-sterilized glucose stock solution. It should be noted that PO4-buffered medium was usedto facilitate substrate and end product determinations as well as transcriptomic analyses.

In all experiments, growth was initiated by injecting 0.5 ml of inoculum (late log-/early-stationary-phase cultures) into tubes of sterile medium with sterile 23-gauge needles and 1-ml syringes. Allinoculated tubes were incubated vertically without shaking and in the dark at 37°C; growth wasmeasured over time as the optical density (OD) at 600 nm using a spectrophotometer (SpectronicInstruments, Inc., Rochester, NY, USA); the optical path width (inner diameter of culture tubes) was1.6 cm. Uninoculated culture media served as references. A culture medium was considered growthsupportive if OD values of cultures in the medium after three sequential transfers were �0.05. All ODvalues reported are the means of duplicate cultures or experiments.

Nutritional requirements. To determine amino acid and vitamin requirements for C. scindens ATCC35704, the leave-one-amino-acid-group-out and leave-one-vitamin-out approaches were used, respec-tively. First, amino acids were divided into six groups based on known interconversions: (i) glutamategroup (Glu grp; glutamine, glutamic acid, proline, and arginine), (ii) serine (Ser) group (serine, glycine, andcystine), (iii) aspartate (Asp) group (aspartate, asparagine, methionine, lysine, threonine, and isoleucine),(iv) pyruvate (Pyu) group (alanine, valine, and leucine), (v) aromatic (Aro) group (tryptophan, tyrosine, andphenylalanine), and (vi) histidine (His) group (histidine). Six amino acid group-deficient versions of DM(i.e., each version of DM had one of the six amino acid groups omitted) were inoculated with active DMcultures. Amino acid group-deficient versions of DM that supported growth indicated which amino acidswere not required and were eliminated from further testing. Next, each amino acid from a growth-supportive group was tested individually in DM, which contained the remaining amino acids from theother growth-supportive groups; growth under these conditions indicated which amino acid(s) wasrequired.

For vitamin requirements, 10 vitamin-deficient versions of DM (i.e., each version of DM had 1 of the10 vitamins omitted) were inoculated with active DM cultures; the absence of growth under theseconditions indicated which vitamin(s) was required.

Carbohydrate screening. To determine which carbohydrates were growth supportive, glucose inDM was replaced with individual (different) carbohydrates. Stock solutions (10%) of monosaccharides(D-glucose, D-fructose, D-mannose, D-galactose, L-rhamnose, L-sorbose, L-arabinose, D-ribose, and D-xylose),sugar alcohols (D-lactitol, myo-inositol, D-mannitol, D-sorbitol, D-xylitol, erythritol, and glycerol), disaccha-rides (cellobiose, lactose, lactulose, maltose, melibiose, sucrose, and trehalose), trisaccharides (melezitoseand raffinose), polysaccharides (polydextrose and starch), and synthetic sweeteners (saccharin andsucralose) were prepared in tubes using distilled, reverse osmosis (DRO) water, stoppered and crimpsealed, degassed by sparging and flushing the headspace gas with Ar gas, and sterilized by autoclaving.Due to solubility issues, some substrates were prepared as 3.85% stock solutions: sugar alcohols(D-adonitol and dulcitol), polysaccharides (dextrin, glycogen, inulin, mucin, pectin, and stachyose), andglycosides (amygdalin, esculin, and salicin). Stock solutions (10% or 3.85%) were added (0.2 ml) via sterileneedles and syringes to tubes of DM (7 ml) to achieve initial substrate concentrations (after inoculation)of 0.25% or 0.1%, respectively.

TLC of bile acids. For analysis of bile acids in cultures and in standards prepared in sterile media,thin-layer chromatography (TLC) was used. Samples of stationary-phase cultures were removed asepti-cally, and bile acids were extracted immediately, or samples were stored at �20°C until extracted. For bileacid extraction, a 1-ml sample of a culture or bile acid standard was transferred to duplicate microcen-trifuge tubes (500 �l per tube), and 50 �l of 3 N HCl and 500 �l of ethyl acetate were added to each tube.Tubes were capped, vortexed, and spun for 1 min at 16,000 g. The organic phase (top layer) wasremoved from each tube and transferred to a 20-ml glass scintillation vial. Steps from the addition ofethyl acetate to the collection of the organic phase were repeated for each sample. The collected(pooled) organic phase was dried at room temperature under N2. Methanol (100 �l) was added to eachvial, and methanol-dissolved extracts were spotted (50 �l) onto TLC plates (20- by 20-cm AL Sil G, 250 �mthick, non-UV; Macherey-Nagel, Duren, Germany). Nonradiolabeled bile acids separated by TLC werevisualized as recently described (15). To visualize 24-14C-labeled bile acids, TLC plates containing bile acidintermediates were placed on Kodak Biomax MS film overnight and developed as described previously

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 18

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

(55). Bile acids were identified by comparing Rf values of bile acid standards to those of bile acidsdetected in cultures.

HPLC. The following high-performance liquid chromatography (HPLC) equipment and conditionswere used for the analysis of glucose and organic acids in standards and in PO4-buffered DM andPO4-buffered MM cultures: Beckman Gold System (125 solvent module, model 166 variable wavelengthdetector, Jasco RI-1530 refractive index [RI] detector, Eppendorf CH-30 column heater, model 508autosampler, and 32 Karat software, version 5); Aminex HPX-87H (300-mm long) analytical column(Bio-Rad Laboratories, Inc., Hercules, CA, USA); 0.01 N H2SO4 solvent; solvent flow rate of 0.6 ml/min;column temperature of 55°C; injection size of 20 �l; detection by UV (210 nm) for short-chain organicacids or RI for glucose; and run time of 45 min. Culture fluids were clarified by microcentrifugation andmicrofiltration and stored at �20°C until analyzed. All concentrations were expressed on a millimolarbasis.

GC. The following gas chromatography (GC) equipment and conditions were used for the analysis ofhydrogen (H2) gas analysis in standards and in PO4-buffered DM and PO4-buffered MM cultures: ThermoFisher Trace 1310 gas chromatograph (thermal conductivity detector and Dionex chromeleon, version7.1.2.1478, software); stainless steel column (2 mm by 2 m) containing a molecular sieve (13 by 60/80;Restek, Bellefonte, PA, USA); 100% Ar carrier gas; carrier gas flow rate of 20 ml/min; 150°C injection port;60°C column oven,; 175°C detector; injection size of 100 �l; run time of 2 min. Before chromatographicanalysis, headspace gas volume was measured (in millibars) with a TensioCheck (Tensio-Technik, Geisen-heim, Germany) manometer. Hydrogen (H2) solubility was calculated from a standard solubility table (57),and the amount of gas produced was calculated by considering both gas and liquid phases.

PacBio genome sequencing of Clostridium scindens ATCC 35704. Genomic DNA was isolated aspreviously described (23) with some modifications. Chemical lysis was performed, and pipetting stepswere omitted after gentle phenol and then phenol-chloroform-acetic acid washes. DNA was furtherpurified by a genomic tip 100 column according to the manufacturer’s instructions (Qiagen, Valencia, CA,USA). The genome of C. scindens ATCC 35704 was sequenced by a Pacific Biosciences RSII instrumentafter high-molecular-weight genomic DNA (� 40 kb) was submitted to the Great Lakes Genomics Center,University of Wisconsin at Milwaukee. A standard Pacific Biosciences large-insert library preparation wasperformed. DNA was fragmented to approximately 20 kb using Covaris G tubes. Fragmented DNA wasenzymatically repaired and ligated to a PacBio adapter to form the SMRTbell template. Templates (�10kb) were size selected, depending on library yield and size, using BluePippin (Sage Science, Beverly, MA,USA). Templates were annealed to sequencing primer, bound to polymerase (P6), and then bound toPacBio Mag-Beads and SMRTcell sequenced using C4 chemistry. The C. scindens ATCC 35704 genome wasassembled by Canu, version 1.5, with an estimated genome size of 3.6 Mb.

Pathway analysis. The genome was annotated by both Prokka and the SEED database. For each, twomapping methods were used. The first was mapping to the KEGG database, and the other method usedthe Python-based tool DuctApe.

Bile acid-induction and RNA isolation from Clostridium scindens ATCC 35704. Clostridiumscindens ATCC 35704 was grown to mid-log phase in PO4-buffered DM (control) or in PO4-buffered DMsupplemented with two additions (i.e., one at the time of inoculation and one during mid-log phase) of50 �M cholate or deoxycholate (bile acids were added from methanolic stock solutions). Cells werequenched with RNAprotect (Qiagen) and stored at �80°C until further processing. Total RNA wasrecovered from cells following disruption by bead beating in the presence of acid phenol as previouslyreported (58). In brief, stored cells were dissolved in lysis buffer (200 mM NaCl, 20 mM EDTA, anddiethylpyrocarbonate-treated water) and transferred to 2-ml screw-cap bead-beating tubes (Sarstedt AG& Co. KG, Nümbrecht, Germany). Cells were washed with the same buffer and resuspended in 500 �l oflysis buffer. To each tube, 200 �l of zirconium beads, 210 �l of 20% SDS (Ambion, Thermo FisherScientific, Waltham, USA), and 1 ml of 5:1 acid phenol were added. Cells were disrupted on a Mini-BeadBeater (Biospec Products, Inc., Bartlesville, OK, USA) at maximum speed twice for 1 min, with tubeskept on ice in between treatments. The aqueous and phenol phases were separated by centrifugationat 5,000 rpm for 1 min, and the aqueous phase was transferred to a new tube and washed once with 1 mlof 5:1 acid-phenol. Nucleic acids in the aqueous phase were precipitated at �80°C for 60 min by additionof 1/10 volume of 5 M ammonium acetate (Ambion), 1 �l of Glycoblue (Ambion), 1 �l of RNase inhibitor(New England Biolabs [NEB], Ipswich, MA, USA), and 1 volume of ice-cold isopropanol, followed bycentrifugation at 13,600 g for 30 min. Precipitated nucleic acids were treated with RNase-free TurboDNase (Ambion) according to the manufacturer’s instructions to remove contaminated genomic DNA.Total RNA was purified using a MEGAclear kit (Ambion) according to the manufacturer’s instructions. RNApurity and integrity were checked spectrophotometrically by the A260/A280 ratio and by separating 16Sand 23S rRNA on a 1% agarose gel, respectively.

RNA-Seq analysis. Stranded RNA-Seq libraries were constructed and sequenced at the DNA Serviceslaboratory of the Roy J. Carver Biotechnology Center at the University of Illinois at Urbana-Champaignusing aTruSeq Stranded RNA sample preparation kit (Illumina, San Diego, CA, USA). Briefly, total RNA wasquantitated by Qubit (Life Technologies, Grand Island, NY, USA) and checked for integrity on a 2% eGel(Life Technologies). Starting with 1 �g of RNA, rRNA was removed from the total RNA using a Ribo-ZeroMagnetic Bacteria kit (Illumina, CA). First-strand synthesis was performed with a random hexamer andSuperScript II (Life Technologies). Double-stranded DNA was blunt ended, 3= end A-tailed, and ligated tounique dual-indexed adaptors. The adaptor-ligated double-stranded cDNA was amplified by PCR for 8cycles with Kapa HiFi polymerase (Kapa Biosystems, Wilmington, MA, USA). The final libraries werequantitated on Qubit, and the average size was determined on an AATI Fragment Analyzer (AdvancedAnalytics, Ames, IA, USA). Libraries were pooled evenly, and the pool was cleaned one additional time

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 19

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

using a 50/50 ratio with AxyPrep Mag PCR Cleanup beads (Axygen, Inc., Union City, CA, USA) to ensureremoval of primer and adaptor dimers and then evaluated on an AATI Fragment Analyzer. The final poolwas diluted to a 5 nM concentration and further quantitated by quantitative PCR (qPCR) on a Bio-Rad CFXConnect Real-Time System (Bio-Rad).

The final pool containing 6 libraries was loaded onto one lane of an eight-lane flow cell each forcluster formation on the cBOT and then sequenced on an Illumina HiSeq 4000 with SBS (version 1)sequencing reagents from one end of the molecules, for a total read length of 100 nucleotides (nt). Therun generates bcl files which are converted into adaptor-trimmed demultiplexed fastq files usingbcl2fastq, version 2.20, conversion software (Illumina).

Bioinformatics. Quality control of raw RNA-Seq reads was performed by using the FastQC, version0.11.5. Reads that had quality scores of �32 were used for further analysis. Raw reads were aligned withrRNA sequences prepared from the genome of C. scindens ATCC 35704 using bowtie2 (version 2.3.3.1).Unaligned files were saved and realigned with the genome of C. scindens ATCC 35704 using the sametool. Output sam files were converted to the bam format using Samtools (version 0.1.6) and were namesorted prior to input into HTSeq (version 0.9.1). HTSeq counting was performed in union mode againsta gene feature format (GFF) annotation of C. scindens ATCC 35704. Reads were counted against codingDNA sequences (CDS) of the organism. Differential gene expression analysis was performed using edgeR(59, 60) and limma (61) R packages. A minimum P value of �0.05 was accepted as indicating differentiallyexpressed genes. Gene expression (Bonferroni corrected) is reported as log2FC with a cutoff of 1.58(3-fold change) and a false discovery rate (FDR) cutoff of �0.05. Genes were binned according to knownfunctionality, and totals were generated for upregulated and downregulated genes. Category analysiswas performed using the eggNOG eggnog-mapper (web interface with bacteria as taxonomic scope andDIAMOND as search program) (62).

Quantitative reverse transcription-PCR (qRT-PCR). Differential expression of selected genes undercontrol, CA-induced, and DCA-induced conditions was observed using a Light Cycler 480 Real-Time PCRSystem (Roche Diagnostics, Risch-Rotkreuz, Switzerland). Primers for the C. scindens genes baiE (forward,5=-CTGGAGACCACTCTGTCACC-3=; reverse, 5=-ATACCATCTGCCCGTAGCC-3=) and urocanate reductase(forward, 5=-AGGGAGATGGAATCTGGGCT-3=; reverse, 5=-TGAAGCAGCATGGTCTGGAG-3=) were amplifiedin triplicate from four replicates and compared to the housekeeping gene recA (forward, 5=-TGGAAAAGGCACGGTTATGAAAC-3=; reverse 5=-CAACCACACGGATGTTCTTACG-3=). Relative abundance was quan-tified using the ΔΔCT (where CT is threshold cycle) method, and average log2FC was determined. A t testwas performed using triplicate averages to observe significant differences in expression levels betweenCA- and DCA-induced and control cultures.

Accession number(s) RNA-Seq and PacBio genome data sets were submitted to NCBI underBioProject number PRJNA508260.

SUPPLEMENTAL MATERIALSupplemental material for this article may be found at https://doi.org/10.1128/AEM

.00052-19.SUPPLEMENTAL FILE 1, PDF file, 0.3 MB.SUPPLEMENTAL FILE 2, XLSX file, 0.4 MB.SUPPLEMENTAL FILE 3, XLSX file, 0.1 MB.SUPPLEMENTAL FILE 4, XLSX file, 0.1 MB.

ACKNOWLEDGMENTSWe gratefully acknowledge the financial support provided to J.M.R. for new

faculty start-up through the Department of Animal Sciences at the University ofIllinois at Urbana-Champaign (grant Hatch ILLU-538-916) as well as Illinois CampusResearch Board RB18068. This work was also supported by grants (J.M.R) 1RO1CA204808-01, NIH R01CA179243, and the Young Investigators Grant for ProbioticResearch (Danone, Yakult). L.L. is supported by a Graduate Research Fellowshipthrough the National Science Foundation. P.G.W. is supported by an UIC CancerEducation and Career Development Training Program award by the Institute forHealth Research and Policy at the University of Illinois at Chicago with funding bythe National Cancer Institute (grant no. T32CA057699). We also acknowledgeresearch grants from the Department of Biological Sciences (A.I., J.W., A.A., and E.S.),College of Sciences (J.W., A.A., E.S., and O.P.), Graduate School (R.S. and O.P.), andthe Sandra & Jack Pines Honors College (A.A.) at Eastern Illinois University.

We thank Phillip B. Hylemon for careful review of the manuscript. This project usedthe UW-Milwaukee Great Lakes Genomics Center Pacific Biosciences RSII DNA sequenc-ing and bioinformatics services.

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 20

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

REFERENCES1. Vlahcevic ZR, Heuman DM, Hylemon PB. 1996. Physiology and patho-

physiology of enterohepatic circulation of bile acids, p 376 – 417. InZakim D, Boyer T (ed), Hepatology: a textbook of liver disease, 3rd ed, vol1. Saunders, Philadelphia, PA.

2. Hofmann AF. 1984. Chemistry and enterohepatic circulation of bile acids.Hepatology 4:4S–14S. https://doi.org/10.1002/hep.1840040803.

3. Ridlon JM, Kang DJ, Hylemon PB. 2006. Bile salt biotransformations byhuman intestinal bacteria. J Lipid Res 47:241–259. https://doi.org/10.1194/jlr.R500013-JLR200.

4. Ridlon JM, Harris SC, Bhowmik S, Kang D-J, Hylemon PB. 2016. Conse-quences of bile salt metabolism by intestinal bacteria. Gut Microbes7:22–39. https://doi.org/10.1080/19490976.2015.1127483.

5. Ridlon JM, Alves JM, Hylemon PB, Bajaj JS. 2013. Cirrhosis, bile acids andgut microbiota: unraveling a complex relationship. Gut Microbes4:382–387. https://doi.org/10.4161/gmic.25723.

6. Doerner KC, Takamine F, LaVoie CP, Mallonee DH, Hylemon PB. 1997.Assessment of fecal bacteria with bile acid 7�-dehydroxylating activityfor the presence of bai-like genes. Appl Environ Microbiol 63:1185–1188.

7. White BA, Lipsky RL, Fricke RJ, Hylemon PB. 1980. Bile acid inductionspecificity of 7�-dehydroxylase activity in an intestinal Eubacterium species.Steroids 35:103–109. https://doi.org/10.1016/0039-128X(80)90115-4.

8. Kitahara M, Takamine F, Imamura T, Benno Y. 2000. Assignment ofEubacterium sp. VPI 12708 and related strains with high bile acid 7�-dehydroxylating activity to Clostridium scindens and proposal of Clos-tridium hylemonae sp. nov., isolated from human faeces. Int J Syst EvolMicrobiol 5:971–978. https://doi.org/10.1099/00207713-50-3-971.

9. Bokkenheuser VD, Morris GN, Ritchie AE, Holdeman LV, Winter J. 1984.Biosynthesis of androgen from cortisol by a species of Clostridiumrecovered from human fecal flora. J Infect Dis 149:489 – 494. https://doi.org/10.1093/infdis/149.4.489.

10. Winter J, Morris GN, O’Rourke-Locascio S, Bokkenheuser VD, MosbachEH, Cohen BI, Hylemon PB. 1984. Mode of action of steroid desmolaseand reductases synthesized by Clostridium “scindens” (formerly Clostrid-ium strain 19). J Lipid Res 25:1124 –1131.

11. Morris GN, Winter J, Cato EP, Ritchie AE, Bokkenheuser VD. 1985. Clos-tridium scindens sp. nov., a human intestinal bacterium with desmolyticactivity on corticoids. Int J Sys Bacteriol 35:478 – 481. https://doi.org/10.1099/00207713-35-4-478.

12. Bergstrom S, Lindstedt S, Samuelsson B. 1959. Bile acids and steroids.LXXXII. On the mechanism of deoxycholic acid formation in the rabbit.J Biol Chem 234:2022–2025.

13. Bhowmik S, Chiu HP, Jones DH, Chiu HJ, Miller MD, Xu Q, Farr CL, RidlonJM, Wells JE, Elsliger MA, Wilson IA, Hylemon PB, Lesley SA. 2016.Structure and functional characterization of a bile acid 7� dehydrataseBaiE in secondary bile acid synthesis. Proteins Mar 84:316 –331. https://doi.org/10.1002/prot.24971.

14. Bhowmik S, Jones DH, Chiu HP, Park IH, Chiu HJ, Axelrod HL, Farr CL, TienHJ, Agarwalla S, Lesley SA. 2014. Structural and functional characteriza-tion of BaiA, an enzyme involved in secondary bile acid synthesis inhuman gut microbe. Proteins 82:216 –229. https://doi.org/10.1002/prot.24353.

15. Doden H, Sallam LA, Devendran S, Ly L, Doden G, Daniel SL, Alves JMP,Ridlon JM. 2018. Metabolism of oxo-bile acids and characterization ofrecombinant 12�-hydroxysteroid dehydrogenases from bile acid 7�-dehydroxylating human gut bacteria. Appl Environ Microbiol 84:e00235-18.https://doi.org/10.1128/AEM.00235-18.

16. Marion S, Studer N, Desharnais L, Menin L, Escrig S, Meibom A, Hap-felmeier S, Bernier-Latmani R. 27 December 2018. In vitro and in vivocharacterization of Clostridium scindens bile acid transformations. GutMicrobes https://doi.org/10.1080/19490976.2018.1549420.

17. Kamekura M, Oesterhelt D, Wallace R, Anderson P, Kushner DJ. 1988.Lysis of halobacteria in bacto-peptone by bile acids. Appl Environ Mi-crobiol 54:990 –995.

18. Ditscheid B, Keller S, Jahreis G. 2009. Faecal steroid excretion in humans isaffected by calcium supplementation and shows gender-specific differ-ences. Eur J Nutr 48:22–30. https://doi.org/10.1007/s00394-008-0755-2.

19. Hamilton JP, Xie G, Raufman JP, Hogan S, Griffin TL, Packard CA, Chat-field DA, Hagey LR, Steinbach JH, Hofmann AF. 2007. Human cecal bileacids: concentration and spectrum. Am J Physiol Gastrointest LiverPhysiol 293:G256 –G263. https://doi.org/10.1152/ajpgi.00027.2007.

20. Allison MJ, Robinson IM, Baetz AL. 1974. Tryptophan biosynthesis from

idole-3-acetic acid by anaerobic bacteria from the rumen. J Bacteriol117:175–180.

21. Lee JH, Wood TK, Lee J. 2015. Role of indole as an interspecies andinterkingdom signaling molecule. Trends Microbiol 23:707–718. https://doi.org/10.1016/j.tim.2015.08.001.

22. Scherzer R, Gdalevsky GY, Goldgur Y, Cohen-Luria R, Bittner S, Parola AH.2009. New tryptophanase inhibitors: towards prevention of bacterialbiofilm formation. J Enzyme Inhib Med Chem 24:350 –355. https://doi.org/10.1080/14756360802187612.

23. Ridlon JM, McGarr SE, Hylemon PB. 2005. Development of methods forthe detection and quantification of 7�-dehydroxylating clostridia, De-sulfovibrio vulgaris, Methanobrevibacter smithii, and Lactobacillus planta-rum in human feces. Clin Chim Acta 357:55– 64. https://doi.org/10.1016/j.cccn.2005.02.004.

24. Harris SC, Devendran S, Alves JMP, Mythen SM, Hylemon PB, Ridlon JM.2018. Identification of a gene encoding a flavoprotein involved in bileacid metabolism by the human gut bacterium Clostridium scindens ATCC35704. Biochim Biophys Acta Mol Cell Biol Lipids 1863:276 –283. https://doi.org/10.1016/j.bbalip.2017.12.001.

25. Baron SF, Franklund CV, Hylemon PB. 1991. Cloning, sequencing, andexpression of the gene coding for bile acid 7�-hydroxysteroid dehydro-genase from Eubacterium sp. strain VPI 12708. J Bacteriol 173:4558 – 4569. https://doi.org/10.1128/jb.173.15.4558-4569.1991.

26. Ridlon JM, Hylemon PB. 2012. Identification and characterization of twobile acid coenzyme A transferases from Clostridium scindens, a bile acid7�-dehydroxylating intestinal bacterium. J Lipid Res 53:66 –76. https://doi.org/10.1194/jlr.M020313.

27. Guo Y, Kalathur RC, Liu Q, Kloss B, Bruni R, Ginter C, Kloppmann E, RostB, Hendrickson WA. 2015. Protein structure. Structure and activity oftryptophan-rich TSPO proteins. Science 347:551–555. https://doi.org/10.1126/science.aaa1534.

28. Hylemon PB, Melone PD, Franklund CV, Lund E, Björkhem I. 1991.Mechanism of intestinal 7 �-dehydroxylation of cholic acid: evidencethat allo-deoxycholic acid is an inducible side-product. J Lipid Res32:89 –96.

29. Thanassi DG, Cheng LW, Nikaido H. 1997. Active efflux of bile salts byEscherichia coli. J Bacteriol 179:2512–2518. https://doi.org/10.1128/jb.179.8.2512-2518.1997.

30. Luo Y, Helmann JD. 2012. Analysis of the role of Bacillus subtilis �(M) in�-lactam resistance reveals an essential role for c-di-AMP in peptidogly-can homeostasis. Mol Microbiol 83:623– 639. https://doi.org/10.1111/j.1365-2958.2011.07953.x.

31. Thackray PD, Moir A. 2003. SigM, an extracytoplasmic function sigmafactor of Bacillus subtilis, is activated in response to cell wall antibiotics,ethanol, heat, acid, and superoxide stress. J Bacteriol 185:3491–3498.https://doi.org/10.1128/JB.185.12.3491-3498.2003.

32. Urdaneta V, Casadesús J. 2017. Interactions between bacteria and bilesalts in the gastrointestinal and hepatobiliary tracts. Front Med (Laus-anne) 4:163. https://doi.org/10.3389/fmed.2017.00163.

33. Begley M, Gahan CG, Hill C. 2005. The interaction between bacteria andbile. FEMS Microbiol Rev 29:625– 651. https://doi.org/10.1016/j.femsre.2004.09.003.

34. Kurdi P, van Veen HW, Tanaka H, Mierau I, Konings WN, Tannock GW,Tomita F, Yokota A. 2000. Cholic acid is accumulated spontaneously,driven by membrane ΔpH, in many lactobacilli. J Bacteriol 182:6525– 6528. https://doi.org/10.1128/JB.182.22.6525-6528.2000.

35. Watanabe M, Fukiya S, Yokota A. 2017. Comprehensive evaluation of thebactericidal activities of free bile acids in the large intestine of humans androdents. J Lipid Res 58:1143–1152. https://doi.org/10.1194/jlr.M075143.

36. Yasugi M, Okuzaki D, Kuwana R, Takamatsu H, Fujita M, Sarker MR,Miyake M. 2016. Transcriptional profile during deoxycholate-inducedsporulation in a Clostridium perfringens isolate causing foodborneillness. Appl Environ Microbiol 82:2929 –2942. https://doi.org/10.1128/AEM.00252-16.

37. Bernstein C, Bernstein H, Payne CM, Beard SE, Schneider J. 1999. Bile saltactivation of stress response promoters in Escherichia coli. Curr Microbiol39:68 –72. https://doi.org/10.1007/s002849900420.

38. Breton YL, Mazé A, Hartke A, Lemarinier S, Auffray Y, Rincé A. 2002.Isolation and characterization of bile salts-sensitive mutants of Entero-coccus faecalis. Curr Microbiol 45:434 – 439. https://doi.org/10.1007/s00284-002-3714-3.

Genomic and Transcriptomic Analyses of Clostridium scindens Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 21

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

39. White BA, Cacciapuoti AF, Fricke RJ, Whitehead TR, Mosbach EH, Hyl-emon PB. 1981. Cofactor requirements for 7 �-dehydroxylation of cholicand chenodeoxycholic acid in cell extracts of the intestinal anaerobicbacterium, Eubacterium species V.P.I. 12708. J Lipid Res 22:891– 898.

40. Wolf PG, Biswas A, Morales SE, Greening C, Gaskins HR. 2016. H2 metab-olism is widespread and diverse among human colonic microbes. GutMicrobes 7:235–245. https://doi.org/10.1080/19490976.2016.1182288.

41. Mead GC. 1971. The amino acid-fermenting clostridia. J Gen Microbiol67:47–56. https://doi.org/10.1099/00221287-67-1-47.

42. Fonknechten N, Chaussonnerie S, Tricot S, Lajus A, Andreesen JR,Perchat N, Pelletier E, Gouyvenoux M, Barbe V, Salanoubat M, Le PaslierD, Weissenbach J, Cohen GN, Kreimeyer A. 2010. Clostridium sticklandii,a specialist in amino acid degradation: revisiting its metabolism throughits genome sequence. BMC Genomics 11:555. https://doi.org/10.1186/1471-2164-11-555.

43. Ridlon JM, Wolf PG, Gaskins HR. 2016. Taurocholic acid metabolism bygut microbes and colon cancer. Gut Microbes 22:1–15. https://doi.org/10.1080/19490976.2016.1150414.

44. Ma C, Han M, Heinrich B, Fu Q, Zhang Q, Sandhu M, Agdashian D, TerabeM, Berzofsky JA, Fako V, Ritz T, Longerich T, Theriot CM, McCulloch JA,Roy S, Yuan W, Thovarai V, Sen SK, Ruchirawat M, Korangy F, Wang XW,Trinchieri G, Greten TF, 2018, Gut microbiome-mediated bile acid me-tabolism regulates liver cancer via NKT cells. Science 360:eaan5931.https://doi.org/10.1126/science.aan5931.

45. Yoshimoto S, Loo TM, Atarashi K, Kanda H, Sato S, Oyadomari S, IwakuraY, Oshima K, Morita H, Hattori M, Hattori M, Honda K, Ishikawa Y, Hara E,Ohtani N. 2013. Obesity-induced gut microbial metabolite promotesliver cancer through senescence secretome. Nature 499:97–101. https://doi.org/10.1038/nature12347.

46. Hale VL, Chen J, Johnson S, Harrington SC, Yab TC, Smyrk TC, Nelson H,Boardman LA, Druliner BR, Levin TR, Rex DK, Ahnen DJ, Lance P, AhlquistDA, Chia N. 2017. Shifts in the fecal microbiota associated with adenom-atous polyps. Cancer Epidemiol Biomarkers Prev 26:85–94. https://doi.org/10.1158/1055-9965.EPI-16-0337.

47. Bernstein H, Bernstein C, Payne CM, Dvorakova K, Garewal H. 2005. Bileacids as carcinogens in human gastrointestinal cancers. Mutat Res 589:47– 65. https://doi.org/10.1016/j.mrrev.2004.08.001.

48. Wells JE, Berr F, Thomas LA, Dowling RH, Hylemon PB. 2000. Isolationand characterization of cholic acid 7�-dehydroxylating fecal bacteriafrom cholesterol gallstone patients. J Hepatol 32:4 –10. https://doi.org/10.1016/S0168-8278(00)80183-X.

49. Berr F, Kullak-Ublick GA, Paumgartner G, Münzing W, Hylemon PB. 1996.7�-Dehydroxylating bacteria enhance deoxycholic acid input and cho-lesterol saturation of bile in patients with gallstones. Gastroenterology111:1611–1620. https://doi.org/10.1016/S0016-5085(96)70024-0.

50. O’Keefe SJ. 2016. Diet, microorganisms and their metabolites, and coloncancer. Nat Rev Gastroenterol Hepatol 13:691–706. https://doi.org/10.1038/nrgastro.2016.165.

51. Buffie CG, Bucci V, Stein RR, McKenney PT, Ling L, Gobourne A, No D,

Liu H, Kinnebrew M, Viale A, Littmann E, van den Brink MR, Jenq RR,Taur Y, Sander C, Cross JR, Toussaint NC, Xavier JB, Pamer EG. 2015.Precision microbiome reconstitution restores bile acid mediated re-sistance to Clostridium difficile. Nature 517:205–208. https://doi.org/10.1038/nature13828.

52. Theriot CM, Bowman AA, Young VB. 2016. Antibiotic-induced alterationsof the gut microbiota alter secondary bile acid production and allow forClostridium difficile spore germination and outgrowth in the large intes-tine. mSphere 1:e00045-15. https://doi.org/10.1128/mSphere.00045-15.

53. Sorg JA, Sonenshein AL. 2008. Bile salts and glycine as cogerminants forClostridium difficile spores. J Bacteriol 190:2505–2512. https://doi.org/10.1128/JB.01765-07.

54. Kang JD, Myers CJ, Harris SC, Kakiyama G, Lee IK, Yun BS, Matsuzaki K,Furukawa M, Min HK, Bajaj JS, Zhou H, Hylemon PB. 2018. Bile acid7�-dehydroxylating gut bacteria secrete antibiotics that inhibit Clostrid-ium difficile: Role of secondary bile acids. Cell Chem Biol 26:27–34.e4.https://doi.org/10.1016/j.chembiol.2018.10.003.

55. Ridlon JM, Kang DJ, Hylemon PB. 2010. Isolation and characterization ofa bile acid inducible 7�-dehydroxylating operon in Clostridium hylem-onae TN271. Anaerobe 16:137–146. https://doi.org/10.1016/j.anaerobe.2009.05.004.

56. Gopal-Srivastava R, Mallonee DH, White WB, Hylemon PB. 1990. Multiplecopies of a bile acid-inducible gene in Eubacterium sp. strain VPI 12708.J Bacteriol 172:4420 – 4426. https://doi.org/10.1128/jb.172.8.4420-4426.1990.

57. Wiesenburg DA, Guinasso NL. 1979. Equilibrium solubilities of methane,carbon monoxide, and hydrogen in water and sea water. J Chem EngData 24:356 –360. https://doi.org/10.1021/je60083a006.

58. Ridlon JM, Ikegawa S, Alves JMP, Zhou B, Kobayashi A, Iida T, MitamuraK, Tanabe G, Serrano M, De Guzman A, Cooper P, Buck GA, Hylemon PB.2013. Clostridium scindens: a human gut microbe with a high potential toconvert glucocorticoids into androgens. J Lipid Res 54:2437–2449.https://doi.org/10.1194/jlr.M038869.

59. Robinson MD, McCarthy DJ, Smyth GK. 2010. edgeR: a bioconductor pack-age for differential expression analysis of digital gene expression data.Bioinformatics 26:139–140. https://doi.org/10.1093/bioinformatics/btp616.

60. McCarthy JD, Chen Y, Smyth KG. 2012. Differential expression analysis ofmultifactor RNA-Seq experiments with respect to biological variation.Nucleic Acids Res 40:4288 – 4297. https://doi.org/10.1093/nar/gks042.

61. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. 2015.limma powers differential expression analyses for RNA-sequencing andmicroarray studies. Nucleic Acids Res 43:e4. https://doi.org/10.1093/nar/gkv007.

62. Huerta-Cepas J, Szklarczyk D, Heller D, Hernández-Plaza A, Forslund SK,Cook H, Mende DR, Letunic I, Rattei T, Jensen LJ, von Mering C, Bork P.2018. eggNOG 5.0: a hierarchical, functionally and phylogenetically an-notated orthology resource based on 5090 organisms and 2502 viruses.Nucleic Acids Res 47:D309 –D314. https://doi.org/10.1093/nar/gky1085.

Devendran et al. Applied and Environmental Microbiology

April 2019 Volume 85 Issue 7 e00052-19 aem.asm.org 22

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from