Date post: | 31-Mar-2018 |
Category: |
Documents |
Upload: | nguyenkiet |
View: | 214 times |
Download: | 0 times |
www.sciencemag.org/content/351/6275/aad3311/suppl/DC1
Supplementary Materials for
Gut bacteria that prevent growth impairments transmitted by microbiota from malnourished children
Laura V. Blanton, Mark R. Charbonneau, Tarek Salih, Michael J. Barratt, Siddarth Venkatesh, Olga Ilkaveya, Sathish Subramanian, Mark J. Manary, Indi Trehan, Josh M.
Jorgensen, Yue-mei Fan, Bernard Henrissat, Semen A. Leyn, Dmitry A. Rodionov, Andrei L. Osterman, Kenneth M. Maleta, Christopher B. Newgard, Per Ashorn, Kathryn
G. Dewey, Jeffrey I. Gordon*
*Corresponding author. E-mail: [email protected]
Published 19 February 2016, Science 351, aad3311 (2016) DOI: 10.1126/science.aad3311
This PDF file includes:
Materials and Methods Supplementary Text Figs. S1 to S10 Full Reference List Captions for Tables S1 to S17
Other Supplementary Material for this manuscript includes the following: (available at www.sciencemag.org/content/351/6275/aad3311/suppl/DC1)
Tables S1 to S17 as an Excel file
2
Materials and Methods Human studies
Details of the 317 infant twin pairs and three sets of triplets enrolled in the Malawi Twin Cohort study from birth to 32 months of age are described in an earlier publication (13). Enrollment for the iLiNS-DYAD-M study [clinicaltrials.gov #NCT01239693] was open to consenting women over the age of 15 years, with ultrasound confirmation of a pregnancy that was <20 weeks, who had come for prenatal care at one of three study clinics located in Mangochi District of southern Malawi (one public district hospital in Mangochi, one rural semi-private hospital in Malindi, and one rural public health center in Lungwena). This randomized, controlled clinical trial tested the effects of providing Small Quantity Lipid-based Nutrient Supplements (SQ-LNS) to pregnant and lactating women through 6 months postpartum and to their infants from 6-18 months of age (16).
Bacterial 16S rRNA sequencing DNA was extracted by suspending pulverized human fecal samples (~50mg),
or in the case of mouse studies resuspending fecal pellets, in a solution containing 500µL of extraction buffer [200 mM Tris (pH 8.0), 200 mM NaCl, 20 mM EDTA], 210µL of 20% SDS, 500µL phenol:chloroform:isoamyl alcohol (pH 7.9, 25:24:1, Ambion), and 500µL of 0.1-mm diameter zirconia/silica beads. Cells were mechanically disrupted using a bead beater (BioSpec Products, Bartlesville, OK; maximum setting for 4 minutes at room temperature), followed by extraction with phenol:chloroform:isoamyl alcohol and precipitation with isopropanol.
PCR was used to generate amplicons (~365bp) of variable region 4 (V4) of the bacterial 16S rRNA gene using primers and cycling conditions described in a previous report (11). PCR primers incorporated sample-specific barcodes allowing samples to be subjected to multiplex sequencing using the Illumina MiSeq instrument (paired-end 250nt reads). Paired V4-16S rRNA sequences were trimmed to 200bp and merged into a single sequence with Flash software (24). Merged sequences were filtered for low quality reads and binned according to their sample-specific barcodes. Reads were clustered into 97%ID OTUs using UCLUST (14) and the Greengenes OTU reference database (version May 2013). Reads that failed to hit the reference dataset were clustered de novo using UCLUST. A representative OTU set was created using the most abundant OTU from each bin. Reads were aligned using PyNAST. A custom dataset of manually modified NCBI bacterial taxonomy (18) was used to train the Ribosomal Database Project (RDP) version 2.4 classifier (15) and to assign taxonomy to picked OTUs. Validation of this assignment strategy is described in a previous publication (18). For analysis of bacterial abundance and colonization efficiency, rare OTUs (<0.1% relative abundance within each sample) were removed.
Random Forests-derived models and statistical analysis Models were generated using the randomForests package in R. 1000 trees were
generated per iteration, and 100 iterations were performed to generate average MSE values. The rfcv function was utilized for cross-validation.
3
Univariate analyses (t-tests and ANOVA) were performed using Microsoft Excel or Prism 6.0 (GraphPad Software, Inc.). Heatmaps were generated using Gene-E software (http://www.broadinstitute.org/cancer/software/GENE-E/index.html). Spearman’s rank correlation analyses were performed in the R environment using the ltm package and rcor.test function.
Gnotobiotic mouse studies All gnotobiotic mouse experiments were performed using protocols approved by
the Washington University Animal Studies Committee. Male germ-free C57BL/6J mice were maintained in sterile, flexible, plastic gnotobiotic isolators (Class Biologically Clean Ltd., Madison, WI) under a strict 12-hour cycle (lights on at 0600h, off at 1800h). Mice were fed an autoclaved low-fat, polysaccharide-rich chow (LF/HPP) diet (B&K University, East Yorkshire, U.K.; diet 7378000) from weaning until 3 days prior to the beginning of an experiment. At that time, 4.5-week-old male mice were switched to the M8 diet for the remainder of the experiment. All animals were euthanized without prior fasting.
Preparation of the Malawi-8 (M8) Diet –The macro- and micronutrient content of components of this prototypic Malawian diet was computed using the USDA Nutrient Database. Linear programming was then employed to calculate combinations of ingredients that resembled the mean energy and nutrient values for foods identified from the diet survey.
The M8 diet was prepared as follows. Maseca® corn flour was obtained from Restaurant Depot (College Point, NY). The remaining ingredients were purchased from Whole Foods Supermarkets. A relish containing 2kg mustard greens, 1.5kg onions, and 1.5kg tomatoes was first prepared by pureeing them in a food processor (Robot Coupe Model R23, Jackson, MS) and cooking the puree in 1L water for 60 minutes on a Corning stirrer/hot plot (high setting, until browned). After cooking, the relish was combined with a pureed mixture of 1kg ground peanuts, 700g soaked red kidney beans, 1kg canned pumpkin, and 2.5kg peeled bananas in an industrial mixer (Globe SP30P 30-quart pizza mixer; gear speed 1; Globe Food Equipment Company, Dayton, OH). Corn flour (5kg) and hot, freshly autoclaved water (5L) were then added slowly and mixed using the industrial mixer for 5 minutes to ensure uniformity of the food. Dry pellets of the M8 diet were then generated (extruded diameter, 0.5-inch) (Dyets, Inc. Bethlehem, PA), placed in FDA/USDA-compliant poly-nylon vacuum pouches (#S-7556; Uline, Pleasant Prairie, WI; 500 g aliquots), double-bagged and sterilized by irradiation (20-50 kGy) within 24 hours of production (Steris Co; Chicago, IL). The nutritional content of all cooked and irradiated custom diets was defined by N.P. Analytical Laboratories (St Louis, MO) (Table S5). Irradiated food was stored at 4°C for up to six months. Sterility was determined by resuspending a small aliquot of each batch of food in pre-reduced Gut Microbiota Medium (GMM; (25)) under anaerobic conditions and incubating the suspension for 3 days at 37°C. Sterility was further verified by subculture on pre-reduced anaerobic GMM agar plates.
Generation of clarified stool for gavage into gnotobiotic mice – An aliquot of the frozen sample was pulverized in a Biosafety Class II hood with a ceramic mortar and pestle filled with liquid nitrogen. An aliquot (1g) of the pulverized material, sealed in a sterile screw-capped tube (Axygen SCT-200-C-S), was brought into an anaerobic Coy
4
chamber (atmosphere: 20% CO2, 5% H2, and 75% N2), immediately suspended in 15mL GMM and vortexed on maximum speed (four cycles of blending for 20 seconds followed by a 30 second pause). The sample was allowed to stand for 5 minutes so that particulate matter could settle by gravity; the resulting supernatant was passed through a 100µm pore diameter filter (BD systems, Inc. Franklin Lakes, NJ) to remove remaining particulate material, mixed with an equal volume of pre-reduced GMM containing 30% glycerol (final concentration 15% glycerol) and placed in Wheaton crimp top tubes for storage at -80°C. Mice received a single oral gavage (200mL) of a clarified uncultured human fecal sample 3 days after switching to the M8 diet.
Quantitative magnetic resonance (qMR) analysis of body composition – Body composition was defined using an EchoMRI-3in1 instrument (EchoMRI, Houston, TX). Each mouse was transported from a gnotobiotic isolator to the MR instrument in a 9.5” long, 2” diameter, HEPA filter-capped glass vessel. At each time point surveyed, each mouse was scanned 2-3 times in the instrument over a 5 minute period and the values from each scan were averaged.
Sample collection – Fecal samples were collected at defined times after gavage. At sacrifice, cecal contents, liver, brain, and skeletal muscle (gastrocnemius) were collected and immediately frozen in liquid nitrogen. Gastrocnemius muscle was freeze-clamped immediately after sacrifice using pre-chilled forceps.
Defining the efficiency of microbiota transplantation – Transplantation efficiency was assessed by comparing 97%ID OTUs represented in the input human donor sample, weighted to relative abundance, and in the fecal microbiota of recipient mice sampled throughout the course of the experiment [Table S4B; note that microbiota from four 6-month-old singleton infants from iLiNS-DYAD-M (two healthy, two undernourished) and four 18-month-old co-twins from the Malawian twin study (one healthy, three undernourished) had >50% transplantation efficiency]. Bacterial viability, quantified by the number of colony forming units 24 hours after plating onto GMM medium, was significantly greater (p<3.2x10-6; Student's t-test) in the case of fecal samples from the twins, which had been flash frozen in liquid nitrogen within 5-10 minutes after they were produced. In iLiNS-DYAD-M, the time to freezing was often several hours.
Micro-computed tomography (micro-CT) – Femurs were harvested from mice at time of sacrifice, cleaned of soft tissue and stored at 4°C in 70% ethanol solution until scanned. Micro-CT was performed using a µCT 40 desktop, cone-beam instrument (ScanCO Medical, Brüttisellen, Switzerland). For cortical analyses, 200-300 slices were taken of each femur in the transverse plane with a 6µm voxel size (high resolution). For all cortical scans, slices began at the midpoint of the femur and extended toward the distal femur. Boundaries of and thresholds for bone were drawn manually using the µCT 40 software, and volumetric parameters (bone volume/tissue volume, bone mineral density, and cortical thickness) were calculated using custom scripts.
Cohousing experiments – Four days after gavage of donor microbiota, all mice were placed into HEPA filter-capped glass transport tubes for qMR measurements. Following qMR, control mice were returned to their original gnotobiotic isolator and housed with a cagemate of similar body weight colonized with the same donor microbiota (H-H controls; Un-Un controls). For pairing of UnCH and HCH mice, animals were transported to a new gnotobiotic isolator. All mice received fresh autoclaved
5
bedding (Aspen wood shavings, NEPCO) upon initiation of cohousing. Bedding was changed weekly in all cages in each gnotobiotic isolator.
We defined the direction of invasion by calculating the log odds ratio of the probability of a H origin (PH) or an Un origin (PUn) for each species-level taxon or 97%ID OTU, i:
log! !"! !"#! A positive log odds ratio indicates that a species or 97%ID OTU is derived from a H source, while a negative log odds ratio indicates a Un source. An invasion score was calculated to quantify the success of invasion of each species or 97%ID OTU, i, into each cohousing group, j:
!"#$%&'"!!"#$%!" = log! !!" !!" where !!" is the average relative abundance of taxon i in all fecal samples collected from group j after cohousing, and !!" is its relative abundance in all samples from that group prior to initiation of cohousing. The mean of the distribution of invasion scores for UnCH animals was significantly higher than that for Un-Un controls (P ≤ 0.0085, Welch’s two-sample t-test). This was not the case for HCH animals compared to H-H controls (P > 0.05). Targeted mass spectrometry
Liver, brain, and muscle tissue samples were homogenized in 50% aqueous acetonitrile containing 0.3% formic acid (50mg wet weight tissue/mL solution) using a high-speed homogenizer (IKA Model #EW-04739-21) set at maximum speed for 30-45 seconds. Samples were maintained on ice or dry ice throughout homogenization.
Amino acids, acylcarnitines, organic acids, acyl CoAs and ceramides were analyzed using stable isotope dilution techniques. Amino acids and acylcarnitine measurements were made by flow injection tandem mass spectrometry using sample preparation methods described previously (26, 27). Data were acquired using a Waters AcquityTM UPLC system equipped with a TQ (triple quadrupole) detector and a data system controlled by MassLynx 4.1 operating system (Waters, Milford, MA). Organic acids were quantified according to a previously published protocol (28) using Trace Ultra GC coupled to ISQ MS operating under Xcalibur 2.2 (Thermo Fisher Scientific, Austin, TX). Acyl CoAs were extracted and purified (29-31), and analyzed by flow injection analysis using positive electrospray ionization on a Xevo TQ-S triple quadrupole mass spectrometer (Waters, Milford, MA). Heptadecanoyl CoA was employed as an internal standard. Ceramides were extracted (32) and analyzed by flow injection tandem mass spectrometry using a Xevo TQS spectrometer (Waters Milford, MA) for precursors of m/z 264.
Culturing and characterization of age- and/or growth- discriminatory bacterial strains from the microbiota of healthy Malawian children
Culturing – Fecal samples were resuspended in 10ml of GMM under anaerobic conditions, diluted 1:1000, and then plated onto GMM and incubated at 37°C under anaerobic conditions overnight. Isolated colonies were picked into single wells of a deep 96-well plate, with 600µl liquid GMM per well, and grown anaerobically at 37°C overnight. Cultures were then mixed with an equal volume of pre-reduced GMM containing 30% glycerol (final concentration 15% glycerol) for storage at -80°C. To assign taxonomy, DNA was extracted from each well and V4-16S rRNA gene amplicons were generated and sequenced.
6
Taxonomies were confirmed by sequencing full-length 16S rRNA amplicons produced using primers 8F and 1391R.
Genome sequencing of two strains obtained from the healthy infant Malawian donor 8243C – DNA libraries for sequencing R. gnavus TS8243C and C. symbiosum TS8243C on Illumina MiSeq were prepared using a multistep protocol of end-repair, A-tailing, and ligation to customized sample-specific barcoded Illumina adapters. The resulting adapter-ligated DNA was size-selected (250-350bp) using agarose gel electrophoresis and then used as a template for enrichment PCR (cycling conditions: 98°C for 30 sec, followed by 18 cycles of 98°C for 10 sec, 65°C for 30 sec, 72°C for 45 sec, and then 72°C for 5 minutes). Multiplex sequencing was performed (paired-end 250nt reads). Genomes were assembled with the mira software package (33). For R. gnavus TS8243C, N50 contig length was 2,263,429bp with an aggregate genome size of 3.7Mbp. For C. symbiosum TS8243C, N50 contig length was 328,298bp with a 5.3Mbp genome size. Annotation was initially performed using PROKKA (34).
Genome-based reconstruction of selected metabolic subsystems in R. gnavus and C. symbiosum strains – Reconstruction of selected metabolic pathways (also termed as subsystems) was based on functional gene annotation and prediction using two comparative genomics techniques: (i) homology-based methods and (ii) genome context analysis. First, predicted proteomes were scanned against the KEGG Orthology, Pfam, CAZY, TCDB, Uniprot and SEED databases to identify proteins potentially involved in the carbohydrate utilization, amino acid metabolism and vitamin and cofactor biosynthesis and salvage pathways. The genomic and functional contexts of loci encoding the proteins were then analyzed and the respective metabolic pathways and transcriptional regulons were reconstructed. Additional closely related bacterial genomes available in the Integrated Microbial Genomes (35) and SEED databases (36) were used for comparative genomics-enabled pathway and regulon inference. Transcriptional regulons (sets of genes co-regulated via shared transcription factors (TFs) and transcription factor binding sites (TFBS) or shared RNA elements, riboswitches, responding to respective pathway metabolites) were predicted and reconstructed using the comparative genomics approach (37) implemented in the RegPrecise database (38). This integrative subsystems-based approach was used previously for reconstruction of carbohydrate utilization pathways and regulons in members of Shewanella, Thermotoga, and Bacteroides lineages (39-41).
Supplementary Text Effects of gut microbiota immaturity on host metabolism
As noted in the main text, we cross-referenced the results of metabolic profiling conducted in mice treated with R. gnavus and C. symbiosum and untreated controls, with those obtained from mice harboring the same undernourished donor’s microbiota and the healthy donor’s microbiota used in the cohousing experiments. For the latter experiments, serum, liver, gastrocnemius muscle, and brain were obtained at the time of euthanasia three weeks after gavage of these microbial communities, and analyzed by targeted mass spectrometry (N = 5-6 mice/treatment group). The two donors’ microbiota conferred distinct metabolic phenotypes with the major differences involving fatty acid and amino acid metabolism; mice colonized with the healthy donor’s microbiota had significantly higher concentrations of long chain acylcarnitines in their serum and liver
7
and significantly decreased concentrations of a number of amino acids in their gastrocnemius muscle (and cecal contents) (P < 0.05; Student’s t-test; Fig. S7, Table S16). These data support the notion that mice colonized with the healthy donor microbiota funnel amino acids into new protein synthesis and lean body mass, whereas those colonized with the undernourished donor microbiota preserve the amino acid pools for oxidation and energy production. Consistent with this idea, increased levels of the branched-chain amino acids Val and Leu/Ile in the gastrocnemius muscle of mice colonized with undernourished donor microbiota were accompanied by an increase in a metabolite produced by their oxidation, C3 (propionyl) acylcarnitine (Table S16).
These differences in metabolic profiles extended to the brain (Fig. S8; Table S16). Echoing results from serum and liver, levels of long chain acylcarnitines were significantly elevated in the cerebral cortex of mice harboring the healthy donor microbiota compared to the undernourished donor microbiota (P < 0.05; Student’s t-test). Amino acid metabolism was also impacted with concentrations of histidine, proline, glycine, arginine, and citrulline being significantly higher in the brains of mice with the healthy donor’s microbiota. Additionally, there was evidence of perturbations in the TCA cycle, with citrate levels being significantly higher in animals harboring the undernourished infant’s gut community.
There were strong correlations between the relative abundances of bacterial species in the fecal microbiota of gnotobiotic mice and serum C10, C12, and C14 acylcarnitine levels, as well as levels of amino acids in gastrocnemius muscle. Species enriched in the healthy community that significantly correlated to serum acylcarnitines and amino acids included taxa that were also weight- and/or lean mass gain-discriminatory: R. gnavus (C10:2, C12, C12-OH/C10-DC acylcarnitines; valine, leucine/isoleucine, and methionine), Dorea formicigenerans (C10, C10:1, C12, C12-OH/C10-DC , C14, C14:2 acylcarnitines; methionine), F. prausnitzii (C12-OH/C10-DC, C14 acylcarnitines, leucine/isoleucine), and Clostridium symbiosum (C10:1, C12, C12-OH/C10-DC, C14 acylcarnitines; and 11 amino acids including valine, leucine/isoleucine, and methionine) (P < 0.05, see Figs. 2, S2, S9A,B; Table S17 for a complete list with Spearman’s ranked correlation ρ-values). B. longum and F. prausnitzii, the two age- and growth-discriminatory species, were also correlated with hepatic levels of the TCA cycle intermediate citrate (P < 0.05; Fig. S9C, Table S17).
While these two donors’ microbiota transmitted discordant weight, lean mass gain and metabolic phenotypes, they did not transmit discordant bone morphometric phenotypes (P > 0.05 on all metrics; Student’s t-test). This implies, perhaps not surprisingly, that ‘anthropometrically undernourished’ and ‘anthropometrically healthy’ donor microbiota do not transmit discrete sets of effects, but may produce overlapping phenotypes, each component of which may vary in its degree of discordance from donor to donor. A corollary is that different host phenotypes need to be assayed to obtain a more robust functional classification of ‘healthy’ versus ‘undernourished’ microbiota.
0
5
10
15
Training (N = 113 samples)
Testing (N = 107 samples)
Num
ber o
f sam
ples
0 4 8 12 16 20 24 28 34 0 4 8 12 16 20 24 28 32Age (months)
ns
Fig. S1.
Distribution of ages at the time of sample collection of healthy microbiota donors that were used to generate the Random Forests-derived model of age-discriminatory bacterial 97%ID OTUs. Histogram of donor ages in the training set cohort (N = 113 samples from 14 twin pairs and 1 set of triplets; minimum age 0.6 months, maximum age 33.51 months, median age 13.6 months) and testing set cohort (N = 107 samples from 13 twin pairs and 1 set of triplets; minimum age 0.98 months, maximum age 31.8 months, median age 12.02 months). ns, no significant difference based on Kolmogorov-Smirnov test (P = 0.58).
9
Fig. S2.
Random Forests-derived model of lean body mass gain discriminatory OTUs. The 30 most lean mass gain discriminatory OTUs with their taxonomic assignments, ranked by feature importance as determined by increase in mean squared error. The lean body mass gain model explained ~58% of the observed phenotypic variation (P < 0.0001, permutation test; 999 permutations). Models were iterated 100 times, and average mean squared error values were used to generate the final rankings. Red indicates OTUs that appear within the 30 most discriminatory OTUs for both the weight and lean mass gain models. Purple indicates species that also appear in the 25-taxa sparse model of Malawian microbiota maturation. Colors to the right of the OTU ID numbers represent Spearman’s rank correlation of the same OTU ID to chronological age in the healthy Malawian infant/child cohort (Table S9).
Increase in Mean Squared Error
0 10 20 30
NoneDorea formicigenerans
ClostridiumClostridium innocuum
Bifidobacterium longumClostridium butyricum
ClostridiumClostridium
Ruminococcus gnavusPaenibacillus
Ruminococcus gnavusRuminococcus gnavus
Weissella cibariaEnterococcus
ClostridialesClostridium sordellii
Clostridium disporicumPediococcus pentosaceus
Enterococcus faecalisPediococcus
EggerthellaEnterococcus
Finegoldia magnaClostridium symbiosumRuminococcus gnavus
BifidobacteriumFaecalibacterium prausnitziiFaecalibacterium prausnitzii
Bacteroides vulgatusRuminococcus gnavus
Taxonomic annotationRank
123456789
101112131415161718192021222324252627282930
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
OTU ID
de novo 12617437430232038013709990
72820238205
242473744484922724175214836625756512714942
64384138389
4473509242298
4383953773251
103341343809761141218431692810966104378683342397681370265871195436
36005041860112
P va
lue
Lean mass gain discriminatory taxa
Spear
man’s
ρ
0.4964P value <0.0001
Spearman’s correlation of OTU to age in healthy Malawi cohort
minimum ρ (-0.5) maximum ρ (0.5)
Cortical BV/TVm
g H
A/cc
m
Healthy Undernourished Undernourished0.25
0.30
0.35
0.40
0.45 P = 0.05
Cortical vBMD
mg
HA/
ccm
Healthy300
350
400
450
500
550 P = 0.07
Trabecular BV/TV
Trabecular Connectivity
mg
HA/
ccm
0
200
400
600
mg
HA/
ccm
5
6
7
8
9
10
mg
HA/
ccm
0.10
0.12
0.14
0.16
0.18
0.20
6 months 18 months
6 months 18 months 6 months 18 months
6 months 18 months
P < 0.0001
Trabecular Number
P < 0.0001
Trabecular Spacing
P < 0.0001
A
B Trabecular BMD
mg
HA/
ccm
0
50
100
150
200
250
6 months 18 months
P = 0.0005
mg
HA/
ccm
0
0.1
0.2
0.3
0.4 P < 0.0001
Fig. S3.
10
Effects of microbiota donor health status and age on femoral bone morphology. Micro-CT was performed on femurs harvested from gnotobiotic mice colonized with microbiota from 6- and 18-month-old healthy or undernourished donors (in all cases, the efficiency of transplantation was >50%). Each circle represents a single mouse. (A) Bone morphometric phenotypes expressed as a function of donor nutritional status irrespective of donor age. Differences in the ratio of bone volume to tissue volume (BV/TV) and volumetric bone mineral density (vBMD) in the cortical region of femurs are shown. (B) Bone morphometric phenotypes expressed as a function of age, irrespective of donor nutritional status. Differences in trabecular bone mineral density (BMD), BV/TV, connectivity, number, and spacing are shown. P values are based on a Mann-Whitney test. Y-axis abbreviation; HA/ccm, hydroxyapatite per cubic centimeter.
Cortical BV/TV
Cortical thickness
Trabecular BMD Trabecular BV/TVA B
C D
E
Increase in Mean Squared Error
020
040
060
080
0
ClostridiumRuminococcus gnavusEnterococcus faecalis
Bifidobacterium longumClostridium
Leuconostoc lactisClostridium sordellii
Clostridium xylanolyticumClostridium neonatale
Clostridium disporicumFaecalibacterium prausnitzii
Weissella cibariaEnterococcus
Clostridium paraputrificumFaecalibacterium prausnitzii
Dorea longicatenaClostridium innocuum
Eggerthella lentaEnterococcus
Clostridium
OTU ID
4448492138389
43935323709990188047265871
35761744316928
6438443814304383953
730003197419242298
43651413903651
728201033413342397
3203801
Rank
123456789
1011121314151617181920
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
Increase in Mean Squared Error
0.000
00
0.000
05
0.000
10
0.000
15
0.000
20
0.000
25
BacteroidesEubacterium limosum
Faecalibacterium prausnitziiBifidobacterium bifidum
Faecalibacterium prausnitziiClostridium disporicumClostridium lentocellum
PaenibacillusBifidobacterium longum
Enterococcus faecalisWeissella
Clostridium innocuumClostridium
EnterococcusWeissella cibaria
EnterococcusClostridiales
Clostridium xylanolyticumClostridium neonatale
Eggerthella lenta
OTU ID
439353273000
319741934211013838964384
431692839036513709990663969
103341372820
214836643528054383953265871365385
43814304434268
1992
Rank
123456789
1011121314151617181920
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
Cortical vBMD
Increase in Mean Squared Error0.0
0000
0.000
01
0.000
02
0.000
03
0.000
04
EnterococcusClostridium paraputrificum
EnterococcusBacteroides caccaeEnterobacteriaceae
Odoribacter splanchnicusBifidobacterium longum
Clostridium innocuumWeissella cibaria
ClostridialesEscherichia coli
Eubacterium limosumClostridium xylanolyticum
Clostridium bolteaeEnterococcus
Enterococcus faecalisClostridium neonatale
BifidobacteriumEggerthella lenta
WeissellaOTU ID
663969439353268137073000
1033413431692844695763197419443426878295319593764384
370999072820
44727214425571193528138389
3576174295146
Rank
123456789
1011121314151617181920
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
Increase in Mean Squared Error
010
020
030
040
0
Eggerthella lentaBacteroides fragilis
NoneRuminococcus torques
ClostridialesFaecalibacterium prausnitzii
Paenibacillus graminisRuminococcus sp 5 1 39BFAA
Leuconostoc lactisClostridium lentocellum
Clostridium xylanolyticumEnterococcus
Weissella cibariaEubacterium desmolans
Bacteroides fragilisDorea longicatena
Faecalibacterium prausnitziiDorea longicatena
Faecalibacterium prausnitziiClostridium
OTU ID
4448492265871188047
4381430188194
434362731728664384
4316928319741943528054365141193666179040181422114348
4396688de novo OTU 1870
44338234393532
Rank
123456789
1011121314151617181920
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
Increase in Mean Squared Error0.0
000
0.000
1
0.000
2
0.000
3
0.000
4
Escherichia coliBacteroides ovatus
BacteriaBifidobacterium bifidum
ClostridialesClostridium perfringensBifidobacterium longum
Clostridium paraputrificumClostridium neonatale
Bacteroides ovatusRoseburia intestinalis
BifidobacteriumPaenibacillus
Bacillus licheniformisWeissella
Weissella cibariaClostridium lentocellum
Eggerthella lentaClostridium perfringens
Clostridium xylanolyticumOTU ID
3197419444567343935324352805
64384663969
44423834436752681370392887
410331773000
357617472820
4448928195937365385191153
4467447782953
Rank123456789
1011121314151617181920
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
0 20 40 60 80100400 500 600 700
0
1,000
2,000
3,000
Cro
ss-v
alid
atio
n er
ror
Number of OTUs used forbone metric discrimination
Number of OTUs used forbone metric discrimination
0 20 40 60 80 100
400 500 600 7000.0000
0.0005
0.0010
0.0015
0.0020
0.0025
Cro
ss-v
alid
atio
n er
ror
0 20 40 60 80 100
400 500 600 7000.0000
0.0005
0.0010
0.0015
Cro
ss-v
alid
atio
n er
ror
Number of OTUs used forbone metric discrimination
0 20 40 60 80 100400 500 600 700
0
500
1,000
1,500
Number of OTUs used forbone metric discrimination
Cro
ss-v
alid
atio
n er
ror
0 20 40 60 80100
400 500 600 7000.00000
0.00005
0.00010
0.00015
0.00020
Cro
ss-v
alid
atio
n er
ror
Number of OTUs used forbone metric discrimination
Fig. S4.
11
Random Forests-derived models of OTUs that discriminate femoral bone metrics defined by micro-CT. Top 20 most discriminatory OTUs, ranked by increase in mean squared error, for (A) trabecular bone mineral density, (B) trabecular bone volume per tissue volume, (C) cortical volumetric bone mineral density, (D) cortical bone volume per tissue volume, and (E) cortical thickness. Insets show cross-validation of each model; error decreases as OTUs are added to the model. Red indicates that the OTU is also among the top 30 ranked OTUs the Random Forests-derived models of both weight- and lean mass gain-discriminatory taxa. Boldface indicates that the OTU appears in all five bone metric models.
B
A
0 10 20 3080
100
120
140
160
Days post colonization
% In
itial
lean
mas
sR
elat
ive
abun
danc
e (%
)
HealthyStunted/underweight
0
10
20
30
40
50
60
70
80
90
100
H Screen
H Cohousing 1
H Cohousing 2
UnScreen
UnCohousing 1
UnCohousing 2
Actinomyces radingae Bacteroides Bifidobacterium Bifidobacterium bifidum *
++
+
+
+
+
+
++
+
+
+
+
+
***
***
*
Bifidobacterium longum Blautia producta Clostridiales Clostridium Clostridium bartlettii Clostridium bolteae Clostridium butyricum Clostridium colinum Clostridium difficile Clostridium disporicum Clostridium glycolicum Clostridium hathewayi Clostridium innocuum Clostridium neonatale Clostridium nexile Clostridium paraputrificum Clostridium perfringens Clostridium ramosum Clostridium symbiosum Clostridium xylanolyticum Coprobacillus cateniformis Dorea formicigenerans Eggerthella lenta Enterobacteriaceae Enterococcaceae Enterococcus Enterococcus faecalis Escherichia coli Eubacterium hallii Eubacterium limosum Eubacterium rectale Faecalibacterium Faecalibacterium prausnitzii Gordonibacter pamelaeae Lactobacillus Lactobacillus ruminis Leuconostoc lactis None Paenibacillus Paenibacillus sp P15 9 Paenibacillus taiwanensis Paenibacillus xylanilyticus Pediococcus Pediococcus pentosaceus Ruminococcus gnavus Streptococcus Streptococcus gallolyticus Turicibacter sanguinis Vagococcus Weissella Weissella cibaria
Sparse model age-discriminatory taxon
Within top 30 of weight-discriminatory taxa
Within top 30 of lean mass gain-discriminatory taxa
Fig. S5.
12
Discordant growth phenotypes transmitted by two exemplary iLiNS-DYAD-M donor gut communities: one from a healthy 6-month-old donor with normal gut microbiota maturity and the other from a severely stunted/underweight 6-month-old donor with significant microbiota immaturity. (A) Gain of lean body mass (N = 5 recipients/donor microbiota). Mean±SEM values are shown. P < 0.0001 (2-way ANOVA). (B) Bacterial composition of the fecal microbiota of gnotobiotic mice in the cohousing experiments (relative abundances, averaged over all time points, of taxa from all animals in each treatment group). Abbreviations: ‘screen’, initial survey of mice harboring donor microbiota; ‘H’, cagemates colonized with the healthy donor’s microbiota in each cohousing experiment; ‘Un’, cagemates colonized with the severely stunted/underweight donor’s microbiota.
Citrate
Pyk2.7.1.40
PEP
Eno 4.2.1.11
Glycerate-2P
Gpm 5.4.2.11
Pgk 2.7.2.3
Gap 1.2.1.12
Glyceraldehyde-3PKDG-6P KDGKdgK
2.7.1.454.1.2.14KdgR
PpdK
Tpi 5.3.1.1
Fba 4.1.2.13
Fructose-1,6P
Pfk 2.7.1.11
Fructose-6P
Pgi 5.3.1.9
Glucose-6P Gluconate-6PZwf
1.1.1.49/3.1.1.31Gnd1,2 1.1.1.44
D-Ribulose-5P
Rpe5.1.3.1
Xylulose-5P
Ribose-5P5.3.1.6
RpiTkt
Tal
2.2.1.12.2.1.2
Glk2.7.1.2
Glucose
3.1.3.11 Fbp
Pgl
PduP1.1.1.27
L-lactaldehyde
FucA*
4.1.2.17
FucK
2.7.1.51
FucI
5.3.1.25
FucU
Fucose Fuc* (ABC)
Propane-1,2-diol
Propionaldehyde
Propionyl-CoA
Propanol
FucO1.1.1.77
4.2.1.28PduABCD
Glycerol Glycerol-P GlpDGlpKUgp(ABC)1.1.5.32.7.1.30
PduL2.3.1.222
XyluloseXylB
2.7.1.17XyloseXylA*
5.3.1.5
L-Ribulose-5PAraD
5.1.3.4RibuloseAraK
2.7.1.16ArabinoseAraA
5.3.1.4
XylR*
Ara* (ABC)
FucR*
AraXY*
Mannitol MtlAF(PTS)
MtlRMannitol-1P
UxuR1*GlucuronateUxuA
4.2.1.8
UxuB
1.1.1.57
UxaC2
5.3.1.12GlucuronidesUxu* (TRAP)Agl1,2
UxuR2*
KduD
1.1.1.125
KduI
5.3.1.17KduR*
FucA*
4.1.2.19
RhaB
2.7.1.5
RhaA
5.3.1.14Rhamnose
Rhi* (TRAP)
Fuculose-1P
Rhamnulose-1P
KdgA
Rhamnoga-lacturonides
Rhi* (ABC)
RhaR*RhgR*
Acetoacetyl-CoA
Acetyl-CoA
Thl
1.1.1.157
2.3.1.9/2.3.1.16
Hbd
3-hydroxybutanoyl-CoA
4.2.1.55Crt
Crotonoyl-CoA
1.3.99.2
Butyryl-CoA2.8.3.8
But
Acetate AckA Pta
Pyruvate
Por1.2.7.-
PflABFormate
Ethanol AdhE
Butyrate
LdhALactate
Propionate
Acr
Lcd
Pct 2.8.3.1
4.2.1.54
1.3.1.95
1.1.1.27
EtfABButanol
1.1.1.-Bdh
Ethanolamine EutE1.2.1.10
1.1.1.1/1.2.1.10
2.7.2.1 2.3.1.8
2.3.1.54
Gluta-conyl-CoA
GcdABC
4.1.1.70
HgdABC
4.2.1.-2-hydroxyglutaryl-CoA GctAB
2.8.3.122-hydroxyglutarate
1.1.99.39 ?
GlutamateGdh1.4.1.4
GltS
4-hydroxybutyrate Succinatesemialdehyde
AbfH1.1.1.61
4-hydroxy-butyl-CoA
AbfD4.2.1.120 / 5.3.3.3
Cat
2.8.3.-
AldD* 1.2.1.16
FocA
EutABC4.3.1.7
AraI*
BglF1,2,34,5,6 (PTS)
BglG1,2,3*, BglR*, BglQ*
-glucosides-6P
-glucosides
BglT*(MFS) -glucosides
BglK 2.7.1.85
BglA1,2;BglC;AbgA1,2,33.2.1.86
PtsG (PTS)Glucose
Glucose-1P GalE
5.1.3.2
GalT GalK
2.7.1.6
GalM
5.1.3.32.7.7.10Galactose
Pgm 5.4.2.2 GalR*
Aga* (ABC)AgaL1,2
BgaT1,2,3*(MFS)BgaH1,2,3
Unknownoligosaccharide
RggT*(MFS)Rhamnose
BglS1,2,3,4,5,6,7 RamA1,2 BgaH4
Fructose FruA (PTS) Fructose-1P FruK
2.7.1.56
MtlD*
1.1.1.17
FruR
Sucrose ScrA (PTS) Sucrose-6P
ScrRFructooligo-
saccha-rides (FOS)
Bfr1*(ABC)
ScrB1,23.2.1.26
FructoseBfrA1
3.2.1.26
FOS ScrK2.7.1.4BfrR*
Glycerol
RhiN1,2
Fucose
Arabinose
-galactosides
-galactosides
Alf* (ABC)
AlfA1,2,3,4,5 3.2.1.51
-fucosides
MaltodextrinMal1,2 (ABC)MalQ NplT, NepU PulA1,2
GlgP2.4.1.1
2.4.1.25 3.2.1.135 3.2.1.41MalL
3.2.1.10
3.2.1.22
3.2.1.233.2.1.403.2.1.21
GluconateGntK2.7.1.12
5-keto-gluconateIdnO1.1.1.69
Gnt* (TRAP) Gluconate,5-keto-gluconate
GntR*
Glycerate GarKGrtP*(MFS)2.7.1.31
GlycerateSdaR
Galactitol-1PGalactitol4.1.2.402.7.1.1441.1.1.251
GatYLacC*GatDGatK*
GatR*
GlpP
N-acetylglucosamine (GlcNAc)
GlcN-6P NagB3.5.99.6
NagA 3.5.1.25 NagT*
5.1.3.9ManNAc-6P ManNAcNanK
2.7.1.60
NANaNanA4.1.3.3
Sialic acidsN-acetylneuraminate (NANa) NanH3.2.1.18
NanX*, NanR*
Hyaluronate
Oligohyaluronate-6P
Unsaturatedglucuronate
ddGlcARegR
?
Rex
2.7.9.1
Glutamate
?
PropionatePrpP*
?PduQ
?
?
?
?
Bcd
OadABG4.1.1.3
Oxaloacetate
Citrate Isocitrate
2-Oxoglutarate
Aco*
4.2.1.3
Icd 1.1.1.41
SuccinateFumarate
Malate
FumA 4.2.1.2
PckA4.1.1.49
MaeB 1.1.1.38CitZ
2.3.3.1
CitFEDC 4.1.3.6 CitN1,2
Acetate
FrdABC
Glutamatebiosynthesis
Amino acidbiosynthesis
Utilized carbon sourcesFermentation products
Aspartatebiosynthesis
AldH/AdhB
D-Lactate?
LdhD
1.1.1.28
Histidine YuiF Histidine
HutH 4.3.13
HutU 4.2.1.49
HutI 3.5.2.7
TCA Cycle(incomplete)
GalacturonateUxaA
4.2.1.7
UxaB
1.1.1.58
UxaC1
5.3.1.12
Pgl?
Polyga-lacturonate
UxaR
XynT (MFS)XynB
3.2.1.20
3.2.1.37-xylosides
3.2.1.15
3.2.1.172
AmyAc3.2.1.54
Galactose
Lac*(ABC)EbgA1,2,3,4
LacR*, LacI*
Lactose3.2.1.23
MelB(MFS)MelA
MelR*
MeliobioseGlucose
BglY1
BglY1
BgaL3.2.1.23
EabC
AgaR1,2* BgaR*
-fucosides
FUC13.2.1.51
FucosyllactoseFulT*(PTS)
Fucose
Axy1,2FulR*
sisylocylG
PPpathway
Fermentation / SCFAs production
Amino acid degradation
Carbohydrate Utilization pathways
NanEGlcNAc-6P
Nan1*,2* (ABC)NagP (PTS) Hya (PTS)
Ugl Ogl
(ddGlcA)
Cellobiose
-glucan
BglB 3.2.1.21
Cbp*(ABC)CbpA1,22.4.120-
Cellobiose
CbpR*LicB 3.2.1.73
GahP2.4.1.211
Galacto-N-biose
Galactose
Extracytoplasmic space
Cytoplasm
Bga* (ABC) -galactosidesBgaL1,23.2.1.23
Chitobiose NagZ3.2.1.52
GlcNAc
??
?
Pyruvatemetabolism
SCFAs production
Bfr2*(ABC) BfrA2
Regulators
Start/end metabolites:
Metabolic intermediates
Function/Localization:Enzymes
Transporters
Genomes:
Ruminococcus gnavus TS8243C
Clostridium symbiosum TS8243C
Central carbohydrate metabolism and fermentation
Fig. S6A.
T-box riboswitch
ArgR repressor
Transcriptionalregulation by:
SAM riboswitch
Lys riboswitch
HisR repressor
CymR repressor
Amino Acid Biosynthesis7-Phospho-2-dehydro-
-3-deoxy-arabinoheptonate
4.2.3.4AroB
Erythrose-4-phosphate
2.5.1.54A Gro
3-Dehydroshikimate
Shikimate
1.1.1.25AroE
3-Dehydroquinate
4.2.1.10AroD
Shikimate-3-phosphate
5-O-(1-Carboxyvinyl)-3-phosphoshikimate
2.5.1.19AroA
2.7.1.71AroK
Chorismate
4.2.3.5AroC
Prephenate
5.4.99.5
TyrB
TyrosinePhenylalanine
4.2.1.51 TyrA 1.3.1.12
TrpG4.1.3.27
Anthranilate
N-(5-Phospho-ribosyl)anthranilate
TrpD
TrpE
1-(2-Carboxyphenylamino)-1-deoxy-ribulose 5-phosphate
5.3.1.24TrpF
Indole-3-glycerol phosphate
TrpC
Indole
TrpA
Tryptophan
4-Hydroxy-
AroH
Phenylpyruvatephenylpyruvate
2.6.1.57
TrpB
2.4.2.18
4.1.1.48
4.2.1.20
4.2.1.20
T *rpXYZ
PheA
2-Oxoisovalerate
IlvE
Valine
2.3.3.13LeuA
LeuC 4.2.1.33
3-Isopropylmalate
LeuB 1.1.1.85
nonenzymatic 2-Isopropyl-3-oxosuccinate2-OxoisocaproateLeucine
2-Isopropylmalate
2-Acetolactate
IlvC 1.1.1.86
2,3-Dihydroxy-isovalerate
IlvD 4.2.1.9
LeuD 4.2.1.33
3-Methyl-2-oxopentanoate
IlvE 2.6.1.42
Isoleucine
2-Aceto-2-hydroxybutanoate
IlvC 1.1.1.86
2,3-Dihydroxy-3-methylpentanoate
IlvD 4.2.1.9
2-Oxobutanoate
IlvE
4.3.1.19
IlvM 2.2.1.6
IlvG 2.2.1.6
2.6.1.42
2.6.1.42
Threonine
I lvA
I Mlv 2.2.1.6
I lvG 2.2.1.6
Pyruvate
IlvBHMGF
4-Aspartyl-phosphate
1.2.1.11Asd
Aspartate-4-semialdehyde
Hom 1.1.1.3
DapA4.3.3.7
Dihydrodipicolinate
DapB 1.17.1.8
MetK 2.5.1.6
Homoserine
metA 2.3.1.46
MetY 2.5.1.49
O-Succinylhomoserine
Homocysteine
metLmetLmetLMetH 2.1.1.13
Methionine
S-Adenosylmethionine (SAM)
5-Methyl-THF
5,10-Methylene-THF
T *hrB
2.7.1.394.2.3.1
Threonine
O-Phospho-homoserine
ThrC
2,6-Diamino-Tetra
Ddh 1.4.1.16
hydrodipicolinatepimelate
LysA 4.1.1.20
Meso-diaminopimelate
Lysine
DapL2.6.1.83
DapF5.1.1.7
Met NPQ
MetF 1.5.1.20
Alanine
Ald
Pyruvate
gcSA
4 11. .1.
N-Acetyl-glutamate
ArgB 2.7.2.8
N-Acetyl-glutamyl 5-phosphate
ArgC 1.2.1.38
N-Acetyl-glutamate 5-semialdehyde
2.6.1.11ArgD
ArgF 2.1.3.9
N-Acetylornithine
N-acetylcitrulline
ArgE 3.5.1.16
Citrulline
ArgG 6.3.4.5
ArgH 4.3.2.1
Argininosuccinate
Arginine
Glutamate2.3.1.1ArgA
2-Oxoglutarate
2.7.2.11ProB
Glutamyl-phosphate
ProA 1.2.1.41
nonenzymatic
Glutamate-5-semialdehyde
1-pyrroline-5-carboxylate
ProC1.5.1.2
Proline
Glutamine
6.3.1.2GlnA
GlnN
GlsA3.5.1.2
6.3.1.2
G DltB1.4.1.13
GlnHPQGlnT
Hydroxypyruvate-3-phosphate
Phosphoserine
PspA* 3.1.3.3
2.6.1.52
Serine
2.1.2.1
Glycine
SerC
Glycerate-3-phosphate
1.1.1.95SerA
CysE 2.3.1.30
O-Acetyl-serine
Cysteine
2.5.1.47CysK
GlyA
SgaA2.6.1.45
GlyB
4.1.2.48
1.4.4.2CO2
Pyruvate
SdaAB4.3.1.17
Phosphoribosyl-AMP
HisI 3.5.4.19
3.6.1.31HisE
5-Phosphoribosyldiphosphate
2.4.2.17is GH Z
-5-amino-1-(5''-phospho-D-ribosyl)N-(5'-Phospho-D-1'-ribulosylformimino)
Phosphoribosyl-ATP
HisA 5.3.1.16
5-(5-Phospho-D-ribosylaminoformimino)-1-(5-phosphoribosyl)-imidazole-
4-carboxamide
-4-imidazolecarboxamide
HisH 2.4.2.-
HisB 4.2.1.19
Imidazole-glycerol phosphate
Imidazole-acetol phosphate
HisC 2.6.1.9
Histidinol phosphate
HisJ 3.1.3.15
Histidinol
HisF 4.1.3.-
1.1.1.23
Histidine
HisD
4.3.1.3HutH
Urocanate
GcvPAB
H *isXYZ
YuiF
4.2.1.49HutU
3-(5-oxo-4,5-dihydro-3H--imidazol-4-yl)propanoate
Formimidoyl-L-glutamate
3.5.2.7HutI
ArgW*
BraB
TyrT*
G tS1,2lGdh1.4.1.4
Trd1 84. .1.2
Tryptamine
Aspartate
Oxaloacetate
Asparagine
AsnA6.3.1.1
AspC2.6.1.1
2.7.2.4
GlutamateGlutamine
AsnB6.3.5.4
AnsA3.5.1.1
LysC
Glutamate
2-oxoglutarate
Function/Localization:Biosynthetic enzymes
Degradation enzymes
Transporters
Genomes:
Ruminococcus gnavus TS8243C
Clostridium symbiosum TS8243C
Fig. S6B.
TPPriboswitchTHF riboswitch
FMN riboswitchB12 riboswitchBirA repressorNaiR repressorPdxr repressor
Vitamin and Cofactor Biosynthesis
Transcriptionalregulation by:
Genomes:
Ruminococcus gnavus TS8243C
Clostridium symbiosum TS8243C
Function/Localization:
Biosynthetic enzymes
Degradation enzymes
Transporters
Precorrin 2
Nicotinamide2.4.2.11
(NaMN)
2.7.7.18
Deamido-NAD
6.3.5.1
Iminoaspartate
NadA
Quinolinate
2.4.2.19
2.5.1.72
NadC
Aspartate
NadB 1.4.3.16
2.7.1.23NadF
NadD
NadE
NAD
NADP
PncB
3.5.1.19
PncANicotinate
B3
NiaY
Biotin
Dethiobiotin
Biotinyl-5'-AMP
7,8-Diaminononanoate
BioD
8-Amino-7-oxononanoate
BioA
2.3.1.47
2.6.1.62
6.3.3.3
2.8.1.6
Biotinylatedproteins
BioB
BioF
Malonyl-ACP
Pimeloyl-ACP
6.2.1.11
BioCG
BioY1BirA B7
Riboflavin
6,7-Dimethyl-8-(D-ribityl)lumazine
FMN
5-Amino-6-(D-ribitylamino)uracil
RibH
2,5-Diamino-6-ribosylamino-4(3H)-pyrimidin
RibD 3.5.4.26
2.5.1.78
2.5.1.9
2.7.1.26
FAD
RibE
RibF RibF
RibA
2.7.7.2
4.1.99.12RibB
GTP3.5.4.25
L-3,4-Dihydroxy-2-butanone4-phosphate
D-Ribulose5-phosphate
5-Amino-6-(5'-phospho-D-ribitylamino)uracil
1.1.1.193
B2RibU
4-Amino-5-hydroxymethyl-2-methylpyrimidine
( )HMP
2.5.1.3
Aminoimidazoleribotide (AIR)
ThiC
2.7.1.49
HMP-PPThiazole-
phosphate
2.7.1.7
Thiamin
Thiamin pyrophosphate(TPP)
ThiN 2.7.4.16
Cysteine
2.7.6.2ThiN
Iminoglycine
4.1.99.17
1-deoxy-D-xylulose-5P
ThiS -SH
ThiG
Tyrosine
ThiH 4.1.99.19
2.8.1.10
ThiF2.7.7.73
ThiazoleThiM
ThiE
ThiD
B1ThiT
ThiXYZ
ThiW
PdxHPyridoxine phosphate
PyridoxalPyridoxine
Pyridoxal phosphate
PdxH
PdxK PdxK 2.7.1.352.7.1.351.4.3.5
PdxTS 2.4.2.-
ribulose-5P
PdxUB6
(R)-4'-Phosphopantothenoyl-L-cysteine
(R)-4'-Phosphopantothenate
-alanine
PanC
Dephospho-CoA
Coenzyme A (CoA)
6.3.2.1
6.3.2.5
4.1.1.36
2.7.1.24
2.7.7.3
Pantothenate
2.7.1.33
Pantoate
Aspartate
PanD
2-Dehydropantoate
1.1.1.1694.1.1.11
3-Methyl-2-oxobutanoate
PanB2.1.2.11
CoaX
CoaB
CoaC
CoaD
CoaE
PanG
PanTB5
Pantetheine
2.7.1.34 CoaX
p-aminobenzoate
Dihydrofolate
7,8-Dihydropteroate
FolC
2-Amino-4-hydroxy-6-hydroxymethyl-
FolP
2-Amino-4-hydroxy-6-hydroxymethyl-7,8-dihydropteridine
FolK
7,8-Dihydroneopterin
FolB
Folate
THF-polyglutamate
4.1.2.25
2.7.6.3
2.5.1.15
6.3.2.12
6.3.2.17
GTP
3.5.4.16
phosphatase
1.5.1.31.5.1.3
FolA
FolA
FolC
FolE
PabA
PabB
4-amino-4-deoxychorismate
Chorismate
4.1.3.38
2.6.1.85
Tetrahydrofolate
FolTB9
a,c-diamidea,c-diamide
Cobalamin
aminopropanol-2P
L-threonine-3P
Adenosylcobinamide
Adenosylcobinamide-GDP
Dimethylbenz midazolei
-ribazole-5P
-ribazoleAdenosylcobalamin
2.5.1.17BtuR
2.7.8.26
2.7.1.1562.7.7.62
6.3.1.10
6.3.5.10
3.1.3.73
2.4.2.21
4.1.1.81
6.3.5.11
CbiC 5.4.1.2
CbiT 2.1.1.196
CbiJ 1.3.1.54
CbiD 2.1.1.195
CbiF 2.1.1.271
CbiG 3.7.1.12
CbiH 2.1.1.31
CbiL 2.1.1.151
CbiE 2.1.1.289
1.16.8.1
CbiK
CbiA
4.99.1.3
CbiP
CbiB
CobU
CobS
CobD
CobT
CobC
BtuFCD B12
2.5.1.17BtuR
Cobalt CoCbiMNQO2+
NiaY
Nicotinamide
Pantetheine
PanF*BioY2*
?
Fig. S6C.
Reconstructed Ruminococcus gnavus TS8243C and Clostridium symbiosum TS8243C metabolic pathways involved in carbohydrate and central carbon metabolism, amino acid biosynthesis and degradation, and vitamin and cofactor biosynthesis. Enzymatic reactions and transporters are depicted by solid and dashed lines, respectively. Red and purple lines indicate the presence of the enzyme or transporter in the cultured strains of R. gnavus and C. symbiosum, respectively, while gray lines indicate absence of the enzyme or transporter in either strain. Common names of enzymes and transporters are shown in black and blue font, respectively. Enzymes are also identified by their enzyme commission (EC) numbers. Major families of carbohydrate transporters indicated in parentheses are ATP-Binding Cassette (ABC) transporters, Major Facilitator Superfamily (MFS) permeases, TRipartite ATP-independent Periplasmic (TRAP) transporters, and phosphoenolpyruvate-dependent PhosphoTransferase System (PTS) transporters. Cognate transcriptional regulators of sugar utilization pathways are highlighted in yellow ovals. Central carbon metabolism genes controlled by the global redox-responsive regulator Rex in both genomes are marked by yellow vertical stripes. Genes involved in amino acid and vitamin metabolism that are controlled by local regulators and metabolite responsive riboswitches are denoted by colored circles. Gene locus tags and functional annotations are listed in Table S13. Genes with novel predicted functions are marked with an asterisk. Putative pathways and transporters for which no gene annotation could be identified are indicated with a question mark. (A) Carbon and energy: central carbon metabolism, carbohydrate utilization, and short chain fatty acid production
Central carbon metabolism – Both the R. gnavus and C. symbiosum strains possess the complete glycolysis/gluconeogenesis and pentose phosphate (PP) pathways, as well as the key enzymes involved in conversion of pyruvate to acetyl-CoA or oxaloacetate, and reverse conversion of oxaloacetate to phosphoenolpyruvate (PEP). The PP pathway in C. symbiosum includes multiple paralogs of genes encoding transketolase (TktAB) and transaldolase (Tal). In addition, C. symbiosum (but not R. gnavus) encodes the oxidative pathway for utilization of glucose-6 phosphate via the PP pathway (Zwf, Pgl, Gnd). The tricarboxylic acid (TCA) cycle is incomplete in both strains. In R. gnavus, it involves two parts: (i) interconversion of pyruvate to malate and fumarate, and (ii) fusion of oxaloacetate and acetyl-CoA to produce 2-oxoglutarate, a key precursor for glutamate biosynthesis. C. symbiosum possesses only the former pathway for conversion of pyruvate to malate and fumarate.
Carbohydrate utilization (CU) pathways – The reconstructed carbohydrate utilization (CU) potential of the two Malawian TS8243C strains differ substantially and include novel carbohydrate-specific regulators, transporters, and enzymes that are not orthologous to previously characterized proteins. R. gnavus has 28 predicted novel regulators, most of which belong to the AraC family, and 35 transporter genes encoding 16 transporter complexes (including 7 ABC transporters). C. symbiosum contains 5 novel regulators from various structural families and 17 transporter genes (5 ABC transporter complexes). Notable novel monosaccharide catabolizing enzymes include xylose isomerase (XylA), tagatose-6P isomerase (LacC), galactitol kinase (GatK), and bifunctional fuculose-P/rhamnulose-P aldolase (FucA) in R. gnavus, as well as arabinose isomerase (AraI) and mannitol dehydrogenase (MtlD) in C. symbiosum.
R. gnavus has predicted pathways for utilization of glucose, fructose and its oligosaccharides, sucrose, lactose, galactosides, fucose and fucosides, N-acetylglucosamine, galacto-N-biose, N-acetylneuraminate, sialic acids, beta-xylosides, rhamnogalacturonides, glucuronides, polygalacturonate, maltodextrins, beta-glucosides and cellobiose, meliobiose, fucosyllactose, hyaluronate and galactitol. C. symbiosum is predicted to utilize glucose, fructose, beta-galactosides, N-acetylneuraminate, chitobiose, galacto-N-biose, arabinose, mannitol and gluconate.
Utilization of other carbon sources – R. gnavus and C. symbiosum have shared and unique pathways for utilization of other carbon sources. Both strains share the glycerol and D-lactate utilization pathways. C. symbiosum has unique pathways for utilization of ethanolamine and citrate. R. gnavus has a unique pathway for utilization of D-glycerate.
Short-chain fatty acid (SCFA) production – R. gnavus and C. symbiosum possess both shared and strain-specific pathways for fermentation of various carbon sources and production of SCFAs. Both strains have complete pathways for production of formate, lactate, acetate and ethanol from pyruvate. R. gnavus has a unique pathway for production of propionate/propanol by fermentation of fucose and rhamnose. This pathway involves lactaldehyde reductase (FucO) and B12-dependent propanediol dehydratase (PduABCD), plus several other downstream enzymes for conversion of propionaldehyde to propionate and propanol. C. symbiosum has the complete pathway for conversion of acetyl-CoA to butyrate and butanol via crotonoyl-CoA and butyryl-CoA intermediates. Most of the butyrate synthesis pathway genes are organized into a single locus controlled by the Rex repressor. C. symbiosum also has the complete pathway for further conversion of lactate to propionate, which involves propionate CoA-transferase (Pct), lactoyl-CoA dehydratase (Lcd), acryloyl-CoA reductase (Acr) and is equipped with predicted propionate permease (PrpP*) for end product excretion. Thus, C. symbiosum and R. gnavus possess distinct pathways for production of propionate. (B) Amino acids
Both strains are capable of synthesizing most amino acids de novo. The histidine biosynthesis pathway in R. gnavus is composed of 11 committed enzymes, most of which are organized into a single operon controlled by a HisR repressor. The HisR regulon in R. gnavus also includes the predicted ABC-type His transporter HisXYZ* suggesting it can also salvage exogenous His. In contrast, the reconstructed HisR regulon in C. symbiosum includes the hutH-yuiF operon involved in His uptake and degradation; other His degradation genes were also identified in its genome. Thus, C. symbiosum is a predicted His auxotroph able to salvage exogenous His and use it both for protein synthesis and as a source of carbon, energy and nitrogen. Glutamate is synthesized from 2-oxoglutarate using the committed enzyme glutamate synthase GltBD, which is present in R. gnavus but not in C. symbiosum, suggesting the latter is a Glu auxotroph. In support of this prediction, there are two copies of the glutamate uptake transporter gene gltS in C. symbiosum but not in R. gnavus. Furthermore, C. symbiosum but not R. gnavus has a unique pathway for degradation of exogenous glutamate to produce butyrate.
In addition to the His and Glu uptake transporters described above, specific amino acid transporters were predicted for arginine, tyrosine, tryptophan, glutamine, branched chain amino acids (isoleucine/leucine/valine), alanine and methionine. These transporters are present in both R. gnavus and C. symbiosum suggesting that they can salvage these
amino acids from the environment, thus providing an alternative way to supply their biosynthetic needs. Tryptophan is a precursor of the neurotransmitter tryptamine. In contrast to C. symbiosum, R. gnavus contains a tryptophan decarboxylase ortholog (Trd) suggesting that is able to produce tryptamine. (C) B-vitamins
Thiamin (vitamin B1) is a common precursor of thiamin pyrophosphate (TPP), an essential cofactor of several ubiquitous enzymes in central carbon metabolism, most notably the pyruvate dehydrogenase complex. Neither R. gnavus nor C. symbiosum have a complete set of enzymes for de novo synthesis of TPP, and thus should be considered B1 auxotrophs. Both strains contain the thiamin salvage enzyme ThiN (thiamin pyrophosphokinase). The B1-specific transporter ThiT was identified in R. gnavus, but not in C. symbiosum. Alternatively, or in addition to thiamin salvage, both strains are capable of synthesizing TPP using exogenous thiazole via the dedicated salvage enzyme ThiM (hydroxyethylthiazole kinase). However, the respective thiazole transporter, ThiW was confidently identified only in C. symbiosum. Both strains have the complete machinery for production of the second biosynthetic intermediate (HMP-PP), joining it with thiazole-P, and downstream phosphorylation to yield the mature TPP cofactor. C. symbiosum, but not R. gnavus, also has the dedicated transporter ThiXYZ for the opportunistic salvage of the HMP intermediate.
Riboflavin (vitamin B2) is a precursor and biosynthetic intermediate of the essential redox cofactors flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD). The complete canonical de novo riboflavin biosynthetic pathway, from GTP and D-ribulose-5-phosphate, is present in C. symbiosum but absent in R. gnavus. This is one of the two primary distinctions between these two strains, which are otherwise quite similar in terms of B-vitamin proto- and auxotrophies. R. gnavus, unlike C. symbiosum, lacks fumarate reductase, one of the primary flavin-dependent enzymes in central carbon metabolism. Both strains have a committed riboflavin transporter RibU, plus an essential bi-functional enzyme RibF for converting riboflavin to FMN and FAD. A FMN-sensitive riboswitch is present to regulate RibU expression in both strains and de novo synthesis in C. symbiosum.
Niacin (vitamin B3), or nicotinic acid, as well as nicotinamide are converted to the most abundant redox cofactors, NAD and NADP; these cofactors are collectively involved in ~10% of all cellular biochemical transformations. Reconstruction of this subsystem in both R. gnavus and C. symbiosum includes both de novo synthesis of NAD from aspartate (prototrophy), and potential (optional) salvage of B3 and its amidated precursor. However, the B3 transporter NiaY is only present in C. symbiosum. In the absence of additional data, it is impossible to distinguish between two possible interpretations: (i) R. gnavus has a B3 transporter of another, yet unknown, type or (ii) the enzymes PncA and PncB are predominantly involved in recycling of intracellular nicotinamide, a well-known product of NAD degradation.
Pantothenate (vitamin B5) is an intermediate in the biosynthesis of coenzyme A (CoA), an indispensable cofactor in a variety of acyl transfer reactions, most importantly in fatty acid biosynthesis, degradation and the TCA cycle. Both strains lack components for de novo synthesis of pantothenate. However, auxotrophy is fully applicable only to R. gnavus, which has a complete set of downstream enzymes for the 5-step conversion of pantothenate to CoA. An interesting distinction of C. symbiosum is the absence of two
enzymes CoaB and CoaC. Pantetheine, a product of CoA degradation, can be phosphorylated (apparently via the second activity of pantothenate kinase), bypassing CoaBC-mediated steps, yielding a substrate for the next step catalyzed by CoaD. Therefore, C. symbiosum should be considered a pantetheine rather than a pantothenate auxotroph.
Pyridoxine (B6) is a precursor of the pyridoxal phosphate (PLP) cofactor that drives numerous transamination, decarboxylation and epimerization reactions of various amino acids. This pathway reveals a second major phenotypic distinction between R. gnavus and C. symbiosum. In contrast to the pattern with vitamin B2, C. symbiosum is a B6 auxotroph, whereas R. gnavus is a B6 prototroph. Consistent with this conclusion, only C. symbiosum contains a B6 transporter (PdxU).
Biotin (B7) is the critical cofactor in transcarboxylation reactions (including the initiation of fatty acid synthesis); it performs its function after covalent attachment to carrier protein(s) by the universally conserved and essential enzyme BirA (biotin ligase). Neither R. gnavus nor C. symbiosum has recognizable enzymes involved in the upstream steps of B7 biosynthesis, including the critical enzyme BioF (8-amino-7-oxononanoate synthase). Both strains contain the biotin transporter BioY, predicted to be under control of the same transcriptional repressor (the second function of the bifunctional protein BirA). However, both species also contain BioB, pointing to dethiobiotin as a potential alternative vitamin. Moreover, the presence of BioA and BioD, upstream of BioB, in C. symbiosum (but not in R. gnavus) points to two other possibilities; (i) salvage of 8-amino- 7-oxononanoate as yet another undocumented vitamin, and (ii) an alternative upstream pathway merging with the canonical route at the stage of 8-amino-7-oxononanoate. The first possibility appears more likely, and is indirectly supported by the presence of the second BioY paralog in C. symbiosum (but not in R. gnavus) in the chromosomal neighborhood of the bioA-bioB operon. The second possibility would make C. symbiosum a B7 prototroph.
Folate (B9) is a precursor of folate-derived cofactors DHF, THF and its polyglutamate derivatives that play a critical role in single carbon metabolism, including numerous methylation reactions. Both strains are obvious B9 auxotrophs containing only downstream enzymes and the folate transporter FolT. In C. symbiosum, this transporter and FolC are under regulation of the THF-responding riboswitch.
Cobalamin (B12) is involved in a limited number of very important reactions (e.g., methionine biosynthesis). Both analyzed strains have a complete B12 biosynthesis pathway. In addition, C. symbiosum (but not R. gnavus) contains the B12-specific transporter BtuFCD under regulation of a B12 riboswitch.
14
Metabolic features that distinguish gnotobiotic mice colonized with a microbiota from a 6-month-old healthy infant or an immature microbiota from a 6-month-old severely stunted/underweight infant. Metabolic profiles generated by targeted mass spectrometry of (A) acylcarnitines in serum (B) acylcarnitines in liver, and (C) amino acids in gastrocnemius muscle. Each metabolite’s concentration differs significantly between the two recipient groups of mice (P < 0.05; Student’s t-test).
B
A
C
Serum acylcarnitines
C6 C8C10
C10:1
C12
C10
-C12
-OH/
DCC14
C14:2
C16:2
C18:2
C18:2-
OH
Healthy (H)Stunted/underweight (Un)
Healthy (H)Stunted/underweight (Un)
Liver acylcarnitines
μm
ol/g
C10-O
H/C8-D
CC12 C14
C14-O
H/C12
-DC
C14:1
C14:2
C14:2-
OH/C12
:2-DC
16:1-C16
-OH/C
14-D
OH/
C
C C16:2-
C14
OH
:/
1-DC
C16:3-
C14
OH
:/C
2-DC
14:3-
DC
18:1-C18
-OH/C
16-D
OH/
C
C
C16:1-
DCC18
:3C20
:20
0.0005
0.0010
0.0015
mol
/g
Leucine/Isoleucine
0
0.2
0.4
0.6
0.8
1.0
H Un
Valine
0
0.05
0.10
0.15
0.20
H Un0
0.05
0.10
0.15
H Un
Methionine Alanine
0
1
2
3
4
H Un
Phenylalanine
0
0.02
0.04
0.06
0.08
H Un
Histidine
0
0.05
0.10
0.15
H Un
Gastrocnemius muscle amino acids
0
0.04
0.08
0.12
μM
Fig. S7.
15
Differences in brain metabolism in gnotobiotic recipients of 6-month-old healthy compared to severely stunted/underweight donor microbiota. (A) Metabolic profile of the cerebral cortex recovered at sacrifice. Columns represent individual animals. Rows represent metabolites whose concentrations are significantly different between the two treatment groups (P < 0.05 for all metabolites shown, Student’s t-test). (B) Concentrations of amino acids in the cerebral cortex of mice colonized with the healthy (H) or stunted/underweight (Un) donor microbiota (mean±SEM values plotted; P < 0.05 for all metabolites shown; Student’s t-test).
HistidineProlineGlycineArginineCitrullineC16:3-OH/C14:3-DCC8C20:1C16C14C18:1C18C18:2C12C18-OH/C16-DCC16:3-OH/C14:3-DCC20:1linoleoylC14:2Citrate
Healthy
Relative
Row min Row max
Stunted/underweight
Cerebral cortex
Am
ino
acid
sA
cylc
arni
tines
Acy
l coA
Arginine
0
0.05
0.10
0.15
0.20
0.25
Glycine
0
0.5
1.0
1.5
µg/m
olµg
/mol
Histidine
0
0.05
0.10
0.15
Ornithine
0
0.02
0.04
0.06
Proline
0
0.05
0.10
0.15
Citrulline
0
0.01
0.02
0.03
H Un
H Un H UnH Un
H Un H Un
B
AFig. S .
C5-
OH
/C3-
DC
C4-
DC
:Ci4
-DC
C7-
DC
C10
:3C
6-D
C:C
8-O
HC
8:1-
DC
C22
C5-
DC
C8:
1-O
H/C
6:1-
DC
C4-
OH
C18
C20
C8:
1C
10:2
C18
:1-D
C C3
C5
C18
-OH
/C16
-DC
C20
-OH
/C18
-DC
C5:
1C
2C
4:C
i4C
10-O
H/C
8-D
C C6
C12
-OH
/C10
-DC
C10
C10
:1C
14 C8
C12
C14
:2C
14:1
C18
:2C
16:2
C16
:1C
16-O
H/C
14-D
CC
14-O
H/C
12-D
CC
16C
18:1
C20
:4C
18:1
-OH
/C16
:1-D
CC
18:2
-OH
C12
:1C
16:1
-OH
/C14
:1-D
CC
14:1
-OH
Taxonomy
Clostridium bolteaeClostridium difficileEnterococcusClostridium perfringensClostridium neonataleClostridialesCoprobacillus cateniformisEubacterium limosumBifidobacterium longumBifidobacterium bifidumClostridium innocuumGordonibacter pamelaeaePaenibacillusClostridium ramosumNoneClostridiumClostridium nexileClostridium bartlettiiRuminococcus gnavusClostridium symbiosumClostridium butyricumFaecalibacterium prausnitziiClostridium paraputrificumDorea formicigeneransClostridium colinumClostridium disporicumFaecalibacteriumClostridium sartagoformeWeissella cibariaEggerthella lentaLactobacillus
A Acylcarnitines
Significantly enriched in healthy
Significantly enriched in stunted/underweight
Row max Row min
B Amino acids
Liver Gastrocnemius
Pro
Cit
Met
Gly
Glx
Ser
Val
Phe
Tyr
Arg
Leu/
IleA
sxO
rn
Ser
Ala
His
Gly
Cit
Ala
Leu/
IleH
isM
etV
alP
heTy
rO
rnP
roA
sxG
lxA
rg Taxonomy
Clostridium ramosumFaecalibacterium prausnitziiClostridium colinumClostridium disporicumClostridiumClostridium symbiosumClostridium nexileClostridium bartlettiiDorea formicigeneransRuminococcus gnavusClostridium butyricumClostridium paraputrificumWeissella cibariaClostridium sartagoformeFaecalibacteriumClostridium bolteaeClostridium neonataleClostridialesEnterococcusClostridium innocuumEubacterium limosumBifidobacterium longumClostridium difficileClostridium perfringensGordonibacter pamelaeaeCoprobacillus cateniformisBifidobacterium bifidumLactobacillusPaenibacillusEggerthella lentaNone
Liver Gastrocnemius
C Organic acids
Lact
ate
Pyr
uvat
eS
ucci
nate
Fum
arat
eM
alat
ea-
Ket
oglu
tara
teC
itrat
eLa
ctat
eP
yruv
ate
Suc
cina
teFu
mar
ate
Mal
ate
a-K
etog
luta
rate
Citr
ate
Taxonomy
PaenibacillusNoneClostridium ramosumClostridium sartagoformeFaecalibacteriumRuminococcus gnavusDorea formicigeneransFaecalibacterium prausnitziiClostridium paraputrificumClostridium disporicumClostridium colinumClostridium symbiosumClostridiumClostridium nexileClostridium bartlettiiClostridium butyricumWeissella cibariaEggerthella lentaLactobacillusCoprobacillus cateniformisEnterococcusGordonibacter pamelaeaeClostridium perfringensEubacterium limosumBifidobacterium longumClostridium innocuumClostridialesBifidobacterium bifidumClostridium neonataleClostridium difficileClostridium bolteae
Fig. S .
16
Spearman’s rank correlation coefficients for bacterial species relative abundances in the fecal microbiota versus concentrations of metabolites in gnotobiotic mouse recipients of healthy and severely stunted/underweight infant donor gut communities. Spearman correlations involving (A) acylcarnitines in serum, (B) amino acids in liver and gastrocnemius muscle, and (C) organic acids in liver and gastrocnemius muscle. Rows are hierarchically clustered based on pairwise distances calculated from Pearson correlation values. Bars to the right of each heatmap indicate significant species enrichment (P < 0.05; Student’s t-test) in the fecal microbiota of mice colonized with the healthy (purple) or stunted/underweight (yellow) donor’s gut community.
17
0 0.05 0.10 0.15 0.20
Bifidobacterium longumBifidobacterium
Faecalibacterium prausnitziiClostridiumClostridium
Enterococcus faecalisClostridium disporicumDorea formicigenerans
Eubacterium desmolansBacteroides vulgatus
EnterococcusRuminococcus gnavus
Clostridium innocuumRuminococcus gnavusRuminococcus gnavusRuminococcus gnavusRuminococcus gnavus
ClostridiumEggerthella
PaenibacillusEnterococcus
NonePediococcus pentosaceus
Weissella cibariaClostridium
Ruminococcus gnavusClostridium butyricum
Finegoldia magnaEnterococcus
Clostridium
Increase in Mean Squared Error
728206813701954362424737320380110334134383953437430227008833600504431692827241753709990186011227149421825063423972148365114121821483664348151
de novo 1261777325164384170652257565123820510966101383894448492
123456789101112131415161718192021222324252627282930
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
0 0.05 0.10 0.15 0.20
Bifidobacterium longumFaecalibacterium prausnitzii
BifidobacteriumClostridium
Faecalibacterium prausnitziiClostridium
Enterococcus faecalisClostridiales
Dorea formicigeneransClostridium disporicum
Bacteroides vulgatusEnterococcus
Ruminococcus gnavusClostridium innocuum
Ruminococcus gnavusRuminococcus gnavusRuminococcus gnavus
NonePaenibacillus
Clostridium sordelliiPediococcus
EggerthellaClostridium symbiosum
Pediococcus pentosaceusRuminococcus gnavus
Clostridium butyricumFinegoldia magnaWeissella cibaria
EnterococcusClostridium
Increase in Mean Squared Error
72820265871681370242473719543632038011033413447350943743024383953360050443169282724175370999018601122714942342397
de novo 12617214836624229843809761141218437868377325125756512382051096610643841383894448492
Rank OTU ID Taxonomic annotation
123456789
101112131415161718192021222324252627282930
Incr
easi
ng im
porta
nce
in a
ccur
acy
of m
odel
Rank OTU ID Taxonomic annotationA
B
Feature importance scores for OTUs, identified as weight gain and lean body mass gain discriminatory in mice, applied to members of the Malawian twin birth cohort. (A) The 30 most weight gain discriminatory OTUs (as shown in Fig. 2D), and (B) the 30 most lean body mass gain discriminatory OTUs (as shown in Fig. S2), ranked by increase in mean squared error (MSE) when these taxa are regressed against the WHZ scores of infants and children in the healthy Malawian cohort. Bars show mean±SD of the MSE.
Fig. S1 .
18
Supplementary Data Tables
Table S1. Summary of the ages, anthropometry, and time of fecal collection from Malawian cohorts. (A) Concordant healthy Malawian twins pairs/triplet sets. (B) Members of the iLiNS-DYAD-M birth cohort with fewer than 3 reported days of antibiotic consumption.
19
Table S2. Summary of bacterial V4-16S rRNA datasets from human donors. (A) Malawi Twin cohort. (B) iLiNS-DYAD-M cohort.
20
Table S3. Random Forests-derived sparse model of microbiota maturation in healthy Malawian infants/children. (A) Rank order of OTU feature importance. (B) Comparison of OTUs comprising the sparse Malawian and sparse Bangladeshi models.
21
Table S4. V4-16S rRNA-based characterization of fecal microbiota sampled from transplant recipients. (A) Maturity indices (microbiota age and MAZ scores) for microbiota selected for transplantation into gnotobiotic mice. (B) Summary of 16S rRNA datasets generated from the fecal microbiota of recipient mice in the initial screen of 19 donor microbiota. (C) Summary of 16S rRNA datasets generated from the follow-up transplant studies using microbiota from a 6-month-old healthy and a 6-month-old severely stunted/underweight donor.
22
Table S5. Composition of the Malawi-8 (M8) diet. (A) Summary of ingredients. (B) Nutritional analysis.
23
Table S6. Summary of growth phenotypes in recipients of the 19 microbiota from 6- and 18-month-old healthy and undernourished Malawian donors. (A) Weight gain. (B) Lean body mass gain and fat body mass gain. Spearman’s rank correlation of Shannon Index to lean body mass gain at day 28 following gavage of the donors’ microbiota is non-significant (P = 0.3275).
24
Table S7. Consumption of M8 chow in each donor group of gnotobiotic recipient mice from the screen of 19 donor microbiota.
25
Table S8. Rank order list of feature importance scores of OTUs in Random Forest models based on the growth phenotypes of gnotobiotic mice with 19 different transplanted microbiota from 6- and 18-month-old Malawian donors. (A) Weight gain. (B) Lean body mass gain.
26
Table S9. Spearman’s rank correlations of OTUs to mouse weight and lean body mass gain as well as to the chronological ages of members of concordant healthy Malawian twin pairs/triplet sets and the of iLiNS-DYAD-M cohort. (A) Weight gain. (B) Lean body mass gain.
27
Table S10. Rank order list of OTU feature importance scores in Random Forest models based on the femoral bone metrics of groups of recipient mice with 19 different transplanted microbiota from 6- and 18-month-old Malawian donors.
28
Table S11. List of species and their relative abundances used to construct the bargraphs shown in Fig. S4B.
29
Table S12. Direction and success of invasion in cohousing experiments involving HCH and UnCH mice.
30
Table S13. Reconstructed metabolic pathways for Ruminococcus gnavus TS8243C and Clostridium symbiosum TS8243C. (A) Overall number of genes in the reconstructed metabolic pathways of the two strains. The three major functional categories of studied genes are those involved in (i) carbohydrate and central carbon metabolism, (ii) amino acid metabolism and (iii) vitamin and cofactor biosynthesis. (B,C) Detailed lists of genes in R. gnavus TS8243C and C. symbiosum TS8243C, and their orthologs in the type strains, R. gnavus ATCC 29149 and C. symbiosum WAL-14163. Enzymes are identified by their enzyme commission (EC) numbers. Functional descriptions for carbohydrate active enzymes (CAZymes) include their assigned glycoside hydrolase (GH) families and their predicted localization (intracellular versus secreted as defined by SignalP). Transcriptional regulators and riboswitches that are predicted to control expression of individual genes from the reconstructed pathways (according to conservative propagation in the RegPrecise database) are listed in the last column. Central carbon metabolism genes controlled by the global redox-responsive regulator Rex are also indicated. Genes with novel predicted functions are marked with an asterisk. (D) Summarized metabolic capabilities. Predicted ability (+) or inability (-) to: (i) synthesize amino acids, (ii) synthesize B-vitamin related cofactors, (iii) utilize sugars, (iv) utilize other sources of carbon and energy, and (v) generate fermentation products.
31
Table S14. List of taxa and their relative abundances (%) used to construct the bargraphs shown in Fig. 4C.
32
Table S15. Metabolic phenotypes, as defined by targeted mass spectrometry, of gnotobiotic mice colonized with microbiota from the severely stunted/underweight infant donor or with the donor’s microbiota together with R. gnavus TS8243C and C. symbiosum TS8243C.
33
Table S16. Metabolic phenotypes of gnotobiotic mice colonized with a normally mature microbiota from a 6-month-old healthy (H) infant and a significantly immature microbiota from a 6-month-old severely stunted/underweight (Un) donor.
34
Table S17. Spearman’s rank correlations between levels of metabolites and the abundances of bacterial species in the fecal microbiota of mice colonized with the healthy versus the severely stunted/underweight infant microbiota. (A) Serum acylcarnitines. (B) Amino acids in liver and gastrocnemius muscle. (C) Organic acids assayed in liver and gastrocnemius muscle.
REFERENCES AND NOTES 1. C. G. Victora, L. Adair, C. Fall, P. C. Hallal, R. Martorell, L. Richter, H. S. Sachdev,
Maternal and Child Undernutrition Study Group, Maternal and child undernutrition: Consequences for adult health and human capital. Lancet 371, 340–357 (2008). Medline doi:10.1016/S0140-6736(07)61692-4
2. L. Gaayeb, J. B. Sarr, C. Cames, C. Pinçon, J. B. Hanon, M. O. Ndiath, M. Seck, F. Herbert, A. B. Sagna, A. M. Schacht, F. Remoue, G. Riveau, E. Hermann, Effects of malnutrition on children’s immunity to bacterial antigens in Northern Senegal. Am. J. Trop. Med. Hyg. 90, 566–573 (2014). Medline doi:10.4269/ajtmh.12-0657
3. M. Kosek, R. Haque, A. Lima, S. Babji, S. Shrestha, S. Qureshi, S. Amidou, E. Mduma, G. Lee, P. P. Yori, R. L. Guerrant, Z. Bhutta, C. Mason, G. Kang, M. Kabir, C. Amour, P. Bessong, A. Turab, J. Seidman, M. P. Olortegui, J. Quetz, D. Lang, J. Gratz, M. Miller, M. Gottlieb, MAL-ED network, Fecal markers of intestinal inflammation and permeability associated with the subsequent acquisition of linear growth deficits in infants. Am. J. Trop. Med. Hyg. 88, 390–396 (2013). Medline doi:10.4269/ajtmh.2012.12-0549
4. D. P. Waber, C. P. Bryce, J. M. Girard, M. Zichlin, G. M. Fitzmaurice, J. R. Galler, Impaired IQ and academic skills in adults who experienced moderate to severe infantile malnutrition: A 40-year study. Nutr. Neurosci. 17, 58–64 (2014). Medline doi:10.1179/1476830513Y.0000000061
5. R. E. Black, C. G. Victora, S. P. Walker, Z. A. Bhutta, P. Christian, M. de Onis, M. Ezzati, S. Grantham-McGregor, J. Katz, R. Martorell, R. Uauy, Maternal and Child Nutrition Study Group, Maternal and child undernutrition and overweight in low-income and middle-income countries. Lancet 382, 427–451 (2013). Medline doi:10.1016/S0140-6736(13)60937-X
6. S. A. Richard, B. J. McCormick, M. A. Miller, L. E. Caulfield, W. Checkley, MAL-ED Network Investigators, Modeling environmental influences on child growth in the MAL-ED cohort study: Opportunities and challenges. Clin. Infect. Dis. 59, S255–S260 (2014). Medline doi:10.1093/cid/ciu436
7. G. T. Keusch, D. M. Denno, R. E. Black, C. Duggan, R. L. Guerrant, J. V. Lavery, J. P. Nataro, I. H. Rosenberg, E. T. Ryan, P. I. Tarr, H. Ward, Z. A. Bhutta, H. Coovadia, A. Lima, B. Ramakrishna, A. K. Zaidi, D. C. Hay Burgess, T. Brewer, Environmental enteric dysfunction: Pathogenesis, diagnosis, and clinical consequences. Clin. Infect. Dis. 59, S207–S212 (2014). Medline
8. K. G. Dewey, The challenge of meeting nutrient needs of infants and young children during the period of complementary feeding: An evolutionary perspective. J. Nutr. 143, 2050–2054 (2013). Medline doi:10.3945/jn.113.182527
9. M. de Onis, C. Garza, C. G. Victora, A. W. Onyango, E. A. Frongillo, J. Martines, The WHO Multicentre Growth Reference Study: Planning, study design, and methodology. Food Nutr. Bull. 25, S15–S26 (2004). Medline doi:10.1177/15648265040251S103
10. M. de Onis, A. W. Onyango, J. Van den Broeck, W. C. Chumlea, R. Martorell, Measurement and standardization protocols for anthropometry used in the construction of a new international growth reference. Food Nutr. Bull. 25, S27–S36 (2004). Medline doi:10.1177/15648265040251S104
11. S. Subramanian, S. Huq, T. Yatsunenko, R. Haque, M. Mahfuz, M. A. Alam, A. Benezra, J. DeStefano, M. F. Meier, B. D. Muegge, M. J. Barratt, L. G. VanArendonk, Q. Zhang, M. A. Province, W. A. Petri Jr., T. Ahmed, J. I. Gordon, Persistent gut microbiota immaturity in malnourished Bangladeshi children. Nature 510, 417–421 (2014). Medline
12. L. Breiman, Random forests. Mach. Learn. 45, 5–32 (2001). doi:10.1023/A:1010933404324
13. M. I. Smith, T. Yatsunenko, M. J. Manary, I. Trehan, R. Mkakosya, J. Cheng, A. L. Kau, S. S. Rich, P. Concannon, J. C. Mychaleckyj, J. Liu, E. Houpt, J. V. Li, E. Holmes, J. Nicholson, D. Knights, L. K. Ursell, R. Knight, J. I. Gordon, Gut microbiomes of Malawian twin pairs discordant for kwashiorkor. Science 339, 548–554 (2013). Medline doi:10.1126/science.1229000
14. R. C. Edgar, Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461 (2010). Medline doi:10.1093/bioinformatics/btq461
15. Q. Wang, G. M. Garrity, J. M. Tiedje, J. R. Cole, Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ. Microbiol. 73, 5261–5267 (2007). Medline doi:10.1128/AEM.00062-07
16. P. Ashorn, L. Alho, U. Ashorn, Y. B. Cheung, K. G. Dewey, U. Harjunmaa, A. Lartey, M. Nkhoma, N. Phiri, J. Phuka, S. A. Vosti, M. Zeilani, K. Maleta, The impact of lipid-based nutrient supplement provision to pregnant women on newborn size in rural Malawi: A randomized controlled trial. Am. J. Clin. Nutr. 101, 387–397 (2015). Medline doi:10.3945/ajcn.114.088617
17. K. Sjögren, C. Engdahl, P. Henning, U. H. Lerner, V. Tremaroli, M. K. Lagerquist, F. Bäckhed, C. Ohlsson, The gut microbiota regulates bone mass in mice. J. Bone Miner. Res. 27, 1357–1367 (2012). Medline doi:10.1002/jbmr.1588
18. V. K. Ridaura, J. J. Faith, F. E. Rey, J. Cheng, A. E. Duncan, A. L. Kau, N. W. Griffin, V. Lombard, B. Henrissat, J. R. Bain, M. J. Muehlbauer, O. Ilkayeva, C. F. Semenkovich, K. Funai, D. K. Hayashi, B. J. Lyle, M. C. Martini, L. K. Ursell, J. C. Clemente, W. Van Treuren, W. A. Walters, R. Knight, C. B. Newgard, A. C. Heath, J. I. Gordon, Gut microbiota from twins discordant for obesity modulate metabolism in mice. Science 341, 1241214 (2013). Medline doi:10.1126/science.1241214
19. D. Knights, J. Kuczynski, E. S. Charlson, J. Zaneveld, M. C. Mozer, R. G. Collman, F. D. Bushman, R. Knight, S. T. Kelley, Bayesian community-wide culture-independent microbial source tracking. Nat. Methods 8, 761–763 (2011). Medline doi:10.1038/nmeth.1650
20. B. Cleary, I. L. Brito, K. Huang, D. Gevers, T. Shea, S. Young, E. J. Alm, Detection of low-abundance bacterial strains in metagenomic datasets by eigengenome partitioning. Nat. Biotechnol. 33, 1053–1060 (2015). Medline doi:10.1038/nbt.3329
21. A. L. Kau, J. D. Planer, J. Liu, S. Rao, T. Yatsunenko, I. Trehan, M. J. Manary, T. C. Liu, T. S. Stappenbeck, K. M. Maleta, P. Ashorn, K. G. Dewey, E. R. Houpt, C. S. Hsieh, J. I. Gordon, Functional characterization of IgA-targeted bacterial taxa from undernourished Malawian children that produce diet-dependent enteropathy. Sci. Transl. Med. 7, 276ra24 (2015). Medline doi:10.1126/scitranslmed.aaa4877
22. R. J. Crane, K. D. J. Jones, J. A. Berkley, Environmental enteric dysfunction: An overview. Food Nutr. Bull. 36, S76–S87 (2015). Medline doi:10.1177/15648265150361S113
23. K. G. Dewey, S. Adu-Afarwuah, Systematic review of the efficacy and effectiveness of complementary feeding interventions in developing countries. Matern. Child Nutr. 4, 24–85 (2008). Medline doi:10.1111/j.1740-8709.2007.00124.x
24. T. Magoč, S. L. Salzberg, FLASH: Fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957–2963 (2011). Medline doi:10.1093/bioinformatics/btr507
25. A. L. Goodman, G. Kallstrom, J. J. Faith, A. Reyes, A. Moore, G. Dantas, J. I. Gordon, Extensive personal human gut microbiota culture collections characterized and manipulated in gnotobiotic mice. Proc. Natl. Acad. Sci. U.S.A. 108, 6252–6257 (2011). Medline doi:10.1073/pnas.1102938108
26. C. T. Ferrara, P. Wang, E. C. Neto, R. D. Stevens, J. R. Bain, B. R. Wenner, O. R. Ilkayeva, M. P. Keller, D. A. Blasiole, C. Kendziorski, B. S. Yandell, C. B. Newgard, A. D. Attie, Genetic networks of liver metabolism revealed by integration of metabolic and transcriptional profiling. PLOS Genet. 4, e1000034 (2008). Medline doi:10.1371/journal.pgen.1000034
27. J. An, D. M. Muoio, M. Shiota, Y. Fujimoto, G. W. Cline, G. I. Shulman, T. Koves, R. D. Stevens, D. S. Millington, C. B. Newgard, Hepatic expression of malonyl-CoA decarboxylase reverses muscle, liver and whole-animal insulin resistance. Nat. Med. 10, 268–274 (2004).
28. M. V. Jensen, J. W. Joseph, O. Ilkayeva, S. Burgess, D. Lu, S. M. Ronnebaum, M. Odegaard, T. C. Becker, A. D. Sherry, C. B. Newgard, Compensatory responses to pyruvate carboxylase suppression in islet beta-cells. Preservation of glucose-stimulated insulin secretion. J. Biol. Chem. 281, 22342–22351 (2006). Medline doi:10.1074/jbc.M604350200
29. J. Deutsch, E. Grange, S. I. Rapoport, A. D. Purdon, Isolation and quantitation of long-chain acyl-coenzyme A esters in brain tissue by solid-phase extraction. Anal. Biochem. 220, 321–323 (1994). Medline doi:10.1006/abio.1994.1344
30. C. Magnes, F. M. Sinner, W. Regittnig, T. R. Pieber, LC/MS/MS method for quantitative determination of long-chain fatty acyl-CoAs. Anal. Chem. 77, 2889–2894 (2005). Medline doi:10.1021/ac048314i
31. P. E. Minkler, J. Kerner, S. T. Ingalls, C. L. Hoppel, Novel isolation procedure for short-, medium-, and long-chain acyl-coenzyme A esters from tissue. Anal. Biochem. 376, 275–276 (2008). Medline doi:10.1016/j.ab.2008.02.022
32. A. H. Merrill Jr., M. C. Sullards, J. C. Allegood, S. Kelly, E. Wang, Sphingolipidomics: High-throughput, structure-specific, and quantitative analysis of sphingolipids by liquid chromatography tandem mass spectrometry. Methods 36, 207–224 (2005). Medline doi:10.1016/j.ymeth.2005.01.009
33. B. Chevreux, T. Pfisterer, B. Drescher, A. J. Driesel, W. E. Müller, T. Wetter, S. Suhai, Using the miraEST assembler for reliable and automated mRNA transcript assembly and SNP detection in sequenced ESTs. Genome Res. 14, 1147–1159 (2004). Medline doi:10.1101/gr.1917404
34. T. Seemann, Prokka: Rapid prokaryotic genome annotation. Bioinformatics 30, 2068–2069 (2014). Medline doi:10.1093/bioinformatics/btu153
35. V. M. Markowitz, I. M. Chen, K. Palaniappan, K. Chu, E. Szeto, Y. Grechkin, A. Ratner, B. Jacob, J. Huang, P. Williams, M. Huntemann, I. Anderson, K. Mavromatis, N. N. Ivanova, N. C. Kyrpides, IMG: The Integrated Microbial Genomes database and comparative analysis system. Nucleic Acids Res. 40, D115–D122 (2012). Medline doi:10.1093/nar/gkr1044
36. R. Overbeek, R. Olson, G. D. Pusch, G. J. Olsen, J. J. Davis, T. Disz, R. A. Edwards, S. Gerdes, B. Parrello, M. Shukla, V. Vonstein, A. R. Wattam, F. Xia, R. Stevens, The SEED and the Rapid Annotation of microbial genomes using Subsystems Technology (RAST). Nucleic Acids Res. 42, D206–D214 (2014). Medline doi:10.1093/nar/gkt1226
37. D. A. Rodionov, Comparative genomic reconstruction of transcriptional regulatory networks in bacteria. Chem. Rev. 107, 3467–3497 (2007). Medline doi:10.1021/cr068309+
38. P. S. Novichkov, A. E. Kazakov, D. A. Ravcheev, S. A. Leyn, G. Y. Kovaleva, R. A. Sutormin, M. D. Kazanov, W. Riehl, A. P. Arkin, I. Dubchak, D. A. Rodionov, RegPrecise 3.0 – A resource for genome-scale exploration of transcriptional regulation in bacteria. BMC Genomics 14, 745 (2013). Medline doi:10.1186/1471-2164-14-745
39. D. A. Ravcheev, A. Godzik, A. L. Osterman, D. A. Rodionov, Polysaccharides utilization in human gut bacterium Bacteroides thetaiotaomicron: Comparative genomics reconstruction of metabolic and regulatory networks. BMC Genomics 14, 873 (2013). Medline doi:10.1186/1471-2164-14-873
40. D. A. Rodionov, I. A. Rodionova, X. Li, D. A. Ravcheev, Y. Tarasova, V. A. Portnoy, K. Zengler, A. L. Osterman, Transcriptional regulation of the carbohydrate utilization network in Thermotoga maritima. Front. Microbiol. 4, 244 (2013). Medline doi:10.3389/fmicb.2013.00244
41. D. A. Rodionov, C. Yang, X. Li, I. A. Rodionova, Y. Wang, A. Y. Obraztsova, O. P. Zagnitko, R. Overbeek, M. F. Romine, S. Reed, J. K. Fredrickson, K. H. Nealson, A. L. Osterman, Genomic encyclopedia of sugar utilization pathways in the
Shewanella genus. BMC Genomics 11, 494 (2010). Medline doi:10.1186/1471-2164-11-494