MICROBIOTA Effects of microbiota-directed foods in gnotobiotic … · Andrei L. Osterman16, Md Iqbal Hossain 17, Munirul Islam , Nuzhat Choudhury , Shafiqul Alam Sarker 17, Sayeeda

RESEARCH ARTICLE SUMMARY

MICROBIOTA

Effects of microbiota-directedfoods in gnotobiotic animals andundernourished childrenJeanette L Gehrig Siddarth Venkatesh Hao-Wei Chang Matthew C HibberdVanderlene L Kung Jiye Cheng Robert Y Chen Sathish SubramanianCarrie A Cowardin Martin F Meier David OrsquoDonnell Michael Talcott Larry D SpearsClay F Semenkovich Bernard Henrissat Richard J Giannone Robert L HettichOlga Ilkayeva Michael Muehlbauer Christopher B Newgard Christopher SawyerRichard D Head Dmitry A Rodionov Aleksandr A Arzamasov Semen A LeynAndrei L Osterman Md Iqbal Hossain Munirul Islam Nuzhat ChoudhuryShafiqul Alam Sarker Sayeeda Huq Imteaz Mahmud Ishita Mostafa Mustafa MahfuzMichael J Barratt Tahmeed Ahmed Jeffrey I Gordondagger

INTRODUCTION There is adimension topost-natal human development that involves assem-bly ofmicrobial communities in different bodyhabitats including the gut Children with acutemalnutritionhave impaireddevelopmentof theirgut microbiota leaving them with communitiesthat appear younger (more immature) than thoseof chronologically age-matched healthy individu-als Current therapeutic foods given to childrenwith acute malnutrition have not been formu-lated based on knowledge of how they affect thedevelopmental biology of the gut microbiotaMoreover they are largely ineffective in amelio-rating the long-term sequelae of malnutritionthat include persistent stunting neurodevelop-mental abnormalities and immunedysfunction

RATIONALE Repairing microbiota immatu-rity and determining the degree to which such

repair restores healthy growth requires iden-tification of microbial targets that are not onlybiomarkers of community assembly but alsomediators of various aspects of growth Iden-tifying ingredients in complementary foodsconsumed during the transition from exclusivemilk feeding to a fully weaned state that in-crease the representation and expressed bene-ficial functions of growth-promoting bacterialtaxa in the developing microbiota could pro-vide an effective affordable culturally accept-able and sustainable approach to treatment

RESULTSMetabolomic and proteomic analy-ses of serially collected plasma samples werecombined with metagenomic analyses of se-rially collected fecal samples from Bangladeshichildren with severe acute malnutrition (SAM)treated with standard therapy The results

provided a readout of their biological featuresas they transitioned from SAM to a state ofpersistentmoderate acutemalnutrition (MAM)with accompanying persistent microbiotaimmaturity Significant correlations were iden-tified between levels of plasma proteins an-thropometry plasma metabolites and therepresentation of bacteria in their microbiotaGnotobiotic mice were subsequently colonizedwith a defined consortium of bacterial strainsthat represent various phases of microbiotadevelopment in healthy Bangladeshi childrenAdministration of different combinations ofBangladeshi complementary food ingredientsto colonized mice and germ-free controls re-vealed diet-dependent increases in the abun-dance and changes in the metabolic activities

of targeted weaning-phasestrains as well as diet- andcolonization-dependentaugmentation of growth-promoting host signalingpathways Host andmicro-bial effects of microbiota-

directedcomplementary food (MDCF)prototypeswere subsequently examined in gnotobioticmice colonized with immature microbiota fromchildren with post-SAM MAM and in gnoto-biotic piglets colonized with a defined consor-tiumof targeted age- and growth-discriminatorytaxa A randomized double-blind study of stan-dard therapy versus various MDCF prototypesemerging from these preclinical models con-ducted in Bangladeshi children with MAMidentified a leadMDCF that increased levels ofbiomarkers and mediators of growth bone for-mation neurodevelopment and immune func-tion toward a state resembling healthy childrenUsing an approach inspired by statistical meth-ods applied to financialmarkets we show in theaccompanying paper by Raman et al that thislead MDCF was most effective in repairing themicrobiota

CONCLUSION These findings demonstratethe translatability of results obtained from pre-clinical gnotobiotic animal models to humansdirectly support the hypothesis that healthymicrobiota development is causally linked tohealthy growth illustrate an approach for treat-ing childhood undernutrition and with thecapacity to deliberately reconfigure immaturemicrobiota suggest a means to decipher howelements of the gut microbial community op-erate to regulate various host systems involvedin healthy growth

RESEARCH

Gehrig et al Science 365 139 (2019) 12 July 2019 1 of 1

The list of author affiliations is available in the full article onlineThese authors contributed equally to this workdaggerCorresponding author Email jgordonwustleduThis is an open-access article distributed under the termsof the Creative Commons Attribution license (httpcreativecommonsorglicensesby40) which permitsunrestricted use distribution and reproduction in anymedium provided the original work is properly citedCite this article as J L Gehrig et al Science 365 eaau4732(2019) DOI 101126scienceaau4732

Overview of therapeutic food discovery and testing The approach used for integratingpreclinical gnotobiotic animal models with human studies to understand the contributions ofperturbed gut microbiota development to childhood malnutrition and to identify MDCFs

ON OUR WEBSITE

Read the full articleat httpdxdoiorg101126scienceaau4732

Corrected 11 July 2019 See full text on A

ugust 30 2020

httpsciencesciencemagorg

Dow

nloaded from

RESEARCH ARTICLE

MICROBIOTA

Effects of microbiota-directedfoods in gnotobiotic animals andundernourished childrenJeanette L Gehrig12 Siddarth Venkatesh12 Hao-Wei Chang12Matthew C Hibberd12 Vanderlene L Kung123 Jiye Cheng12 Robert Y Chen12Sathish Subramanian12dagger Carrie A Cowardin12 Martin F Meier12 David OrsquoDonnell12Michael Talcott4 Larry D Spears5 Clay F Semenkovich5 Bernard Henrissat67Richard J Giannone8 Robert L Hettich8 Olga Ilkayeva910 Michael Muehlbauer910Christopher B Newgard9101112 Christopher Sawyer1314 Richard D Head1314Dmitry A Rodionov1516 Aleksandr A Arzamasov1516 Semen A Leyn1516Andrei L Osterman16 Md Iqbal Hossain17 Munirul Islam17 Nuzhat Choudhury17Shafiqul Alam Sarker17 Sayeeda Huq17 Imteaz Mahmud17 Ishita Mostafa17Mustafa Mahfuz17 Michael J Barratt12 Tahmeed Ahmed17 Jeffrey I Gordon12Dagger

To examine the contributions of impaired gut microbial community development tochildhood undernutrition we combined metabolomic and proteomic analyses of plasmasamples with metagenomic analyses of fecal samples to characterize the biological stateof Bangladeshi children with severe acute malnutrition (SAM) as they transitioned afterstandard treatment to moderate acute malnutrition (MAM) with persistent microbiotaimmaturity Host and microbial effects of microbiota-directed complementary food(MDCF) prototypes targeting weaning-phase bacterial taxa underrepresented in SAM andMAM microbiota were characterized in gnotobiotic mice and gnotobiotic piglets colonizedwith age- and growth-discriminatory bacteria A randomized double-blind controlledfeeding study identified a lead MDCF that changes the abundances of targeted bacteriaand increases plasma biomarkers and mediators of growth bone formationneurodevelopment and immune function in children with MAM

Evidence is accumulating that disruptionof ldquonormalrdquo gut community (microbiota)development may contribute to the path-ogenesis of undernutrition Using culture-independent surveys bacterial membership

has beendefined in fecal samples collectedmonth-ly during the first 2 postnatal years from healthymembers of a birth cohort living in an urbanslum (Mirpur) in Dhaka Bangladesh (1 2) Ap-plying machine learning [Random Forests (RF)]to the resulting 16S ribosomal DNA (rDNA) data-set yielded a ldquosparserdquomodel composed of themostage-discriminatory bacterial strains changes inthe relative abundances of these organisms de-scribed a program of normal microbiota develop-ment (2) ThisRF-derivedmodelwas subsequentlyused to characterize fecal samples collected from

Bangladeshi children with severe acute mal-nutrition [SAM defined as a weight-for-heightz-score (WHZ) gt3 standard deviations below themedian for a World Health Organization (WHO)reference healthy growth cohort (3)] The resultsrevealed gut communities that resembled thoseof healthy children who were chronologicallyyounger Thismicrobiota ldquoimmaturityrdquowasmorepronounced in children with SAM as comparedwith those with moderate acute malnutrition(MAM WHZ score between ndash2 and ndash3) and wasnot repaired in a clinical study that tested theeffects of two therapeutic foods (2)Impaired microbiota development has also

been documented in undernourished Malawianchildren (4) To examine the functional impor-tance of this impairment microbial communi-

ties from healthy and stunted or underweight6- and 18-month-old Malawian children weretransplanted into groups of recently weanedgerm-free mice fed a diet representative of thatconsumed by the human donor population Theresults disclosed that compared with mice col-onized with normally maturing microbiota fromthe healthy donors animals harboring imma-ture microbiota exhibited reduced rates of leanbody mass gain alterations in bone growth andmetabolic abnormalities (4) These studies pro-vided preclinical evidence for a causal link be-tween microbiota immaturity and undernutritionthey also revealed that a subset of the age-discriminatory strains is growth-discriminatoryMoreover a cultured consortium of these age-and growth-discriminatory taxa amelioratedthe impaired growth phenotype transmitted torecipient gnotobiotic mice by an immature mi-crobiota (4)One question arising from these observations

is how do we design optimal foods that steer amicrobiota into an age-appropriate and healthystate Breastfeeding plays a major role in reduc-ing childhood malnutrition As such WHO andthe United Nations Childrenrsquos Fund (UNICEF)recommend exclusive breastfeeding for the first6 months of postnatal life and continued breast-feeding after the introduction of complementaryfoods up to 24 months of age (5) Suboptimalcomplementary feeding practices are importantcontributors to malnutrition in children less than2 years of age (6) However current complemen-tary feeding recommendations are not based onknowledge of how foods affect the develop-mental biology of the gut microbiota during theweaning process Together these observationsraise the question Do certain complementaryfood ingredients or combinations of ingredientshave the ability to selectively increase the rep-resentation and expressed beneficial functions ofage- and growth-discriminatory strains deficientin SAM- or MAM-associated microbiota If theanswer is yes then prescribed feeding of theseingredients could help ldquorepairrdquo or prevent de-velopment of microbiota immaturity in chil-dren with potentially long-lived health-promotingeffectsHere we describe a process for identifying

microbiota-directed complementary foods (MDCFs)designed to treat children with acute malnutri-tion We first characterized gut microbial com-munity and host responses over the course of12 months in Bangladeshi children who weretreated for SAM with one of three conventionaltherapeutic foods Measurement of the levels of

RESEARCH

Gehrig et al Science 365 eaau4732 (2019) 12 July 2019 1 of 12

1Edison Family Center for Genome Sciences and Systems Biology Washington University School of Medicine St Louis MO 63110 USA 2Center for Gut Microbiome and Nutrition ResearchWashington University School of Medicine St Louis MO 63110 USA 3Department of Pathology and Immunology Washington University School of Medicine St Louis MO 63110 USA 4Divisionof Comparative Medicine Washington University St Louis MO 63110 USA 5Department of Medicine Washington University School of Medicine St Louis MO 63110 USA 6Architecture etFonction des Macromoleacutecules Biologiques Centre National de la Recherche Scientifique and Aix-Marseille Universiteacute 13288 Marseille cedex 9 France 7Department of Biological Sciences KingAbdulaziz University Jeddah Saudi Arabia 8Chemical Sciences Division Oak Ridge National Laboratory Oak Ridge TN 37830 USA 9Sarah W Stedman Nutrition and Metabolism Center DukeUniversity Medical Center Durham NC 27710 USA 10Duke Molecular Physiology Institute Duke University Medical Center Durham NC 27710 USA 11Department of Pharmacology and CancerBiology Duke University Medical Center Durham NC 27710 USA 12Department of Medicine Duke University Medical Center Durham NC 27710 USA 13Department of Genetics WashingtonUniversity School of Medicine St Louis MO 63110 USA 14Genome Technology Access Center Washington University School of Medicine St Louis MO 63110 USA 15A A Kharkevich Institutefor Information Transmission Problems Russian Academy of Sciences Moscow 127994 Russia 16Infectious and Inflammatory Disease Center Sanford Burnham Prebys Medical DiscoveryInstitute La Jolla CA 92037 USA 17International Centre for Diarrhoeal Disease Research Bangladesh (icddrb) Dhaka 1212 BangladeshThese authors contributed equally to this work daggerPresent address Department of Medicine Massachusetts General Hospital Boston MA 02114 USADaggerCorresponding author Email jgordonwustledu


ugust 30 2020


Dow

nloaded from

1305 human plasma proteinsmdashincluding regu-lators and effectors of physiologic metabolic andimmune functionsmdashcombined with mass spec-trometric profiling of plasma metabolites andculture-independent analyses of serially col-lected fecal samples provided a ldquoreadoutrdquo of thebiological features of these children as they pro-gressed from SAM to a state of incomplete recov-ery (post-SAMMAM) with persistent microbiotaimmaturity This readout included correlationsbetween plasma proteins anthropometry plas-ma metabolites and the representation of age-discriminatory members of their microbiota Wethen screened complementary foods in gnotobi-otic mice colonized with a consortium of bacte-rial strains that had been cultured from childrenliving in Mirpur to identify ingredients thatpromote the representation of constituent age-discriminatory strains that are underrepresentedin the setting of acutemalnutrition Subsequentlya representative microbiota from a child withpost-SAM MAM was transplanted into gnoto-biotic mice Recipient animals were fed a dietresembling that consumed by children inMirpurbut supplemented with ingredients identified inthe screen in order to establish whether one ormore of these MDCF formulations could repaira microbiota from a subject who had alreadyreceived conventional therapy Lead formu-lations were subsequently tested in gnotobioticpiglets colonized with a defined consortium ofage- and growth-discriminatory strains to testtheir biological effects in a host species that isphysiologically and metabolically more similarto humans than mice Last three MDCF proto-types were administered to children with MAMand their effects on the microbiota and host bio-logical state were determined

Effects of conventional therapeuticfoods on the biological stateof children with SAM

A total of 343 Bangladeshi children aged 6 to36 months with SAM were enrolled in a multi-center randomized double-blind ldquononinferiorityrdquostudy designed to compare two locally producedtherapeutic foods (supplementarymaterialsmate-rials andmethods)with a commercially availableready-to-use therapeutic food (RUTF) (7) (studydesign is provided in Fig 1A and the composi-tions of these therapeutic foods are available intable S1A) Children received standard manage-ment for SAMduring the acute stabilization phaseof in-hospital treatment including a short courseof antibiotics Eligible children were then ran-domized to one of the three therapeutic foodarms (~200 kcalkgday mean duration 161 plusmn103 days) (table S1B) Children were dischargedafter meeting criteria described in the supple-mentary materials materials and methods In asubset of 54 children fecal sampleswere collectedat enrollment [age 152 plusmn 51months (meanplusmn SD)]before randomization twice during treatmentwith a therapeutic food and at regular intervalsup to 12months after discharge (Fig 1A clinicalmetadata is available in table S1B) Blood samples(plasma) were also obtained at enrollment dis-

charge and 6 months after discharge for tar-geted mass spectrometry (MS)ndashbased metabolicprofiling a sufficient quantity of blood was ob-tained fromeight children at all three time pointsfor aptamer-based proteomics analysis (8ndash10)Of these children 44 had MAM at 12 monthsof followup None of the therapeutic foods pro-duced a significant effect on their severe stunt-ing [height-for-age z-score (HAZ)] (Fig 1B andtable S1B)

Metabolic phenotypes

Targeted MS of plasma samples obtained at en-rollment revealed high levels of ketones non-esterified fatty acids (NEFA) andmedium to longeven-chain acylcarnitines (Fig 2 A and B andtable S2) which is consistent with the knownacute malnutrition-induced lipolytic responsethat raises circulating fatty acids and activatesfatty acid oxidation (11) By discharge this meta-bolic feature had normalized whereas levels of anumber of amino acids had increased signifi-cantly including the gluconeogenic amino acidalanine the branched-chain amino acids leucineisoleucine and valine plus products of branched-chain amino acid metabolism [C3 (propionyl)-carnitine and their ketoacids] (Fig 2 A to C)These findings suggest that the increased proteinprovided by the therapeutic foods prompted aswitch from fatty acid to amino acid oxidationleading to repletion of fat depots increases in

plasma leptin (Fig 2A) and weight gain (tableS1B) However 6 months after treatment multi-ple plasma amino acids and their metaboliteshad declined to levels comparable with those atadmission whereas fatty acids and fatty acidndashderived metabolites remained at similar concen-trations to those observed at discharge (Fig 2 Ato C) Insulin-like growth factor 1 (IGF-1) levelsdid not change significantly during this period(Fig 2A) potentially explaining the absence of asignature of pronounced lipolysis that had beenobserved at enrollment Although the suppressionof lipolysis at 6months after discharge suggests asustained effect of nutritional resuscitation thefall in essential amino acids and the lower levelof IGF-1 compared with that found in similarlyaged healthy children from the same community(445 versus 694 ngmL P = 002 t test) maycontribute to the observed failure to achieve catch-up growth

The plasma proteome

Significant correlations between levels of plasmaproteins anthropometric indices plasma metab-olites and host signaling pathways regulatingkey facets of growth are described in the supple-mentary text results (table S3 A and B)mdashforexample components of the growth hormone(GH)ndashIGF axis including soluble growth hor-mone receptor (also known as growth hormonebinding protein) multiple IGF binding proteins


Fig 1 Longitudinal study of Bangladeshi children with SAM treated with therapeutic foods(A) Study design (B) Anthropometry and MAZ scores Gray bars represent the three time pointsat which blood samples were collected (C) Summary of MAZ scores for children with SAM(WHZ lt ndash3 n = 96 fecal samples) and subsequently (post-SAM) MAM (WHZ gt ndash3 and ltndash2 n =151 fecal samples) plus healthy children aged 6 to 24 months living in the same area in which theSAM study was conducted (n = 450 fecal samples) Mean values for WHZ WAZ HAZ andMAZ plusmn SEM are plotted on the x axes of (B) and (C) P lt 00001 (one-way ANOVA followedby Tukeyrsquos multiple comparisons test)

RESEARCH | RESEARCH ARTICLE


ugust 30 2020


Dow

nloaded from

(IGFBPs) and regulators of IGFBP turnover (themetalloprotease pappalysin-1 and its inhibitorstanniocalcin-1)

The gut microbiome

A sparse RF-derived model of normal gut micro-biota development comprising 30 bacterial taxa[operational taxonomic units (OTUs)] and ob-tained from25 healthymembers of a birth cohortliving in Mirpur (table S4 A to C) was appliedto bacterial V4-16S rDNA datasets generatedfrom fecal samples serially collected from thechildren in the SAMstudy (n= 539 samples) Thismodel allowed us to define microbiota-for-agez-scores (MAZ) as a function of treatment armand time [93 plusmn 37 sampleschild (mean plusmn SD)]The MAZ score measures the deviation in de-velopment of a childrsquos microbiota from that ofchronologically age-matched reference healthychildren on the basis of the representation ofthe ensemble of age-discriminatory strains con-tained in the RF-derived model (2) Significantmicrobiota immaturity was apparent in the SAMand post-SAM MAM groups (Fig 1C and tableS5A) Moreover MAZ-scores in this SAM cohortwere significantly correlated with WHZ HAZ

and WAZ [Pearson correlation coefficient (r) =016 P = 00004 r = 013 P = 0003 and r = 010P = 002 respectively] The MAZ score was notdifferent at discharge but improved 1month later(P = 00051 versus admission Mann-Whitneytest) This improvement could reflect increaseddietary diversity reduced antibiotic usage (tableS1B) andor other factors associatedwith return-ing to the home environment MAZ-scores didnot change significantly thereafter (Fig 1B)A number of the age-discriminatory strains

were significantly correlated with anthropo-metric indices as well as with plasma proteinsinvolved in biological processes that mediategrowth We also identified significant negativecorrelations between these taxa and mediatorsof systemic inflammation and anorexiacachexiaBifidobacterium longum (OTU 559527) had thegreatest number of significant correlations (114)[table S3C further discussion is available in sup-plementary text results]The effects of the therapeutic food interven-

tions on the representation of metabolic path-ways in the gut microbiome were defined byshotgun sequencing of 331 fecal DNA samplesobtained from 30 members of the Mirpur birth

cohort with consistently healthy anthropome-try and 15 of the 54 children enrolled in theSAM study (table S5B) these 15 children wereselected according to their age (12 to 18 months)and that we had corresponding plasma metab-olomic and proteomic datasets for at least two ofthe three time points sampled The abundancesof microbial genes that mapped to pathways inthe microbial communities SEED (mcSEED) data-base (12)mdashrelated to metabolism of amino acidscarbohydrates fermentation products and Bvitamins and related cofactorsmdashwere first de-fined in healthy children sampled monthly frombirth to 2 years of age A set of age-discriminatorymetabolic pathways (mcSEED ldquosubsystemsrdquo orpathway modules) was identified The resultingsparse RF-derived model (fig S1 A and B andmaterials and methods) allowed us to assign astate of development (functional age or ldquomatu-rityrdquo) to the fecal microbiomes of the 15 childrentreated for SAMRelative functionalmaturitywassignificantly correlatedwithMAZWHZ andWAZscores during the course of the trial (Pearson rand P values areMAZ r = 055 P lt 00001WHZr = 030 P = 00011 WAZ r = 023 P = 0013)At enrollment and just before administration of


Fig 2 Metabolic features of children with SAM before and after treatment (A to C) Levels of (A) standard clinical metabolites and selectedhormones (B) acylcarnitines and (C) amino acids and ketoacids in plasma collected from children at enrollment (Fig 1A B1 blood sample)discharge (Fig 1A B2 sample) and 6 months after discharge (Fig 1A B3 sample) Abbreviations for branched chain ketoacids in (C) are KICa-ketoisocaproate KIV a-ketoisovalerate and KMV a-keto-b-methylvalerate Mean values plusmn SEM are plotted P lt 005 P lt 001 P lt 0001P lt 00001 (paired t test followed by FDR correction)



ugust 30 2020


Dow

nloaded from

therapeutic foods children with SAM had moreimmaturemicrobiomes [one-way analysis of var-iance (ANOVA) P = 00002 Dunnettrsquos multiplecomparisons test for healthy versus SAMadjustedP values at the two time points 0027 and 00001respectively] There was a statistically significantimprovement in functionalmaturity from initia-tion of therapeutic food treatment to dischargeand at 1 and 6 months after discharge (Tukeyrsquosmultiple comparisons test adjusted P values =0039 00028 and 0025 respectively) Howeverthis improvement was not sustained at later timepoints (fig S1D) Comparing the relative abun-dances of the 30most age-discriminatory path-ways at six time points revealed that the SAMmicrobiome had significantly reduced represen-tation of (i) amino acid metabolic pathwaysincluding those involved in isoleucine leucinevaline biosynthesis and uptake (ii) several car-bohydrate utilization pathways (arabinose andarabinosides rhamnose and rhamnogalacturonanand sialic acid) and (iii) multiple pathwaysinvolved in B-vitamin metabolism includingldquoniacinNADP (nicotinamide adenine dinucleo-tide phosphate) biosynthesisrdquo (fig S1E and tableS5C) The observed underrepresentation of age-discriminatory OTUs and metabolic pathwaysin the gut communities of children with post-SAMMAM provided the rationale for developinga pipeline to test complementary food ingredientsfor their ability to repair this immaturity

Screening complementaryfood ingredients

Nine age-discriminatory bacterial strains werecultured from the fecal microbiota of threehealthy children aged 6 to 23 months who livedin Mirpur and genomes of these isolates weresequenced (table S6 A and C) Seven of thenine isolates had V4-16S rDNA sequences thatcorresponded to age-discriminatoryOTUswhoserepresentation is associatedwith the period of com-plementary food consumption (ldquoweaning-phaserdquoOTUs) (fig S2A) whereas two Bifidobacteriumlongum subsp infantis and Bifidobacteriumbreve are most prominent during the period ofexclusive predominant milk feeding (fig S2A)(13) OTUs representing seven of the nine cul-tured strains were significantly depleted in thefecal microbiota of Bangladeshi children withSAM before treatment (table S7 and fig S3) Sevenadditional age-discriminatory strains were cul-tured from the immature fecal microbiota of a24-month-old childwith SAMenrolled in the samestudy as the subcohort shown in Fig 1 (table S6A and C) Together the consortium of 16 strainsrepresented OTUs that directly matched 656 plusmn228 (mean plusmn SD) of V4-16S rDNA sequencingreads generated from 1039 fecal samples col-lected from 53 healthy members of the Mirpurbirth cohort during their first 2 postnatal yearsand 742 plusmn 252 of the reads produced fromfecal samples obtained from 38 children withSAM (table S7)To identify complementary foods that selec-

tively increase the representation of weaning-phase age-discriminatory strains deficient in

immature SAM-associated microbiota we col-onized 5-week-old germ-free C57Bl6J micewith the consortium of cultured sequenced bac-terial strains After colonization an 8-week pe-riod of diet ldquooscillationsrdquo was initiated (fig S2B)We incorporated 12 complementary food ingre-dients commonly consumed in Mirpur (6) into14 different diets using a random sampling strat-egy (table S8 A to E and materials and meth-ods) The composition of these complementaryfood combinations (CFCs) and their order of ad-ministration tomice were based on considerationsdescribed in the legend to fig S2 B and CSpearmanrsquos rank correlation coefficients were

calculated between the relative abundances ofthe 14 bacterial strains that colonized mice andlevels of complementary food ingredients in the14 CFCs tested (fig S2D and table S9A) Chickpeaand banana had strong positive correlationswith the greatest number of strains representingweaning-phase age-discriminatoryOTUs Tilapiahad a narrower range of significant positive ef-fects (fig S2D) Chickpea banana and tilapiaalso had significant negative correlations withlevels of the preweaning milk-adapted B longumsubsp infantis isolate A sobering observation wasthat a number of complementary food ingre-dients typically represented in diets consumedby 18-month-old children living in Mirpur hadsignificant negative correlations with six of theweaning-phase age-discriminatory strains in-cluding rice milk powder potato spinach andsweet pumpkin (fig S2D) Rice gruel withmilk isthe most common first complementary foodgiven to Bangladeshi children (14) Moreoveregg which is included in a number of regimensfor nutritional rehabilitation of childrenwith acutemalnutrition (15) was negatively correlated withthe abundance of two weaning-phase strainsDorea formicigenerans and Blautia luti

Testing an initial MDCF prototype

Khichuri-halwa (KH) is a therapeutic food com-monly administered together with milk-suji (MS)toMirpur children with SAM A previous studydocumented the inability of this interventionto repair gut microbiota immaturity (2) We pre-pared a diet thatmimickedMS andKH (MSKH)(table S8 D and E) 7 of its 16 ingredients arecommonly consumed complementary foods thathad little if any effect on the representation ofweaning-phase age-discriminatory strains (ricered lentils potato pumpkin spinach whole-wheatflour and powdered milk) (fig S2D) The effectsof MSKH on members of the 14-member con-sortium and the host were compared with thoseproduced by an initial MDCF prototype that con-tained chickpeas banana and tilapia (table S9B)Five-week-old germ-free C57Bl6J mice colonizedwith the consortiumweremonotonously fed eitherof the two diets ad libitum for 25 days

Microbial community responses

Community profiling by means of short readshotgun sequencing (COPRO-seq) of cecal DNArevealed that compared with MSKH consump-tion of the MDCF prototype resulted in signifi-

cantly higher relative abundances of a numberof weaning-phase age-discriminatory taxa in-cluding Faecalibacterium prausnitzii Dorealongicatena and B luti (P lt 001 Mann-Whitneytest) (Fig 3A and table S9B) This prototype didnot promote the fitness of the SAM donor-derived strains with the exception ofEscherichiafergusoniiWe used targeted MS to quantify cecal levels

of carbohydrates short-chain fatty acids plusamino acids and their catabolites (table S10 Ato D) Germ-free animals served as referencecontrols to define levels of cecal nutrients thatby inference would be available for bacterialutilization in the different diet contexts Therewere several noteworthy findings (i) Levels ofbutyrate and succinate were significantly higher incolonized animals consuming MDCF comparedwith MSKH (Fig 3B and table S10B) (ii) Therewere no statistically significant diet-associated dif-ferences in levels of any of the amino acids mea-sured in germ-free animals but when comparedwith their colonized MSKHndashfed counterpartscolonized MDCF-consuming animals had sig-nificantly elevated cecal levels of six amino acidsclassified as essential in humans (the threebranched-chain amino acids leucine isoleucineand valine plus phenylalanine and tryptophan)(Fig 3C and table S10C) And (iii) two tryptophan-derivedmicrobialmetabolites that play importantroles in suppressing inflammation and are neuro-protective 3-hydroxyanthranillic acid (3-HAA)and indole-3-lactic acid (16ndash21) were signifi-cantly elevated in colonized animals fed MDCFcompared with their MSKHndashtreated counter-parts (table S10D)Findings from RNA-sequencing (RNA-seq)

analysis of the transcriptional responses of com-munitymembers to the two diets based onKyotoEncyclopedia of Genes and Genomes (KEGG)ndashand mcSEED-derived annotations of the 40735predicted protein-coding genes present in con-sortium members are described in tables S9Cand S11 A to C and supplementary text resultsand in silico predictions of their ability to produceuse andor share nutrients are provided in tableS6 D and E For example community-level anal-ysis revealed specificmembersmanifestedMDCF-associated increases in expression of genes involvedin biosynthesis of the essential amino acids in-cludingbranched-chainaminoacids (Ruminococcusobeum andRuminococcus torques) and generationof aromatic amino acid metabolites (R obeumR torques and F prausnitzii) (table S11C ii)

Host effects

Serum levels of IGF-1 were significantly higher incolonized mice that consumed the initial MDCFprototype compared with those that consumedMSKH This effect was diet- and colonization-dependent with germ-free animals exhibitingsignificantly lower levels of IGF-1 in both dietcontexts (Fig 3D) Serum insulin levels were alsohigher in colonized animals that consumedMDCFcompared with MSKH [8007 plusmn 3029 ngmL(mean plusmn SD) versus 5187 plusmn 1351 ngmL respec-tively P = 006 unpaired t test]




ugust 30 2020


Dow

nloaded from


Fig 3 Comparison of microbial community and host effects of an initialMDCFprototype versus MSKH Separate groups of germ-free mice or ani-mals colonized with the defined consortium of 14 bacterial strains were fed thetwo diets monotonously for 25 days after which time they were euthanized andcecal contents were analyzed (A) The relative abundances of strains in the cecalmicrobiota of colonized mice Mean values plusmn SD shown (B and C) Diet- andcolonization-dependent effects on (B) cecal levels of short chain fatty acids and(C) essential amino acids plus the tryptophan metabolite indole 3-lactic acidEach dot represents a sample from a mouse in the indicated treatment groupMean values plusmn SD are shown P lt 0001 P lt 00001 [2-way ANOVA fol-lowed by Tukeyrsquosmultiple comparisons test for (A) to (C)] (D) Diet- and colonization-dependent effects on serum IGF-1 levels (E) Effects of diet on levels of liver

proteins involved in IGF-1 signaling and IGF-1 production Levels of phospho-rylated proteins were normalized to the total amount of the correspondingnonphosphorylated protein or to glyceraldehyde-3-phosphate dehydrogenase(GAPDH) (F) Effect of diet and colonization status on the cortical thickness offemoral bone (G) Effects of diet in colonized gnotobiotic mice on branched-chain amino acids in serum and acylcarnitines in muscle and liver [C5-OHC3are isobars that are not resolved through flow injection MSMS C5-OH is amix of 3-hydroxy-2-methylbutyryl carnitine (derived from the classicalisoleucine catabolic intermediate 3-hydroxy-2-methylbutyryl CoA) and3-hydroxyisovaleryl carnitine (a noncanonical leucine metabolite)] For (D)to (G) mean values plusmn SD are shown ns not significant P lt 005P lt 001 P lt 00001 for (D) to (G) (Mann-Whitney test)



ugust 30 2020


Dow

nloaded from

IGF-1 binding to its receptor tyrosine kinaseIGF-1R affects a variety of signal transductionpathways including one involving the serinethreonine kinase AktPKB phosphatidylinositol-3kinase (PI3K) and the mammalian target ofrapamycin (mTOR) Absorption of several aminoacids from the gutmdashnotably branched-chainamino acids and tryptophanmdashleads to activationof mTOR (22) Colonized animals fed MDCF hadsignificantlyhigher levels of hepatic phosphoSer473-Akt which is consistent with activation of Aktby IGF-1 signaling through the PI-3K pathway(Fig 3E) Levels of phosphondashAMPK (5prime adeno-sine monophosphate-activated protein kinase)were not significantly affected by diet (Fig 3E)suggesting that Akt phosphorylation is not causedindirectly by altered hepatic energy status Phos-phorylation of hepatic Jak 2 (Tyr10071008) andmTOR (Ser2448) which are involved in IGF-1production was significantly increased in colon-ized mice consuming MDCF (Fig 3E) whereasphosphorylation of STAT5 also implicated inIGF-1 production was not significantly alteredPrevious studies of adult germ-free mice re-

ported increases in serum IGF-1 after their col-onization with gut microbiota from conventionallyraised mice increased IGF-1 levels were also as-sociated with increased bone formation (23 24)Micro-computed tomography (mCT) of mousefemurs revealed a significant increase in femoralcortical bone area in MDCF-fed animals theeffect was both diet- and microbiota-dependent(Fig 3F)We used targeted MS to quantify levels of

amino acids acylndashcoenzyme As (acylCoAs) acyl-carnitines and organic acids in serum liverand gastrocnemius muscle (table S12) Productsof nonoxidative metabolism of glucose and pyr-uvate (lactate from glycolysis and alanine fromtransamination of pyruvate respectively) weresignificantly lower in mice fed MDCF comparedwith mice fed MSKH this was true for alaninein serum skeletal muscle and liver and for lac-tate in liver (table S12 A to C and H) Oxidativemetabolism of glucose is associated with nutri-tionally replete anabolic conditions These find-ings are consistent with the observed elevationsof the anabolic hormone IGF-1 in MDCF-fed com-pared with MSKHndashfed mice MDCF-fed micehad significantly higher circulating levels ofbranched-chain amino acids than those of theirMSKHndashfed counterparts (Fig 3G and table S12A to C) Skeletal muscle C5 carnitine and theclosely related metabolite C5-OHC3 carnitinewere significantly higher in animals consumingMDCF (Fig 3G and table S12F) In liver C3 and C5acylcarnitines were significantly lower in MDCF-treated mice (Fig 3G and table S12E) suggestingthat the more nutritionally replete state asso-ciated with MDCF may act to limit branched-chain amino acid oxidation in this tissue

Testing additional MDCF prototypesin gnotobiotic mice

Incorporating tilapia into MDCF prototypes posesseveral problems Its organoleptic propertiesare not desirable and its cost is higher than

commonly consumed plant-based sources ofprotein To identify alternatives to tilapia weselected an additional 16 plant-derived comple-mentary food ingredients with varied levels andquality of protein (25) that are culturally ac-ceptable affordable and readily available inBangladesh (fig S4A and table S13 A and B)Their effects were tested in gnotobiotic micecolonized with a defined expanded consortiumof 18 age- and growth-discriminatory bacterialstrains (table S6A) We generated 48 mouse dietsby supplementing a prototypic base diet repre-sentative of that consumed by 18-month-old chil-dren living in Mirpur (Mirpur-18) with each ofthe individual ingredients incorporated at threedifferent concentrations (fig S4A and table S13A)The results revealed that in this defined commu-nity context peanut flour had the greatest effecton the largest number of targeted weaning-phaseage-discriminatory taxa followed by chickpeaflour (fig S4B and table S13C) Soy flour whichpromoted the representation of two of thesetaxa had the second-highest percentage pro-tein after peanut flour (fig S4A) and its proteinquality was among the highest of the ingre-dients tested (table S13B) On the basis of theseobservations we chose soy and peanut flours asreplacements for tilapia in subsequent MDCFformulationsWe reasoned that by transplanting a repre-

sentative immature intact microbiota into younggerm-free mice we could investigate whether guthealth (defined by relative abundances of com-munity members expression of microbial genesin mcSEED metabolic pathways and biomark-ers and mediators of gut barrier function) wasimproved by supplementing the Mirpur-18 dietwith oneormore complementary food ingredientsthat targetweaning-phase age-discriminatory taxaFifteen fecal samples from 12 different childrenobtained during or after treatment for SAM werescreened in gnotobiotic mice to identify samplescontaining the greatest number of transmissi-ble weaning-phase age-discriminatory taxa andto assess their response to supplementation ofMirpur-18 (table S14A) We selected a sampleobtained from a donor (PS064) who had post-SAMMAM in addition to the successful transmis-sion of targeted taxa 887 plusmn 13 (mean plusmn SD) ofthe recipient animalsrsquo gut communities consistedof OTUs that were detected at gt01 relativeabundance in the donor sample (table S14B)Three groups of mice were colonized with thismicrobiota and monotonously fed one of threediets unsupplemented Mirpur-18 Mirpur-18 sup-plemented with peanut flour [Mirpur(P)] orMirpur-18 supplemented with four of the leadingredients [Mirpur(PCSB) with peanut flourchickpea flour soy flour and banana] (Fig 4Aand table S15A) Three control groups were main-tained as germ-free each group was fed one of thethree dietsWe characterized the effects of diet supple-

mentation on cecal and serum levels of metab-olites as well as on expression of genes in variousmicrobial metabolic pathways (tables S15 B Dand E and S16 and supplementary text results)

Eighteen mcSEED pathway modules involvedin amino acid metabolism were significantlyup-regulated in the cecal microbiomes of miceconsuming Mirpur(PCSB) or Mirpur(P) com-pared with those consuming Mirpur-18 withthe most up-regulated being ldquoisoleucine leucinevaline biosynthesisrdquo [other age-discriminatorymcSEED pathway modules that showed signifi-cantly lower abundances in the fecalmicrobiomesof Bangladeshi children with SAM and whoseexpression was increased by Mirpur(PCSB) orMirpur(P) in gnotobiotic mice are provided inFig 4B and fig S1E] Serum levels of a product ofbranched-chain amino acid metabolism C51-acylcarnitine were significantly higher in miceconsuming Mirpur(PCSB) compared with un-supplemented Mirpur-18 (0148 plusmn 0015 versus0086 plusmn 00098 mM respectively P = 0014 un-paired t test) Findings frommass spectrometricanalysis of cecal contents isolation and compara-tive genomic analysis of an F prausnitzii strainprominently represented in the transplantedcommunity and characterization of the in vivotranscriptional responses of this strain to thedifferent diets are described in table S15F andsupplementary text results

Gut mucosal barrier function

Epithelium and overlying mucus from the prox-imal middle and distal thirds of the small intestinewere recovered with laser capture microdissec-tion (LCM) (Fig 4D) Listed in table S15C are the30 most abundant OTUs identified by means ofV4-16S rDNA analysis of LCM mucosal DNA ob-tained from the different small intestinal seg-ments within a given diet group and betweensimilarly positioned segments across the differ-ent diet treatments For example Mirpur(PCSB)produced a statistically significant increase inthe relative abundance of F prausnitzii in theproximal two-thirds of the small intestine with-out significantly affecting the proportional rep-resentation of amilk-associated age-discriminatoryBifidobacteria OTU (Fig 4 C and D)Gene expression was characterized in the

jejunal mucosa (Fig 4D SI-2 segment) recoveredby LCM from mice belonging to all six treat-ment groups Significant differences in expres-sion were categorized based on enriched GeneOntology (GO) terms for ldquoMolecular FunctionrdquoIn colonized mice Mirpur(P) and Mirpur(PCSB)significantly up-regulated genes assigned toldquocadherin bindingrdquo (GO 0045296) and ldquocell ad-hesion molecule bindingrdquo (GO 0050839) com-pared with Mirpur-18 (table S17A) The diet effectwas colonization-dependent there were no signif-icant differences in expression of these genes orthese GO categories in germ-free mice consum-ing supplemented versus unsupplemented diets(table S17) (Further discussion is available in thesupplementary text results and histochemicaland immunohistochemical analyses of tissuesections prepared along the length of the smallintestines of these mice are provided in fig S6)On the basis of its effects on microbial commu-nity organismal composition gene expressionand gut barrier function we deemed Mipur-18




ugust 30 2020


Dow

nloaded from

supplemented with the four lead complementaryfoods [Mirpur(PCSB)] superior to that supple-mented with just peanut flour [Mirpur(P)]

Characterizing MDCF prototypesin gnotobiotic piglets

We examined the effects of MDCF prototypesin a second host species whose physiology andmetabolism are more similar to that of humansGnotobiotic piglets provide an attractive modelfor these purposes pigletsmanifest rapid growthrates in the weeks after birth (26) and methodsfor conducting experiments with gnotobioticpiglets have been described (27)On the basis of the results from the gnotobiotic

mouse studies we designed two MDCF proto-types One prototype was formulated to be anal-ogous to Mirpur-18 which contains milk powderthis prototype was supplemented with peanutflour chickpea flour soy flour and banana[MDCF(PCSB)] The other diet lackedmilk powderand was supplemented with just chickpea flourand soy flour [MDCF(CS)] The two MDCFs wereisocaloric matched in lipid levels and total protein

content (with equivalent representation of aminoacids) and also met current ready-to-use thera-peutic food guidelines for children with respect tomacro- andmicronutrient content (table S18A) (28)Four-day-old germ-free piglets fed a sowmilkndash

based formula were colonized with a 14-memberconsortium of bacterial strains that consistedof the same nine Bangladeshi age-discriminatorystrains used for the diet oscillation experimentsdescribed in fig S2 plus five weaning-phase age-discriminatory strains cultured from Malawianchildren (table S6B) In an earlier study sev-eral members of this consortium (B longumF prausnitzii Clostridium Ruminococcus gnavusand D formicigenerans) had been classified asgrowth-discriminatory by means of a RF-basedanalysis of their representation in gnotobioticmouse recipients of healthy and undernourisheddonor microbiota and the animalsrsquo weight andlean body mass gain phenotypes (4) After ga-vage the two groups of piglets were weaned overthe course of 10 days (supplementary materialsmaterials and methods) onto one or the otherirradiated MDCF prototypes which they con-

sumed ad libitum for the remainder of the ex-periment (n = 4 pigletstreatment arm) (Fig5A) Animals were euthanized on day 31 after a6-hour fastPiglets fed MDCF(PCSB) exhibited signifi-

cantly greater weight gain than those receivingMDCF(CS) (Fig 5B) Micro-computed tomogra-phy of their femurs revealed that they also hadsignificantly greater cortical bone volume (Fig5C) COPRO-seq analysis disclosed that pig-lets fed MDCF(PCSB) had significantly higherrelative abundances of C symbiosum R gnavusD formicigenerans R torques and Bacteroidesfragilis in their cecum and distal colon comparedwith those of piglets consuming MDCF(CS) (Fig5D and table S18B) all are weaning-phase age-discriminatory strains and the former threewere as noted above also defined as growth-discriminatory Conversely the relative abundancesof three members of Bifidobacteria (includingtwo milk-associated age-discriminatory strainsB breve and B longum subsp infantis) weresignificantly higher in the ceca and distal co-lons of piglets fed MDCF(CS) (table S18B) These


Fig 4 Effects of Mirpur-18 diet supplementation on a post-SAMMAM donor microbiota transplanted into gnotobiotic mice (A)Experimental design dpg days post gavage of the donor microbiotaMirpur(P) Mirpur-18 supplemented with peanut flour Mirpur(PCSB)Mirpur-18 supplemented with peanut flour chickpea flour soy flour andbanana (B) Expression of microbial mcSEED metabolic pathwaymodulesin the ceca of gnotobiotic mouse recipients of the post-SAM MAM donorgut community as a function of diet treatment P lt 005 P lt 0001P lt 00001 (statistical comparisons indicate results of gene setenrichment analysis expression on a per-gene basis across the indicated

mcSEED subsystempathway module all P values are FDR-adjusted)(C) Effects of supplementing Mirpur-18 with one or all four complementaryfood ingredients on the relative abundances of a weaning-phasendash and amilk-phasendashassociated taxon in feces obtained at dpg 21 (one-way ANOVAfollowed by Tukeyrsquos multiple comparisons test) (D) Relative abundancesof the two taxa in mucosa harvested by means of LCM from the proximalmiddle and distal thirds of the small intestine (Right) Schematic oflocations in the small intestine where LCM was performed The samecolor code for diets is used in (A) to (D) P lt 005 P lt 001 P lt00001 (Mann-Whitney test)



ugust 30 2020


Dow

nloaded from

findings led us to conclude that MDCF(PCSB)promoted a more weaning-phasendashlike (mature)community configuration than MDCF(CS) (ge-nome annotations microbial RNA-seq and tar-geted MS analyses of cecal metabolites areprovided in tables S18 C and D and S19A andsupplementary text results)The effects of the two MDCF prototypes on

host biology were defined by means of MS-based serum metabolomic and proteomic analy-ses (tables S19 and S20) Notable findingsincluded significant increases in levels of tryp-tophan methionine and C3-acylcarnitine withMDCF(PCSB) as well as changes produced in theserum proteome that are shared with childrenin the SAM trial (Fig 5 E and F and supple-mentary text results)

Testing MDCFs in Bangladeshichildren with MAM

To assess the degree to which results obtainedfrom the gnotobiotic mouse and piglet modelstranslate to humans we performed a pilot ran-domized double-blind controlled feeding studyof the effects of three MDCF formulations The

formulations (MDCF-1 -2 and -3) were designedto be similar in protein energy ratio and fat en-ergy ratio and provide 250 kcalday (divided overtwo servings) MDCF-2 contained all four leadingredients (chickpea flour soy flour peanutflour and banana) at higher concentrations thanin MDCF-1 MDCF-3 contained two lead ingre-dients (chickpea flour and soy flour) A rice- andlentil-based ready-to-use supplementary food(RUSF) included as a control arm lacked allfour ingredients but was otherwise similar inenergy density protein energy ratio fat energyratio and macro- andmicronutrient content tothose of the MDCFs (Fig 6A) Milk powder wasincluded in MDCF-1 and RUSF All formulationswere supplemented with a micronutrient mixturedesigned to provide 70 of the recommendeddaily allowances for 12- to 18-month-old childrenThe formulations were produced locally andtested for organoleptic acceptability before ini-tiating the trial (table S21A)Children fromMirpur withMAMand no prior

history of SAM were enrolled (mean age at en-rollment 152 plusmn 21 months mean WHZ ndash23 plusmn03) Participants were randomized into one of

the four treatment groups (14 to 17 children pergroup) and received 4 weeks of twice-dailyfeeding under supervision at the study centerpreceded and followed by 2 weeks of observa-tion and sample collection Mothers were en-couraged to continue their normal breastfeedingpactices throughout the study (Fig 6A) Therewere no significant differences in the mean dailyamount of each MDCF or RUSF consumed perchild or in the mean incidence of morbidityacross the four treatment groups (table S21 B toD) All three MDCFs and the RUSF control im-proved WHZ scores [ndash19 plusmn 05 (mean plusmn SD) atthe completion of intervention compared withndash22 plusmn 04 at the start of intervention n = 63children P = 206 times 10minus11 for all groups com-bined paired t test] There were no statisticallysignificant differences between the four inter-ventions in the change in WHZ (P = 031 one-way ANOVA) Despite the small group size andthe short length of the study there were signif-icant differences in treatment effects on anotheranthropometric indicator with MDCF-2 produc-ing a significantly greater increase in mid-upperarm circumference (MUAC) thanMDCF-3 (one-way ANOVA P = 0022 with Tukeyrsquos multiplecomparisons test P = 0017) (table S21E)

Effects on biological state

To contextualize the biological effects of the di-etary interventions we performed quantitativeproteomics (SOMAscan) on plasma collectedfrom 21 12- to 24-month-old Mirpur childrenwith healthy growth phenotypes (mean age 192 plusmn51 monthsWHZ 008 plusmn 058 HAZ ndash041 plusmn 056WAZ ndash012 plusmn 060) and 30 children with SAMbefore treatment (Fig 1A B1 sample WHZ lt ndash3mean age 152 plusmn 51 months) [metadata associ-ated with the healthy SAM and MAM (MDCFtrial) cohorts are provided in table S22] We rank-ordered all detected proteins according to folddifferences in their abundances in plasma collectedfrom healthy children compared with childrenwith untreatedSAMThe top 50most differentiallyabundant proteins (P lt 10minus7 R package ldquolimmardquo)that were significantly higher in healthy childrenwere designated ldquohealthy growth-discriminatoryrdquowhereas the top 50 differentially abundant pro-teins that were significantly higher in childrenwith SAMwere designated ldquoSAM-discriminatoryrdquo(table S23A) We next compared the mean dif-ference for each protein in the pre- versus post-intervention plasma samples for all children ineachMDCFRUSF treatment group Proteinswerethen ranked according to the fold differencesof the pre- versus post-treatment levels in eachof the four groups (table S23B) and these treat-ment effects were mapped onto the 50 mosthealthy growth-discriminatory and 50most SAM-discriminatoryproteins StrikinglyMDCF-2 eliciteda biological response characterized by a markedshift in theplasmaproteome toward that of healthychildren and away from that of childrenwith SAMMDCF-2 increased the abundance of proteins thatare higher in plasma from healthy children andreduced the levels of proteins elevated in SAMplasma samples (Fig 6B)


Fig 5 Effects of two different MDCF prototypes in gnotobiotic piglets (A) Experimentaldesign (B) Weight gain in piglets weaned onto isocaloric MDCF prototypes containing either peanutflour chickpea flour soy flour and banana [MDCF(PCSB)] or chickpea and soy flours [MDCF(CS)](C) mCT of femoral bone obtained at euthanasia (D) Effects of the MDCFs on the relative abun-dances of community members in cecal and distal colonic contents (E) Examples of serum proteinswith significantly different post-treatment levels between the two diet groups (F) Effect of dieton serum C3 acylcarnitine levels Mean values plusmn SD are plotted P lt 005 P lt 001 P lt 0005P lt 0001 [two-way ANOVA in (B) unpaired t test in (C) (D) and (F)] The color codeprovided in (B) also applies to (C) (D) and (F)



ugust 30 2020


Dow

nloaded from

Aggregating proteomic datasets from the com-bined cohort of 113 children with SAM MAMand healthy growth phenotypes for whom plas-ma samples were available we identified a totalof 27 plasma proteins that were significantlypositively correlated with HAZ and 57 plasmaproteins that were significantly negatively cor-related with HAZ [absolute value of Pearsoncorrelation gt 025 false discovery rate (FDR)ndashcorrected P value lt 005] Among the treatmentsMDCF-2 was distinctive in its ability to increasethe abundances of a broad range of proteins pos-itively correlated with HAZ including the major

IGF-1 binding protein IGFBP-3 growth hormonereceptor (GHR) and leptin (LEP) (Fig 6C) Growthdifferentiation factor 15 (GDF15) was reducedafter 4 weeks of dietary supplementation withMDCF-2 (Fig 6C) This transforming growthfactorndashb superfamily member which was nega-tively correlated with HAZ has been implicatedin the anorexia and muscle wasting associatedwith cancer and with chronic heart failure inchildren it was elevated in children with SAMand positively correlated with their lipolyticbiomarkers NEFA and ketones (supplementarytext results) Peptide YY an enteroendocrine cell

product elevated in SAM plasma that reducesappetite and negatively correlated with HAZwas also decreased by MDCF-2We identified GO terms that were enriched

among the group of treatment-responsive pro-teins and ranked them according to the P valueof their enrichment (table S23C) Proteins be-longing to GO terms significantly higher in healthycomparedwith SAMplasma sampleswere deemedldquohealthy growth-discriminatoryrdquo whereas thosethatwere significantlyhigher inSAMweredeemedldquoSAM-discriminatoryrdquo (fold-difference gt30 FDR-adjusted P value lt005) This analysis revealedmultiple healthy growth-discriminatory proteinsassociated with GO processes ldquoosteoblast differ-entiationrdquo and ldquoossificationrdquo that were increasedby supplementation with MDCF-2 (Fig 7A andtable S23C) Examples include key markers ormediators of osteoblast differentiation [osteo-pontin (SPP1) bone sialoprotein 2 (IBSP) andbonemorphogenetic protein 7 (BMP7)] as well asmatrix metalloproteases (MMP-2 and MMP-13)involved in terminal differentiation of osteoblastsinto osteocytes and bone mineralizationA number of plasma proteins categorized un-

der the GO process ldquoCNS developmentrdquo includ-ing those involved in axon guidance and neuronaldifferentiation were also affected by MDCF-2 sup-plementation Levels of the SAM-discriminatorysemaphorin SEMA3A a potent inhibitor of axonalgrowth decreased with this treatment where-as healthy growth-discriminatory semaphorins(SEMA5A SEMA6A and SEMA6B) increased(Fig 7B) Other healthy growth-discriminatoryproteins whose abundances increased withMDCF-2 included receptors for neurotrophin(NTRK2 and NTRK3) plus various axonal guidanceproteins [netrin (UNC5D) ephrin A5 (EFNA5)roundabout homolog 2 (ROBO2) and SLIT andNTRK-like protein 5 (SLITRK5)] (Fig 7B) Ex-pression of a number of neurotrophic proteinsbelonging to these families has been reportedto be influenced by nutrient availability in pri-mates (29)Compared with healthy children the plasma

proteome of children with SAM was charac-terized by elevated levels of acute phase proteins[such as C-reactive protein (CRP) or interleukin-6 (IL-6)] and inflammatory mediators includingseveral agonists and components of the nuclearfactorndashkB (NF-kB) signaling pathway (Fig 7C)These components include the pro-inflammatorycytokines IL-1b tumor necrosis factorndasha (TNF-a)and CD40L plus ubiquitin-conjugating enzymeE2 N (UBE2N) which is involved in inductionof NF-kBndash and mitogen-activated protein kinase(MAPK)ndashresponsive inflammatory genes (30)MDCF-2 supplementation was associated withreductions in the levels of all of these SAM-associated proteins (Fig 7C)

Effects on the microbiota

Our analysis of fecal microbiota samples re-vealed no significant change in the representa-tion of enteropathogens within and across thefour treatment groups (fig S7A and table S21F)MDCF-2ndashinduced changes in biological state


Fig 6 Comparing the effects of MDCF formulations on the health status of Bangladeshichildren with MAM (A) Study design and composition of diets Total carbohydrate includes allcomponents except added sugar (B) Quantitative proteomic analysis of the average fold-changeper treatment group in the abundances of the 50 plasma proteins most discriminatory forhealthy growth and the 50 plasma proteins most discriminatory for SAM (protein abundance iscolumn-normalized across treatment groups) (C) Average fold-change in abundances of plasmaproteins that significantly positively or negatively correlate with HAZ [absolute value of Pearsoncorrelation gt 025 FDR-corrected P value lt 005 abundance is column-normalized as in (B)]



ugust 30 2020


Dow

nloaded from

were accompanied by increases in the relativeabundances of several weaning-phase taxa in-cluding OTUs assigned to F prausnitzii (OTU851865) and a Clostridiales sp (OTU 338992)that are closely related to taxa ranked first andsecond in feature importance in the sparseBangladeshi RF-derived model of gut microbiotamaturation (fig S7 B and C) MDCF-2 supple-mentation was associated with a significantdecrease in B longum (OTU 559527) (fig S7B)which is ranked third in feature importance inthe RF-derived model and discriminatory for ayoung milk-oriented microbiota None of theother members of the 30 OTU model showedsignificant changes By contrast MDCF-1 didnot produce significant increases in any of thetaxa in the model The other two formulationswere each associated with a significant changein just one member [an increase in the relativeabundance of an early age-discriminatory OTU(Streptococcus ranked 30th) with MDCF-3 sup-plementation and a decrease in another OTU(Enterococcus faecalis ranked 29th) with RUSFsupplementation] (table S4B)MAZ scores were not significantly different

between groups at enrollment nor were theysignificantly improved by any of the formula-

tions Interpretation of this finding was con-founded by unexpectedly high baseline microbiotamaturity scores in this group of children withMAM [MAZ ndash001 plusmn 112 (mean plusmn SD)] (tableS22) compared with a small previously char-acterized Mirpur cohort with untreated MAM(2) Hence we developed an additional measureof microbiota repair (31) This involved a statis-tical analysis of covariance among bacterial taxain the fecal microbiota of anthropometricallyhealthy members of a Mirpur birth cohort whohad been sampled monthly over a 5-year periodUsing approaches developed in the fields ofeconophysics and protein evolution to charac-terize the underlying organization of interactingsystems with seemingly intractable complexitysuch as financial markets we found that the gutcommunity in healthy children could be decom-posed into a sparse unit of 15 covarying bacterialtaxa termed an ldquoecogrouprdquo (31) These ecogrouptaxa include a number of age-discriminatorystrains in the Bangladeshi RF-derived model(such as B longum F prausnitzii and Prevotellacopri) We used the ecogroup to show that inaddition to its effects on host biological stateMDCF-2 was also the most effective of the fourtreatments in reconfiguring the gut bacterial

community to a mature state similar to thatcharacteristic of healthy Bangladeshi children

Conclusions

We have integrated preclinical gnotobiotic ani-mal models with human studies to understandthe contributions of perturbed gut microbiotadevelopment to childhood undernutrition andto identify new microbiota-directed therapeu-tic approaches We identified a set of proteinsthat distinguish the plasma proteomes of healthychildren from those with SAM Using these datawe have developed a supplemental food proto-type MDCF-2 that shifted the plasma proteomeof children with MAM toward that of healthyindividuals including proteins involved in lineargrowth bone development neurodevelopmentand immune function MDCF-2 is a tool for in-vestigating in larger studies across differentpopulations with varying degrees of undernu-trition how repair of gut microbiota immaturityaffects various facets of child growth

Overview of methodsHuman studies

Children aged 6 to 59 months with SAM (n =343 participants) were enrolled in a study en-titled ldquoDevelopment and field testing of ready-to-use therapeutic foods (RUTF) made of localingredients in Bangladesh for the treatment ofchildren with severe acute malnutritionrdquo Thestudy was approved by the Ethical Review Com-mittee at icddrb (ClinicalTrialsgov identifierNCT01889329) Written informed consent wasobtained from their parent or guardian A sub-set of 54 children were included in a substudythat involved regular fecal sampling and threeblood draws for up to 1 year after dischargeChildren aged 12 to 18 months with MAM whowere no longer exclusively breastfed (n = 63participants) were enrolled in a double-blindrandomized four-group parallel assignmentinterventional trial study (ClinicalTrialsgov iden-tifier NCT03084731) approved by the EthicalReview Committee at icddrb Fecal and plasmasamples were collected as described in the sup-plementary materials materials and methodsand stored at ndash80degC Samples were shipped toWashington University with associated clinicalmetadata and maintained in a dedicated bio-specimen repository with approval from theWashington University Human Research Pro-tection Office

Analysis of plasma samples

Methods for targeted MS-based metabolomicsare described in the supplementary materialsThe SOMAscan 13K Proteomic Assay plasmaserum kit (SomaLogic Boulder Colorado) wasused to measure 1305 proteins The R packageldquolimmardquo (Bioconductor) was used to analyzedifferential protein abundances (32) Spearmancorrelation analyses were performed betweenmeasured proteins and anthropometric scoresplasma metabolites and the abundances of bac-terial taxa in fecal samples Plasma proteins inchildren with healthy growth phenotypes or SAM


Fig 7 The effects of different MDCF formulations on biomarkers and mediators of bone andCNS development plus NF-kB signaling (A to C) Average fold-change (normalized across treat-ment groups) in the abundances of plasma proteins belonging to GO categories related to (A) bone(B) CNS development and (C) agonists and components of the NF-ĸB signaling pathway Proteins inthe GO category that were significantly higher in the plasma of healthy compared with SAM children(fold-difference gt30 FDR-adjustedP value lt 005) are labeled ldquohealthy growth-discriminatoryrdquowhereasthose higher in SAM compared with healthy children (fold-difference gt30 FDR-adjusted P valuelt 005) are labeled ldquoSAM-discriminatoryrdquo Levels of multiple ldquohealthy growth-discriminatoryrdquo proteinsassociated with (A) GO processes ldquoosteoblast differentiationrdquo and ldquoossificationrdquo and (B) the GO processldquoCNS developmentrdquo are enhanced by MDCF-2 treatment while (C) NF-kB signaling is suppressed



ugust 30 2020


Dow

nloaded from

(before treatment) were rank-ordered accordingto the fold-difference in their levels betweenthese two groups The top 50most differentiallyabundant proteins in healthy compared with SAMwere designated as healthy growth-discriminatoryproteins and the top 50 most differentiallyabundant in SAM compared with healthy weredesignated as SAM-discriminatory proteinsThe average fold-change for these healthy growth-and SAM-discriminatory proteins was then cal-culated for each treatment arm in theMDCF trial(before versus after MDCFRUSF treatment) andnormalized to the mean fold-change across allfour arms Limma was used to calculate statisti-cal significance All 1305 proteins were mappedto all GO ldquoBiological Processesrdquo in the GO data-base (wwwgeneontologyorg) SetRank a geneset enrichment analysis (GSEA) algorithm (33)was used to identify GO Biological Processesthat were significantly enriched for proteinsthat exhibited changes in abundance from be-fore to after treatment with MDCFRUSF Enrich-ment was calculated by using the setRankAnalysisfunction in the SetRank R library (parameterswere useranks = TRUE setPCutoff = 001 andfdrCutoff = 005) The average fold-change foreach protein in the statistically significant Bi-ological Process category was calculated for eachtreatment arm and normalized to the mean fold-change across all four arms We defined proteinswithin the GO Biological Process as ldquohealthygrowth-discriminatoryrdquo if they were at least 30more abundant in healthy individuals comparedwith those with SAM and ldquoSAM-discriminatoryrdquoif they were at least 30 more abundant in chil-dren with SAM compared with those who wereclassified as healthy

Characterizing human fecal microbialcommunities

Methods for V4-16S rRNA gene sequencing anddata analysis calculation of MAZ scores andfunctional microbiome maturity and quantifi-cation of enteropathogen burden by means ofmultiplex quantitative polymerase chain reac-tion (qPCR) are described in the supplemen-tary materials

Animal studiesGnotobiotic mice

All mouse experiments were performed by usingprotocols approved by the Washington Uni-versity Animal Studies Committee Mice werehoused in plastic flexible film gnotobiotic iso-lators under a 12-hour light cycle Male germ-freeC57BL6 mice were initially weaned onto an auto-claved low-fat high-plant polysaccharide chowthat was administered ad libitum (diet 2018SEnvigo) Animals were maintained on this dietuntil 3 days before the beginning of experimentsinvolving tests of the effects of complementaryfood ingredients Defined consortia of sequencedbacterial strains cultured from Bangladeshi chil-dren or intact uncultured microbiota from do-nors with post-SAM MAM were introduced bymeans of gavage into recipientmice at 5 weeks ofage Methods for identifying and characterizing

the effectsMDCF prototypesmdashincluding (i) designand preparation of diets (ii) culturing of age-discriminatory and SAM-associated bacterialstrains (iii) shotgun sequencing of DNA isolatedfrom serially collected fecal samples (iv) micro-bial RNA-seq of cecal contents (v) targetedMS ofcecal contents liver gastrocnemius muscle andserum samples for measurement of amino acidsacylcarnitines organic acids and acylCoAs (vi)Western blot analysis of IGF-1 pathway com-ponents in liver (vii) mCT of bone and (viii)characterizing the effects of a transplanted fecalmicrobiome from a donor with post-SAM MAMin recipient gnotobioticmice as a function of diettreatment by histochemical and immunohisto-chemical analysis of their intestinal segmentsLCM of their small intestinal epithelium andRNA-seq analysis of gene expression in LCMmucosamdashare all described in the supplementarymaterials

Gnotobiotic piglets

Experiments were performed under the supervi-sion of a veterinarian by using protocols approvedby the Washington University Animal StudiesCommittee and that followed American Veter-inaryMedical Association guidelines for eutha-nasia The protocol for generating germ-freepiglets preparing diets feeding colonizationand husbandry of piglets measurement of weightgain mCT of femurs and liquid chromatographyndashMS (LC-MS)MSndashbased serum proteomics are alldescribed in the supplementary materials

REFERENCES AND NOTES

1 T Ahmed et al The MAL-ED cohort study in MirpurBangladesh Clin Infect Dis 59 S280ndashS286 (2014)doi 101093cidciu458 pmid 25305298

2 S Subramanian et al Persistent gut microbiota immaturity inmalnourished Bangladeshi children Nature 510 417ndash421(2014) doi 101038nature13421 pmid 24896187

3 World Health Organization Department of Nutrition for Healthand Development WHO child growth standards Lengthheight-for-age weight-for-age weight-for-length weight-for-height and body mass index-for-age methods anddevelopment (2000) wwwwhointchildgrowthen

4 L V Blanton et al Gut bacteria that prevent growthimpairments transmitted by microbiota from malnourishedchildren Science 351 aad3311 (2016) doi 101126scienceaad3311 pmid 26912898

5 World Health Organization (WHO) Infant and Young Childfeeding fact sheet no 342 1ndash5 (WHO 2016)

6 L Manikam et al A systematic review of complementaryfeeding practices in South Asian infants and young childrenThe Bangladesh perspective BMC Nutr 3 56 (2017)doi 101186s40795-017-0176-9

7 H Sandige M J Ndekha A Briend P Ashorn M J ManaryHome-based treatment of malnourished Malawian childrenwith locally produced or imported ready-to-use food J PediatrGastroenterol Nutr 39 141ndash146 (2004) doi 10109700005176-200408000-00003 pmid 15269617

8 L Gold J J Walker S K Wilcox S Williams Advances inhuman proteomics at high scale with the SOMAscanproteomics platform Nat Biotechnol 29 543ndash549 (2012)pmid 22155539

9 B Lollo F Steele L Gold Beyond antibodies New affinityreagents to unlock the proteome Proteomics 14 638ndash644(2014) doi 101002pmic201300187 pmid 24395722

10 J Candia et al Assessment of variability in the SOMAscanassay Sci Rep 7 14248 (2017) doi 101038s41598-017-14755-5 pmid 29079756

11 S Bartz et al Severe acute malnutrition in childhoodHormonal and metabolic status at presentation response totreatment and predictors of mortality J Clin Endocrinol

Metab 99 2128ndash2137 (2014) doi 101210jc2013-4018pmid 24606092

12 R Overbeek et al The SEED and the Rapid Annotation ofmicrobial genomes using Subsystems Technology (RAST)Nucleic Acids Res 42 D206ndashD214 (2014) doi 101093nargkt1226 pmid 24293654

13 D A Sela et al The genome sequence of Bifidobacteriumlongum subsp infantis reveals adaptations for milk utilizationwithin the infant microbiome Proc Natl Acad Sci USA105 18964ndash18969 (2008) doi 101073pnas0809584105pmid 19033196

14 T Ahmed et al Nutrition of children and women in BangladeshTrends and directions for the future J Health Popul Nutr 301ndash11 (2012) doi 103329jhpnv30i111268 pmid 22524113

15 L L Iannotti et al Eggs in early complementary feedingand child growth A randomized controlled trial Pediatrics140 e20163459 (2017) doi 101542peds2016-3459pmid 28588101

16 W R Russell et al Major phenylpropanoid-derived metabolitesin the human gut can arise from microbial fermentation ofprotein Mol Nutr Food Res 57 523ndash535 (2013) doi 101002mnfr201200594 pmid 23349065

17 D Krause et al The tryptophan metabolite 3-hydroxyanthranilicacid plays anti-inflammatory and neuroprotective rolesduring inflammation Role of hemeoxygenase-1 Am J Pathol179 1360ndash1372 (2011) doi 101016jajpath201105048pmid 21855684

18 L Cervantes-Barragan et al Lactobacillus reuteri induces gutintraepithelial CD4+CD8aa+ T cells Science 357 806ndash810(2017) doi 101126scienceaah5825 pmid 28775213

19 G Das et al An important regulatory role for CD4+CD8 a aT cells in the intestinal epithelial layer in the prevention ofinflammatory bowel disease Proc Natl Acad Sci USA100 5324ndash5329 (2003) doi 101073pnas0831037100pmid 12695566

20 H Cheroutre M M Husain CD4 CTL Living up to thechallenge Semin Immunol 25 273ndash281 (2013) doi 101016jsmim201310022 pmid 24246226

21 T Sujino et al Tissue adaptation of regulatory and intraepithelialCD4+ T cells controls gut inflammation Science 352 1581ndash1586(2016) doi 101126scienceaaf3892 pmid 27256884

22 R A Saxton D M Sabatini mTOR signaling in growthmetabolism and disease Cell 169 361ndash371 (2017)doi 101016jcell201703035 pmid 28388417

23 J Yan et al Gut microbiota induce IGF-1 and promote boneformation and growth Proc Natl Acad Sci USA 113E7554ndashE7563 (2016) doi 101073pnas1607235113pmid 27821775

24 M Schwarzer et al Lactobacillus plantarum strain maintainsgrowth of infant mice during chronic undernutritionScience 351 854ndash857 (2016) doi 101126scienceaad8588pmid 26912894

25 D J Millward Amino acid scoring patterns for protein qualityassessment Br J Nutr 108 (Suppl 2) S31ndashS43 (2012)doi 101017S0007114512002462 pmid 23107544

26 P L Altman D S Dittmer Growth Including Reproductionand Morphological Development (Federation of AmericanSocieties for Experimental Biology 1962)

27 M R Charbonneau et al Sialylated milk oligosaccharidespromote microbiota-dependent growth in models of infantundernutrition Cell 164 859ndash871 (2016) doi 101016jcell201601024 pmid 26898329

28 US Department of Agriculture (USDA) ldquoCommercial ItemDescription Ready-to-Use Therapeutic Food (RUTF)rdquoA-A-20363B (USDA 2012)

29 I Antonow-Schlorke et al Vulnerability of the fetal primatebrain to moderate reduction in maternal global nutrientavailability Proc Natl Acad Sci USA 108 3011ndash3016 (2011)doi 101073pnas1009838108 pmid 21252306

30 K Taniguchi M Karin NF-kB inflammation immunity andcancer Coming of age Nat Rev Immunol 18 309ndash324(2018) doi 101038nri2017142 pmid 29379212

31 A S Raman et al A sparse covarying unit that describeshealthy and impaired human gut microbiota developmentScience 365 eaau4735 (2019)

32 M E Ritchie et al limma powers differential expressionanalyses for RNA-sequencing and microarray studiesNucleic Acids Res 43 e47 (2015) doi 101093nargkv007pmid 25605792

33 C Simillion R Liechti H E L Lischer V IoannidisR Bruggmann Avoiding the pitfalls of gene set enrichmentanalysis with SetRank BMC Bioinformatics 18 151 (2017)doi 101186s12859-017-1571-6 pmid 28259142




ugust 30 2020


Dow

nloaded from

ACKNOWLEDGMENTSWe are grateful to the families of members of the human studiesdescribed in this work for their participation and assistance Weare indebted to the staff and health care workers at icddrb fortheir contributions to the recruitment and enrollment of mothersas well as the collection of biospecimens and data from theiroffspring We thank M Karlsson M Meier S Wagoner S DengJ Serugo and J Hoisington-Loacutepez for superb technical assistanceK Ahsan for assistance with maintaining the biospecimenrepository and associated database J Guruge for help withanaerobic microbiology O Delannoy-Bruno for assistance with thegnotobiotic piglet experiment Mars Inc for their assistance withmanufacturing the MDCF(PCSB) and MDCF(CS) diets A Lutzand J Yu (Genome Technology Access Center at WashingtonUniversity) for their contributions to generating SOMAscandatasets R Olson and other members of RASTSEED developmentteam at the Argonne National Laboratory for support with themcSEED-based genome analysis and subsystem curation andD Leib for developing the computer program to quantify CT scandata obtained from the femurs of gnotobiotic piglets FundingThis work was supported by the Bill amp Melinda Gates Foundation aspart of the Breast Milk gut Microbiome and Immunity (BMMI)Project As members of Washington Universityrsquos Medical ScientistTraining Program RYC and SS received support from NIHgrant GM007200 mCT of femoral bone was performed usingresources provided by the Washington University MusculoskeletalResearch Center (NIH P30 AR057235) Histochemical andimmunohistochemical processing of tissue sections was performedat the Washington University Digestive Diseases Research CoreCenter funded by NIH P30 DK052574 Plasma proteomic datasetswere generated by the Genome Technology Access Center atWashington University School of Medicine which is supported inpart by NIH Grants P30 CA91842 and UL1TR002345 DARAAA and SAL were supported in part by the Russian ScienceFoundation (grants 14-14-00289 and 19-14-00305) JIG is therecipient of a Thought Leader Award from Agilent TechnologiesThis work is licensed under a Creative Commons Attribution 40International (CC BY 40) license which permits unrestricted use

distribution and reproduction in any medium provided the originalwork is properly cited To view a copy of this license visit httpcreativecommonsorglicensesby40 This license does not applyto figuresphotosartwork or other content included in the articlethat is credited to a third party obtain authorization from therights holder before using such material Author contributionsTA IH MI NC SH IMa MMa SAS and IMo wereresponsible for the design and conduct of the human studies pluscollection of biospecimens and clinical metadata MJB establishedand maintained biospecimen repositories and associated databasesof de-identified metadata JLG and MFM generated the 16S rDNAand shotgun sequencing datasets from human fecal samples andJLG and MCH analyzed the data SV JLG MJB and JIGdesigned the gnotobiotic mouse studies H-WC MJB and JIGdesigned the gnotobiotic piglet experiments JLG SV andSS cultured bacterial strains JLG SV H-WC and MCHsequenced and assembled the genomes of bacterial strains used ingnotobiotic animal experiments DAR AAA SAL and ALOperformed in silico metabolic reconstructions with the genomes ofthe cultured strains BH provided updated CAZyme annotationsSV and JLG performed gnotobiotic mouse experiments withcultured bacterial strains and intact uncultured communitiesrespectively H-WC DOD and MT conducted the gnotobioticpiglet experiments SV and H-WC generated COPRO-Seqdatasets SV JLG MCH and H-WC produced microbialRNA-seq datasets VLK performed laser capture microdissectionV4-16S rDNA analysis of intestinal mucosa-associated bacterialcommunity composition RNA-seqndashbased characterization ofmouse small intestinal mucosal gene expression and histochemicalassays of intestinal morphometry JC SV JLG H-WC MMuOI and CBN conducted metabolomic analyses of mousepiglet and human biospecimens CAC performed mCT of femursmeasured serum IGF-1 levels in gnotobiotic mice and quantifiedleptin IGF-1 and insulin in plasma obtained from children in the SAMtrial H-WC generated microcomputed tomographic datasets frompiglet femurs LDS and CFS characterized levels of IGF-1 pathwaycomponents in the livers of gnotobiotic mice RJG and RLHwere responsible for MS-based proteomics of piglet serum samples

RDH and MJB produced the quantitative proteomic datasetsfrom plasma samples with DNA aptamer-based arrays and RYCMJB JLG and MCH analyzed the data CS and MCHperformed and analyzed qPCR assays for enteropathogens JLGSV H-WC MJB and JIG wrote the paper with assistanceprovided by co-authors Competing interests JIG is a cofounderof Matatu a company characterizing the role of diet-by-microbiotainteractions in animal health LDS is currently a scientific salesrepresentative at STEMCELL Technologies Data and materialsavailability V4-16S rDNA sequences in raw format prior topost-processing and data analysis COPRO-seq microbial RNA-seqand proteomics datasets plus shotgun sequencing datasets producedfrom human fecal DNA cecal contents of gnotobiotic mice with apost-SAM MAM human donor community and cultured bacterialstrains have been deposited at the European Nucleotide Archive(ENA) under study accession no PRJEB26419 SOMAscan-generatedhuman plasma proteomic datasets have been deposited in theGene Expression Omnibus (GEO) database under accession noGSE119641 All raw mass spectra for quantification of serum proteinsin gnotobiotic piglets have been deposited in the MassIVE andProteomeXchange data repositories under accession nosMSV000082286 (MassIVE) and PXD009534 (ProteomeXchange)with data files available at ftpmassiveucsdeduMSV000082286Fecal and plasma specimens from the SAM and MDCF studies usedfor the analyses described in this study were provided to WashingtonUniversity under materials transfer agreements with icddrb

SUPPLEMENTARY MATERIALS

sciencesciencemagorgcontent3656449eaau4732supplDC1Materials and MethodsSupplementary TextFigs S1 to S7References (34ndash124)Tables S1 to S23

13 June 2018 resubmitted 24 April 2019Accepted 7 June 2019101126scienceaau4732




ugust 30 2020


Dow

nloaded from

Effects of microbiota-directed foods in gnotobiotic animals and undernourished children

Mostafa Mustafa Mahfuz Michael J Barratt Tahmeed Ahmed and Jeffrey I GordonOsterman Md Iqbal Hossain Munirul Islam Nuzhat Choudhury Shafiqul Alam Sarker Sayeeda Huq Imteaz Mahmud IshitaNewgard Christopher Sawyer Richard D Head Dmitry A Rodionov Aleksandr A Arzamasov Semen A Leyn Andrei L Semenkovich Bernard Henrissat Richard J Giannone Robert L Hettich Olga Ilkayeva Michael Muehlbauer Christopher BChen Sathish Subramanian Carrie A Cowardin Martin F Meier David ODonnell Michael Talcott Larry D Spears Clay F Jeanette L Gehrig Siddarth Venkatesh Hao-Wei Chang Matthew C Hibberd Vanderlene L Kung Jiye Cheng Robert Y

DOI 101126scienceaau4732 (6449) eaau4732365Science

this issue p eaau4732 p eaau4735Sciencemetabolic and growth profiles on a healthier trajectoryage-characteristic gut microbiota The designed diets entrained maturation of the childrens microbiota and put theirstate that might be expected to support the growth of a child These were first tested in mice inoculated with recovery Diets were then designed using pig and mouse models to nudge the microbiota into a mature post-weaningmalnutrition The authors investigated the interactions between therapeutic diet microbiota development and growth

monitored metabolic parameters in healthy Bangladeshi children and those recovering from severe acuteet alRaman andet altherapeutic intervention with standard commercial complementary foods children may fail to thrive Gehrig

Childhood malnutrition is accompanied by growth stunting and immaturity of the gut microbiota Even afterMalnutrition and dietary repair

ARTICLE TOOLS httpsciencesciencemagorgcontent3656449eaau4732

MATERIALSSUPPLEMENTARY httpsciencesciencemagorgcontentsuppl201907103656449eaau4732DC1

CONTENTRELATED

httpstmsciencemagorgcontentscitransmed4137137rv6fullhttpstmsciencemagorgcontentscitransmed4137137rv7fullhttpstmsciencemagorgcontentscitransmed6220220ra11fullhttpstmsciencemagorgcontentscitransmed7276276ra24fullhttpstmsciencemagorgcontentscitransmed8366366ra164fullhttpsciencesciencemagorgcontentsci3656449eaau4735fullhttpsciencesciencemagorgcontentsci3656449109full

REFERENCES

httpsciencesciencemagorgcontent3656449eaau4732BIBLThis article cites 118 articles 25 of which you can access for free

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAASScienceScience 1200 New York Avenue NW Washington DC 20005 The title (print ISSN 0036-8075 online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Copyright copy 2018 American Association for the Advancement of Science

on August 30 2020

httpsciencesciencem

agorgD

ownloaded from

PERMISSIONS httpwwwsciencemagorghelpreprints-and-permissions




on August 30 2020

httpsciencesciencem

agorgD

ownloaded from

RESEARCH ARTICLE

MICROBIOTA

Effects of microbiota-directedfoods in gnotobiotic animals andundernourished childrenJeanette L Gehrig12 Siddarth Venkatesh12 Hao-Wei Chang12Matthew C Hibberd12 Vanderlene L Kung123 Jiye Cheng12 Robert Y Chen12Sathish Subramanian12dagger Carrie A Cowardin12 Martin F Meier12 David OrsquoDonnell12Michael Talcott4 Larry D Spears5 Clay F Semenkovich5 Bernard Henrissat67Richard J Giannone8 Robert L Hettich8 Olga Ilkayeva910 Michael Muehlbauer910Christopher B Newgard9101112 Christopher Sawyer1314 Richard D Head1314Dmitry A Rodionov1516 Aleksandr A Arzamasov1516 Semen A Leyn1516Andrei L Osterman16 Md Iqbal Hossain17 Munirul Islam17 Nuzhat Choudhury17Shafiqul Alam Sarker17 Sayeeda Huq17 Imteaz Mahmud17 Ishita Mostafa17Mustafa Mahfuz17 Michael J Barratt12 Tahmeed Ahmed17 Jeffrey I Gordon12Dagger

To examine the contributions of impaired gut microbial community development tochildhood undernutrition we combined metabolomic and proteomic analyses of plasmasamples with metagenomic analyses of fecal samples to characterize the biological stateof Bangladeshi children with severe acute malnutrition (SAM) as they transitioned afterstandard treatment to moderate acute malnutrition (MAM) with persistent microbiotaimmaturity Host and microbial effects of microbiota-directed complementary food(MDCF) prototypes targeting weaning-phase bacterial taxa underrepresented in SAM andMAM microbiota were characterized in gnotobiotic mice and gnotobiotic piglets colonizedwith age- and growth-discriminatory bacteria A randomized double-blind controlledfeeding study identified a lead MDCF that changes the abundances of targeted bacteriaand increases plasma biomarkers and mediators of growth bone formationneurodevelopment and immune function in children with MAM

Evidence is accumulating that disruptionof ldquonormalrdquo gut community (microbiota)development may contribute to the path-ogenesis of undernutrition Using culture-independent surveys bacterial membership

has beendefined in fecal samples collectedmonth-ly during the first 2 postnatal years from healthymembers of a birth cohort living in an urbanslum (Mirpur) in Dhaka Bangladesh (1 2) Ap-plying machine learning [Random Forests (RF)]to the resulting 16S ribosomal DNA (rDNA) data-set yielded a ldquosparserdquomodel composed of themostage-discriminatory bacterial strains changes inthe relative abundances of these organisms de-scribed a program of normal microbiota develop-ment (2) ThisRF-derivedmodelwas subsequentlyused to characterize fecal samples collected from

Bangladeshi children with severe acute mal-nutrition [SAM defined as a weight-for-heightz-score (WHZ) gt3 standard deviations below themedian for a World Health Organization (WHO)reference healthy growth cohort (3)] The resultsrevealed gut communities that resembled thoseof healthy children who were chronologicallyyounger Thismicrobiota ldquoimmaturityrdquowasmorepronounced in children with SAM as comparedwith those with moderate acute malnutrition(MAM WHZ score between ndash2 and ndash3) and wasnot repaired in a clinical study that tested theeffects of two therapeutic foods (2)Impaired microbiota development has also

been documented in undernourished Malawianchildren (4) To examine the functional impor-tance of this impairment microbial communi-

ties from healthy and stunted or underweight6- and 18-month-old Malawian children weretransplanted into groups of recently weanedgerm-free mice fed a diet representative of thatconsumed by the human donor population Theresults disclosed that compared with mice col-onized with normally maturing microbiota fromthe healthy donors animals harboring imma-ture microbiota exhibited reduced rates of leanbody mass gain alterations in bone growth andmetabolic abnormalities (4) These studies pro-vided preclinical evidence for a causal link be-tween microbiota immaturity and undernutritionthey also revealed that a subset of the age-discriminatory strains is growth-discriminatoryMoreover a cultured consortium of these age-and growth-discriminatory taxa amelioratedthe impaired growth phenotype transmitted torecipient gnotobiotic mice by an immature mi-crobiota (4)One question arising from these observations

is how do we design optimal foods that steer amicrobiota into an age-appropriate and healthystate Breastfeeding plays a major role in reduc-ing childhood malnutrition As such WHO andthe United Nations Childrenrsquos Fund (UNICEF)recommend exclusive breastfeeding for the first6 months of postnatal life and continued breast-feeding after the introduction of complementaryfoods up to 24 months of age (5) Suboptimalcomplementary feeding practices are importantcontributors to malnutrition in children less than2 years of age (6) However current complemen-tary feeding recommendations are not based onknowledge of how foods affect the develop-mental biology of the gut microbiota during theweaning process Together these observationsraise the question Do certain complementaryfood ingredients or combinations of ingredientshave the ability to selectively increase the rep-resentation and expressed beneficial functions ofage- and growth-discriminatory strains deficientin SAM- or MAM-associated microbiota If theanswer is yes then prescribed feeding of theseingredients could help ldquorepairrdquo or prevent de-velopment of microbiota immaturity in chil-dren with potentially long-lived health-promotingeffectsHere we describe a process for identifying

microbiota-directed complementary foods (MDCFs)designed to treat children with acute malnutri-tion We first characterized gut microbial com-munity and host responses over the course of12 months in Bangladeshi children who weretreated for SAM with one of three conventionaltherapeutic foods Measurement of the levels of

RESEARCH


1Edison Family Center for Genome Sciences and Systems Biology Washington University School of Medicine St Louis MO 63110 USA 2Center for Gut Microbiome and Nutrition ResearchWashington University School of Medicine St Louis MO 63110 USA 3Department of Pathology and Immunology Washington University School of Medicine St Louis MO 63110 USA 4Divisionof Comparative Medicine Washington University St Louis MO 63110 USA 5Department of Medicine Washington University School of Medicine St Louis MO 63110 USA 6Architecture etFonction des Macromoleacutecules Biologiques Centre National de la Recherche Scientifique and Aix-Marseille Universiteacute 13288 Marseille cedex 9 France 7Department of Biological Sciences KingAbdulaziz University Jeddah Saudi Arabia 8Chemical Sciences Division Oak Ridge National Laboratory Oak Ridge TN 37830 USA 9Sarah W Stedman Nutrition and Metabolism Center DukeUniversity Medical Center Durham NC 27710 USA 10Duke Molecular Physiology Institute Duke University Medical Center Durham NC 27710 USA 11Department of Pharmacology and CancerBiology Duke University Medical Center Durham NC 27710 USA 12Department of Medicine Duke University Medical Center Durham NC 27710 USA 13Department of Genetics WashingtonUniversity School of Medicine St Louis MO 63110 USA 14Genome Technology Access Center Washington University School of Medicine St Louis MO 63110 USA 15A A Kharkevich Institutefor Information Transmission Problems Russian Academy of Sciences Moscow 127994 Russia 16Infectious and Inflammatory Disease Center Sanford Burnham Prebys Medical DiscoveryInstitute La Jolla CA 92037 USA 17International Centre for Diarrhoeal Disease Research Bangladesh (icddrb) Dhaka 1212 BangladeshThese authors contributed equally to this work daggerPresent address Department of Medicine Massachusetts General Hospital Boston MA 02114 USADaggerCorresponding author Email jgordonwustledu


ugust 30 2020


Dow

nloaded from








The plasma proteome






ugust 30 2020


Dow

nloaded from


The gut microbiome










ugust 30 2020


Dow

nloaded from














Host effects





ugust 30 2020


Dow

nloaded from






ugust 30 2020


Dow

nloaded from
















ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from








The plasma proteome






ugust 30 2020


Dow

nloaded from


The gut microbiome










ugust 30 2020


Dow

nloaded from














Host effects





ugust 30 2020


Dow

nloaded from






ugust 30 2020


Dow

nloaded from
















ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from


The gut microbiome










ugust 30 2020


Dow

nloaded from














Host effects





ugust 30 2020


Dow

nloaded from






ugust 30 2020


Dow

nloaded from
















ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from














Host effects





ugust 30 2020


Dow

nloaded from






ugust 30 2020


Dow

nloaded from
















ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from






ugust 30 2020


Dow

nloaded from
















ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from
















ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from














ugust 30 2020


Dow

nloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from














ugust 30 2020


Dow

nloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from













ugust 30 2020


Dow

nloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





Conclusions










ugust 30 2020


Dow

nloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from







Gnotobiotic piglets








































ugust 30 2020


Dow

nloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from










ugust 30 2020


Dow

nloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from









CONTENTRELATED


REFERENCES





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from





on August 30 2020

httpsciencesciencem

agorgD

ownloaded from

MICROBIOTA Effects of microbiota-directed foods in gnotobiotic … · Andrei L. Osterman16, Md Iqbal Hossain 17, Munirul Islam , Nuzhat Choudhury , Shafiqul Alam Sarker 17, Sayeeda

Documents