<<

ARTICLE

https://doi.org/10.1038/s41467-019-13118-0 OPEN Characterization of antibiotic resistance genes in the of the microbiota

Yasmin Neves Vieira Sabino1, Mateus Ferreira Santana1, Linda Boniface Oyama2, Fernanda Godoy Santos2, Ana Júlia Silva Moreira1, Sharon Ann Huws2* & Hilário Cuquetto Mantovani 1*

Infections caused by multidrug resistant represent a therapeutic challenge both in clinical settings and in livestock production, but the prevalence of antibiotic resistance genes

1234567890():,; among the species of bacteria that colonize the of is not well characterized. Here, we investigate the resistome of 435 ruminal microbial genomes in silico and confirm representative phenotypes in vitro. We find a high abundance of genes encoding tetracycline resistance and evidence that the tet(W) gene is under positive selective pres- sure. Our findings reveal that tet(W) is located in a novel integrative and conjugative element in several ruminal bacterial genomes. Analyses of rumen microbial metatranscriptomes confirm the expression of the most abundant antibiotic resistance genes. Our data provide insight into antibiotic resistange gene profiles of the main species of ruminal bacteria and reveal the potential role of mobile genetic elements in shaping the resistome of the rumen microbiome, with implications for human and animal health.

1 Departamento de Microbiologia, Universidade Federal de Viçosa, Viçosa, Minas Gerais, Brazil. 2 Institute for Global Food Security, School of Biological Sciences, Queen’s University Belfast, Belfast, UK. *email: [email protected]; [email protected]

NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0

ithout immediate action to tackle the current and ResFinder13, Resfams14, and ARG-ANNOT15 varied according to escalating threat of antimicrobial resistance (AMR), it differences in specificity, sensitivity, and search method used by W fi is estimated that resistant infections may kill one each computational tool. ResFinder identi ed a total of 141 person every 3 s by the year 2050, raising the death toll worldwide acquired ARGs distributed in 72 rumen microbial genomes and to 10 million annually1. This antibiotic-resistance crisis can be across 11 antibiotic classes (Table 1). ARG-ANNOT detected 754 linked to several issues, including the overuse of these compounds genes from 10 distinct antibiotic classes in 93 genomes of rumen for medical and agricultural purposes, inappropriate antibiotic bacteria, while Resfams predicted 3148 sequences related to the prescribing, and limited discovery/development of novel and resistance of 9 classes of antibiotics in 430 rumen microbial effective antibiotics2. The widespread use of antibiotics in the genomes (Table 1; Supplementary Fig. 1). The most abundant agricultural sector plays an underrepresented role in the AMR ARGs detected by all three bioinformatic tools were related with context. Large amounts of antimicrobials are frequently used in resistance to beta-lactams (726 genes), glycopeptides (510), tet- livestock to prevent diseases and promote animal growth3. More racycline (307), and aminoglycosides (193). than 2 million kilograms of medically important antibiotics were Resfams detected the majority of the aminoglycosides and sold in the USA for use in in 2017, which corresponded to glycopeptides-resistance genes, while ARG-ANNOT identified 42% of the total used in food-producing animals, and ruminants the most beta-lactam-resistance genes in the rumen bacterial have been recognized as a potential reservoir of antibiotic- genomes. Fluoroquinolone, fosfomycin, nitroimidazole, and resistance genes (ARGs)4,5. trimethoprim were the antibiotic classes with the lowest number The rumen ecosystem is colonized by a complex and genetically of ARGs identified using ResFinder, Resfams, and ARG-ANNOT. diverse microbiota, in which dense populations of microbes often These putative ARGs were more prevalent among members of the exist in close proximity within biofilms6. Therefore, mechanisms phyla , , , and Actinobac- that lead to horizontal acquisition of novel genes could provide teria, but their abundances varied according to the bioinformatic competitive advantage for resource competition and facilitate the tool used for ARG detection (Table 1; Supplementary Fig. 2). exchange of genetic material between members of the rumen microbiota and also with allochthonous species that occupy the same site in the gastrointestinal tract (GIT)7–10. In addi- Phylogenetic distribution of ARGs. Although ARGs were widely tion, feeding antimicrobials through ruminant diets can select for distributed across the genomes of ruminal bacteria (Fig. 1), resistant organisms, potentially modifying the autochthonous resistance to specific antibiotic classes were more prevalent in ruminal microbiota11.Thesefindings corroborate with recent ana- some bacterial taxa, especially at the family and level. lyses of the ovine rumen resistome, which identified resistances to 30 known antibiotics, including high abundance of genes for dap- Table 1 In silico detection of ARGs in ruminal microbial 5 tomycin and colistin resistance, two clinically relevant antibiotics . genomes Ruminants represent a major source of animal protein for human consumption worldwide (through milk and meat pro- General features ResFinder ARG- Resfams duction), and cattle are raised in close proximity with humans, ANNOT such as in rural areas, particularly in lower-middle-income ARGs detected 141 754 3148 nations, where a great proportion of the livestock farms are small, Genomes with ARGs 72 93 430 family-owned properties. Nonetheless, little is known about the ARGs by antibiotic class/ occurrence and distribution of ARGs among the bacterial species function of the core rumen microbiome. Recently, hundreds of reference Aminoglycosides 10 26 157 genomes of cultured ruminal bacteria and have been Beta-lactams 11 446 269 made available through the Hungate Project (http://www. Colistin 0 0 0 hungate1000.org.nz/), representing ~75% of the genus-level bac- Fosfomycin 2 2 0 terial and archaeal taxa present in the rumen12. We therefore Fluoroquinolones 4 4 2 hypothesized that genome mining of the Hungate genomic Fusidic acid 0 0 0 resources could reveal the distribution and genetic context of Glycopeptides 22 76 412 MLS 22 44 44 ARGs within major ruminal species represented in the core Nitromidazole 3 0 0 rumen microbiome. Oxazolidinone 0 0 0 As such we analyzed 435 genomes of ruminal bacteria and Phenicols 3 10 19 archaea to search for ARGs using different computational Rifampicin 0 0 0 approaches. Identified ARGs were also evaluated for selective Sulfonamides 2 2 0 pressure, genetic organization, association with mobile genetic Tetracyclines 61 130 116 elements, as well as their expression in different rumen meta- Trimethoprim 1 14 0 transcriptome data sets. The resistant phenotype was confirmed for ABC efllux ––1749 fl –– some cultured representative ruminal bacteria from the sequenced RND e lux 165 Othersa ––215 Hungate1000 collection using in vitro testing and a novel inte- ARG’s by taxonomic category grative and conjugative element (ICE) associated with tetracycline = fi (n 32) 7 11 164 resistance was identi ed in the genomes of rumen bacteria. Our Bacteroidetes (n = 50) 30 67 199 findings shed light on the distribution and frequency of ARGs (n = 2) 0 0 4 among the main species of bacteria colonizing the rumen, thereby Firmicutes (n = 313) 89 233 2502 improving our understanding of the mechanisms underlying (n = 1) 0 0 2 antibiotic resistance within this microbial ecosystem. Proteobacteria (n = 22) 15 443 244 Spirochetes (n = 5) 0 0 27 (n = 10) 0 0 6

Results MLS Macrolides, Lincosamides, Streptogramins Detection of ARGs in ruminal microbial genomes. The number aARGs that could not be ranked in any of the above antibiotic classes/functions; (−) these and classes of ARGs detected in rumen microbial genomes using ARGs are not included in the ResFinder and ARG-ANNOT data sets

2 NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0 ARTICLE

L

a

c

h

L

L 8 n

a 9

L 3 7 a 3 0 o

c 7 4

a 0 3 c 1

Lachnosp s h

Lachnos 1 3

c h 0 0

Eubacte p 196

Lachnos n

h n 1 A 2

i L G70 L Lachnos o

n C r o Lachnos 4 B 1006 G277 D C134 a a a s - o s -

K -zl-G2 Lachnospirace P B c c S Tree scale: 0.1 c p

p s C242 -zl- -zl- e E h

N h X Lachnos i - -zl p Y i -zl

r

Anae E Lachnoc r E a n n a

i a E

r AGR2157

e o m Lachnos o m p c 8 m a c ATCC 255 rium ce irace p s e s u 6 e iracea u 1 s DSM 54 u Lachnoc e c m XB E-zl

b i i i

p e p irace p a r 4 n r 5 e piracea r a u Lachnosp a A m r e NLA i irace i

e e 1 e 0 a e Lachnos r e ovib c r m

t t t AGR2154 e NLA a a 2 1 e t

c c a b c ae NLA pirace e b c c i m lostridi R

a a e bacteri a a b a e a e bacteri r e iofor us M nse NLA

i b b l c tum NLAE tolera r b pirace a G a e bacteri a c iofor luloso u a d

io lipo r S t o

t c ioform l e bacteri e d Butyriv m e

e e A e bacteri e e ostridi e a t D d iofor

r a 0 a e a r e bacteri b e e bacteri b ticum NL Butyriv a i i d 3217 Sarc N i

p u i iracea e 5 e r e u a u a a a e bacteriu - i k irace c c m r c u M L m ami m c a c m citron Butyriv l Butyriv F n a l vens LD20 a e a m aer A m t i yticus LB200 t l u

e bacteri u r r r Butyrivi e u e m lavale u i S i N i i X h x i E m m clostri brio sp MC2021 u r u m ami na sp DS m FE2 r p p p irace N c

K e i p m clostri i B a u m - e bacteri u Butyriv u s A1 i u s AB2 s s z m clostri 1 brio fib s u C

4 w e bacteri m ND20 B m m o m NK4A o n o 1 u AC30 l u m clostri

- A Butyriv a 2 a ibrio fib n 2 n ophilum DS i 0 ibrio fib n G a u i ostridi m KH1 catena conto 07 t i l

1 t G 0 G h ostridiu K2B42 0 h 2 um RM5 3 t catena oro B h l brio fib m KH1 i n 2 u ostridi ucens DS

3 u c 0 0 c a AGR2135 8 0 c l u PX15 u 0 1 utyriv 4 ostridiu ophil 3 a l d 6 a a 8 28 a 8 l ibrio fib a a 1 ostridi Butyriv m NK3 l 18 1 risolve l l 0 1

L L M L ostridi LB2003 0 B u B l i 6 B ostridi Lachnosp a brio fib 0 7 l r m u gnav risolve 144 5 MC2003 T2 Dore isolve 11001 dore i u m P6A 6 Faecali i brio fib r ium sp WCC10 a P Faecal isolve m KH1 AC202 r TCC 192 Lachnoc ium sp N 17 Lachnoc ibrio fibr r ns r ae bacteri A A Lachnoc ltipar 14045 isolve M m C7 Lachnoc u a 20 ium sp KH u M r ns DSM 30 Lachnoc ltipar noschiza M D15d isolve n AB2 10710 r 12 Lachnoc u a ium ox r Butyriv s MD2001 n 3 Butyriv isolve Lachnoc A s P B19 8 sp P6 ltipar n ira m 4 B AR40 0 1 Oribacte u isolven s ND3005 p 004 Butyriv n 20 Oribacte ira m utyriv ultipar 2 5 s FE2007 Oribacterium sp FC n ira pecti e bacteri 0 ibrio sp bovis DS E ibrio sp M s WTE3004 Oribacte p a Butyriv ira m A Eubacter m A10 7 p bovis C6 12 i s ira m Phyla i brio sp N 1 Lachnospirace Butyriv B brio sp p m YRB Lachnos irace utyriv p Butyrivi TB Lachnosp AC20 ibrio sp F Lachnos acteriu 2001 Clostri 2 C2013 bovis S1b D X20 005 Lachnos um i i X m B brio sp V brio sp INll C bacteriu m MD2004 Butyriv PD20 Lachnos acterium bovis NK brio sp WCD2001 3005 u Butyrivi b ae bacterium HCP d Lachnos acterium bovis ium bei K C Lachnob b acteriu 2001 0 m ibrio sp OB Lachno b irace u Butyriv CD2006 2 ae bacteri brio sp 2005 a16 Lachno e bacteri sp jeri a D Butyrivi Lachno m C10 n pirace ae bacterium X u i ckii HUN1 Lachno brio sp AE e bacteri AD3 Lachnosp pirace Butyriv 235 irace rio sp M 2254 brio sp WC bacterium FD20p b Butyriv 0 02 iracea e bacteri Lachnos p a 17 Actinobacteria Butyriv i 3 brio sp LC AC2005 3 4 Lachnos -2 Butyrivi 004 2 butyrivi brio sp 49 V ibrio sp Lachnos irace p rio sp UC1 C E b A Butyriv ibrio sp 2006 Lachnos brio sp WCD3 butyrivi brio sp OR37 AR14 s DSM 10 Pseudo o n Butyrivi M 3010 rio sp B2005 butyrivi b -9 Butyrivi i Lachnos o V brio sp N AE rio sp ivora n Pseud obutyrivi AC B brio sp F 3 utyriv 006 butyrivi Pseud o brio sp minis M 9787 C 002 butyrivib rio xyla u S Butyriv i Pseud o brio sp 2007 seud CS006 P rio r Butyriv X obutyrivibrio sp NOR37b PD20 ibrio sp Pseud butyrivib rio sp C4 minis D Butyrivibrio p V o u CB20 Bacil ibrio sp F Pseud butyrivi 0 o rio r Firmicutes Butyrivibrio sp 6 seud b 1 AE P lus sp MB2 0 butyrivib rio sp YE44 2 1 o b 1b 032 2002 B201 F C Pseud butyrivi roteoclastic S014 Butyriv Pseud butyrivi 7 seudo o minis C 11 AE3009 021 P brio sp L Butyriv i ibrio sp NC rio ru AD201 Butyrivi brio sp INllu Butyriv s P6B7 Pseud rio sp JWminis ibrio p Butyriv b u brio sp butyrivib ibrio sp rio r minis HUN009B3002 11 Butyriv r a14 Pseudobutyrivi b A Butyrivi oteoclastic butyrivi o rio ru AE Su6 Pseudo b ntium F Butyriv a um C6 brio ibrio sp obutyrivi B 2 Pseud 88 utyriv 015 p i u butyrivi rumin Proteobacteria roteoclasticbrio sp s P18 o N269 Y Pseud e bacteri ibrio p AB 2008 B 3 seud S XB 001 P iracea Y 9 roteoclastic uminantiumntium 23 HU 2 u B1 r a Butyriv s FD20070 Eubacterium C20 01 rium A 31 Butyriv rumin m Butyriv i Lachnosp u brio sp us B316 ium r ae bacterium AC20 19 ibrio h Eubacte um ibrio sp L AE irace e bacteri ubacte p a Butyrivi ungatei X 3003 E m LC20 001 irace e bacteri E2 Butyriv B2008 p 4 Lachnos m N brio h iracea BD2006 e bacteriu ibrio sp M Lachnos a AC201 1801 ungatei N m M Butyriv irace B14 Bacteroidetes Lachnosp m DS Butyrivi 55 m P6 Butyriv ibrio sp K4A153 piraceae bacteriue bacteriu u Lachnosp a brio h harolyticu i A irace 22 brio sp INl E Lachnos p ae bacteri ungatei D 2005 irace 004 Butyriv Lachnos p um polysaccs DSM 55 3 SM 14810la21 9D Butyriv umini m V 4 ibrio sp INll Lachnos lostridi u 0 ibrio sp culum r Lachnoc a um NC20 pGA-8 a18 ae bacteri m VCB20 irace Acetitom p e bacteri a e bacteriu 06 Lachnos irace C4 Fibrobacteres 007 P Lachnosp eptococcacea3 AB rans KH Peptostr rium sp uvativo r rans i6 Eubacte ium py KHGC13 r ruvativo rans f135 Eubacte rium py l ruvativo ellii Ca C Eubacte rium py us erobi Eubacte ptococcus russ

Peptostre ptococcus ana Fusobacteria notii LM2 2 Peptostre e eptococcus sp D1 m VA m mang Peptostr bacteriu ostridiu KPPR-9 Peptocl ptococcaceae licus Peptostre Prev obacter glyco otella rum rispor nale 6191 Ter Prev inico GR2141 o ium intesti A tella ru la RM4 averis l-G419 m Clostrid E-z Prev inicol ium cad A otella ru a D31d d Clostri averis NL minico Euryarchaeota Prev la KHP1 nis B3 otella ru cad Prev gidicar o minico dium al AGR2140 tella ru l 6 m a 23 Clostri ricum 5 inico AGR21 Prev la B dium buty ficum otella ru PI-1 Clostri i m 62 raputr 24773 Prev inico ium pa SM o la BPI-3 d inis D tella ru 4 Clostri m rum minico -zl-C224 la Ga6B6 niclasticu ru Protei um NLAE m inico dium cochleari M 17049 la AR32 Clostri Prev ndense DS otella sp KHP dium lu Prev 7 Clostri rium R7 otella sp FD3 diales bacte 2008 004 Clostri E Prev rium WT otella sp P6 001 diales bacte m AB4 B4 Clostri Prev e bacteriu otella sp nococcacea 03 P6B Rumi YAD30 Prev 1 e bacterium otella sp M nococcacea A2016 Rumi m YRB3002 Prevotella ru e bacteriu minicol nococcacea a KHT3 Rumi m AE2021 Prev bacteriu otella sp N nococcaceae E3005 Rumi m KHP2 Prev e bacteriu otella sp T Ruminococcacea c2-24 P7 e bacterium Prevotella sp nococcacea TF2-5 Rumi mii YE282 Prevotella br ococcus bro evis P6B Rumin 11 B98 Prev m NK3 otella brevis A nococcaceae bacteriu TCC 19188 Rumi r sp PC13 Prevotella sp BPI-148 Oscillibacte e bacterium D5 Prevotella sp BPI-145 Ruminococcacea YE71 bacterium HUN156 Ruminococcus sp YE78 Prevotella sp TC2-28 Ruminococcus sp A76 Prevotellaceae bacterium KH2P17 Ruminococcus sp NK3 bacterium FB2012 Prevotella sp KH1P2 Lachnospirace Prevotella sp KH2C16 ae bacterium AD3010 Ruminococcus sp FC2 Prevotella sp HUN102 018 0 Ruminococcus albus 8 Prevotella ruminicola AGR216 Rumin 11370 ococcus albus 7 DSM Rumi 001 nococcus albus SY3 FB3 Ruminococcus albus antii C21a AD20 Prevotella bry 13 Ruminococcus alb antii KHPX14 us AR67 Prevotella bry Ruminococcus alb antii B14 us KH2T6 Prevotella bry Ruminococcus sp HUN0 acterium Ga6A2 oidales b 07 Bacter Ruminococcus flavefaci m Ga6A1 acteriu ens YAD20 b Ruminococcus flavefaci 03 AE-zl-G339 isolvens NL Rumi ens YRD ides nococcus flavefacie 2003 Bactero AE-zl-C202 Rumi ns MA20 isolvens NL nococcus flavefaci oides xylan 07 Bacter AE zl C500 Rumi ens MC202 ides ovatus NL nococcus flavefacie Bactero AE-zl-C57 0 Rumi ns XP oides ovatus NL AR20 nococcus flavefaci D3002 Bacter oides sp Rumin ens 17 Bacter PR3 ococcus flavefaci on KP omicr Rumi e C579 nococcus flavefaci ns FD-1 ides thetaiota E-zl- o Lachnos Bacter on NLA e omicr AR29 pirace ns AE3 ae bacteri 010 ides thetaiota oides sp Rumi o Bacter nococcus flavefaci um MC20 Bacter -zl-G202 AE Rumi 17 tus NL nococcus flavefaci e -zl-C104 ns YL2 ides vulga E Rumi 28 Bactero ium NLA R12 nococcus flavefaci e 3 ns ATCC 192 eae bacter ium KH Ruminococcus flavefacie 08 monadac eae bacter CP3RA ens phyro 3 Rumi Y1 Por monadac 3R9 nococcus flavefaci phyro P ns ND2009 Por eae bacterium KHum KH Eubacte monadac ae bacteri rium cal ens hyro e m WCE2008 Erysi SAb6 Porp p lande 7 monadac acteriu YLM1 elotric ri NL hyro eyae Erysip haceae bacte AE Porp idales b elatoclostr -zl-G225 bacter oll 16632 Bactero Erysi rium N obrevi p idiu eyae DSM m M1 elatoclost m K3D Methan Sha innocuu 112 bacter oll minantiu rpea azab ridiu inii SH m m NLA obrevi acter ru Sha innocuu rpea azab uensis E-zl- Methan brevib acter wol i JH1 m C381 o b Shar K NL brevi orean H2P10 AE Methan M9 pea azab uensis KH -zl-G1 e S Methano lera Sha 97 bacter bovisk 43 rpea azab uenensis D 1P5 obrevi Kandleria vitu SM 166 u brevibacter mil ensis DS SM 204 Methan o lerae D Kand 0 Methan ii DSM 11978 leria vitu lina WCC7 M 18934 6 acter mi 11855 Kand obrevib leria vitu lina WCE2 a sp DSM M 1539 Kand acter gottschalk 5 Methan b 2 leria vitu lina KH osarcin 01 obrevi obile DS Kand 1 m ryantii B 4 Methan leria vitu lina MC T7 Methan obium D1002 Kand nema b XB omicr leria vitu lina D 3001 Trepo yantii 4A124 Lactobaci S Methan 6A8 lina M 20405 nema br yantii NK Lactobaci o r llus ru S3b rep T Lactobaci nema b l m o onema sp SMC 2985 85 lus ru inis DS rep rep T T Pedi m 2 l inis WC1T17M M o lus brev 20403 ogenes S Lactobacicoccus acidi harophilum Dn is R7B Lactobacil AG48 nogenes HKP 82 llus planta lactici obacter succi Lactobaci inicola A A Treponema Fibrsacc SM 279 P lus mucos GR20 obacter succi ns Lactobacil rum llus mucos Fibr R1 AG30 myces rumnicola D P Lactobaci i ae a K 17 A H lus mucos GR63 Actino K S taphyl ae KHP myces denticole llus mucos myces rum rium Staphyl KH10 o a yces nasicol m coccus epid e KHP X Actino Actino m Lachnos 11 ococcus epide ae WCC8 C Actino ceae bacte as sp KH9 Bacil 15 a e bacteriu M4R57 07 p e T B lus cereusirace rmi Ruani lomon V acil dis a ococcineaCellu Enter lus lichen e bacteri rmi AG42 dis NLA Micr m sp MB30 A17 Enter K ococcus faecalis 6 PR7A u coccus luteus iformiss V m M o acteriu m P6 A13 Enter o E s NLAE zlu G260 4 coccus faecalis -zl-G2 Micr e 9 A2 ionib Enter o 020 p coccus mun 3 TM3R7 9 Pro e bacteri M 202 Enter o rium acn a S coccus sp a 43194 92 Enter 8A 0 o 8 coccus gall V cteriace minis D E d TM1R Cutibacte m vitaeruminisu Ga6 nter ococcus casselifl tii C2 u a rugos Lactococcus ga KPP DSM 20 o ioniba m AGR2145 coccus casselifl 9 p R-6 1 m vitaer userell m GR2158 212 Lactococcus lactis 5 inaru Pro u ongu A 0 2 Pra l 3 S ongu treptoc m Corynebacteri l 7 SKF1 bacteri inum 3 Streptoc e DSM 2 a pseudo m 2 vus NLA SM 204 P S rvieae m pseudo treptoc o a Coryn choer ccus sp NL vus NLA m philu AGR21 S o o m 9 treptoc ccus henryi M acteriu 8 E b boum D Calf96 equinu 1 79 -zl- breve R o 1 acterium acteriu m m m ccus equin b b therm longu S E C414 treptoc o -zl-G26 Bifido m m ccus equinu A ifido 6492 Streptoc E B acteriu M 087 -zl- Bifido acteriu 0 b b m bifidu m DSM 64 Streptoc A-4 acteriu C 8 acteriu 3 occus equin us B315 503 b b P Streptoc Bifido Bifido o acteriu inantiu icum DS 19 Streptoc ccus equin Bifido b ta WCP15KH1 6 s JB1 Bifido a o ntis DSM 2 p Streptoc ccus equin e s o s Bifido m rum meryc a Streptoco ccus equin A u G46 m Streptococcus gallol o s pGA ccus equin umbon u sp KH3B4 43 Streptoc occus equin s pR-5 acteriu adolesc a b ata DSM 22 20476 8 6 u acteriu m Olsenell sp KH2P3 9 Streptoc -7 a ccus gallo s ES1 Streptoc u A-3 s Sb04 Bifido Olsenella Streptoco P occus gallol u acteriu umbon Olsenell -zl-G4 i Bifidob a Streptoc s HC5 E o Olsenell s DSM 21 u n A Streptoc ccus gallolyticus VTM s GA-1 zl C269 o l ifidob KPR-6 Streptoc ccus gallo yticus LMG155 B i NL G534 e ichia col AE Streptoc yticus VTM Olsenell r ccus gallo inireducens DSM occus equin -C285 C84 E S 5 yticus E-zl- Streptoco treptoc o A ccus equin E-zl R-2 Streptoc o m detoxifica Esche E-zl- EL5 heliotr lla sonn bacter sp A P S ccus equin l u A B o yticus a e o 0Z treptoc V cter sp NL Streptoc ccus equin 3 o lyticus TM1R29 3R24 a ilis NL Streptoc ccus equin 7 b 2 Shig M 22022 -10 S u Enter ilis NL ccus equin Slacki bacteri b S V S treptoc o s V anii NL ccus sp 45 u TM o C Streptoc treptoc o S 2 Citrob ducens K lvens H5 s C277 V A Streptoc ccus equin N033 R47 ducens M o o u TM s mira Streptococcus equin 1 u ogenes 1 ccus equin s Sb13 Streptoc o R27 1361 u s mira morg ccus equin Denitr u lvens 22B Streptoc o s Sb05 3 icipr M S u R42 o lus RM87 1652 57 occus equin ccus equin icipro n lvens Desulfoto s Sb18 ducens D M 2578 0 74 treptoco A o u Prote o 1 1740 o S 7 A s Sb20 cida o ccus equin M S cida Prote M7 Succini ccus equin dextrinoso 4 Megam uccin u KHC7 Allis occus equin lus DS 2 Megas icipr o 4 s 2B i M 1 o Megas m us S n 1 M UN048 A S ccus equin S minococcu

M Morganella 4 S e us 5 8 inococc ccus equin s DS n 9 e acillus succin dextrinos 0 e 3

6 o allii e g

e 137 D 0 vibri - i m TCC 27 i us M 0 asfia succi n 421 a i g clasticum t o 1 l g nella h l T a T

l 0 e dextrinos s clasticum b17 A S V e S 1 6 H e B yloph M 2872 e a aculu us MPR1 0 s enes DSM 1 onas sp Ca o u n 3 C n phaer antium C3 Z n b10 s r s Basfia succin C 2 u la crassa D g S p m M o o o p

rum H H E o

n m o p s MPR2 A h

W u D P acter sp DS k o A m v haer m

u m K h rio le as sp FC4001 A m i a s H24

antium L e u lfurica i T G R4 Actinob obacter amylophi t N u b ivibri u a Succin b o

o TCC 1256 e i m m p o s S u n l i u n n s fermentan t l r m e staminifo n

a n s p r u n a elsdeni u s Basfia succi i s fermentan antium AC20 A antium m a i rumi n o a u r a s a

t a Alysiel i us a s a n n u lfurican s n j YE01 a t r I croph i s i a elsdeni n s s e s t umin u omon s r A Succinivibri as rumi a n n e l umini i a p antium D l b n n a obacter a a S Rumi m n b s r a R3 Succin la succino l

Acinetob s rumi n l u s antium o l a n n l o d u o i n m R b09 rio des n f98-2 u

a r i d

m n v n o e v is DSM 2 b u

i d antium M m m Desulfovi r e i s rumi s rumi i s i n m s DSM 923 s

i i s Sele omon n m r o u

n n ium ne n s DSM 11005 a a m 5 i i J1 a mans D s u r rio des r i oline n o n 8

a r i i D i u Rumi 0 n a i i b Y r n s pGA4 s rumin e s - T81

n

l D o D S

W s omon n 1 s WCC6

e E

a a

s rumi t s e l

o a n 4 i M S S m n a u 3 a Sele e S n - omon s rumi omon M o M o m m 4 1 n

S o 1 Desulfovi n a n 2 n o

o

m 54 3 S 2 A e 1 Sele n m omon l o

5 m M 152 0 3

e o

B e 6 n l n

omon 9 o Desulfovi 4 Fusobacte 8 Sele n

Sele 3 e S e n 4 n

l 6 1 omon e

0

l S e 1 e 0

n l

0

e Sele

e S

2

S 3

Sele S 0

Sele

Fig. 1 Antibiotic-resistance distribution in ruminal microbes. The 16S rRNA gene sequences used to construct the tree were obtained from the Hungate1000 collection and aligned using RDP Aligner. The phylogenetic tree was generated with the Maximum-Likelihood method using FastTree (1000 replicates), and visualized and annotated using iTOL. Black circles outside the tree represent the genomes, in which ARGs were identified from at least two bioinformatics tools applied in this study

Members of the Proteobacteria phylum showed a high proportion rRNA gene phylogeny (Fig. 2). Majority of the genomes (69.2%) of genomes harboring resistance genes, with ARGs being detected showed resistance to one antibiotic class, with a dominance of in all genomes of the Enterobacteriaceae family. In the Bacter- tetracycline-resistance genes, which were distributed in bacterial oidetes and Actinobacteria phyla, ARGs were concentrated in the genomes within the phyla Actinobacteria, Firmicutes, Proteobac- genus Bacteroides and Bifidobacterium, respectively. In the Fir- teria, and Bacteroidetes. Phylogenetic analysis of the most micutes phylum (which corresponded to the largest number of abundant tetracycline-resistance genes (tet(W), tet(Q), and tet genomes analyzed in this study), resistance genes were detected (O)) revealed that tet(W) was highly conserved across different across members of the Lachnospiraceae (Lachnosclostridium, taxa of ruminal bacteria, with a minimum sequence identity of Clostridium, and Blautia) and Veillonellaceae (Selenomonas, 94.9 and 99% coverage in at least 28 ruminal bacterial genomes Megamonas, and Megasphaera) families, as well as in all species of (Supplementary Figs. 3–5). the genus and among strains of Staphylococcus Moreover, as shown in Fig. 2, metronidazole and fosfomycin epidermidis. No antibiotic-resistance genes were detected in resistance were only identified in the genus Prevotella and in two genomes belonging to the Spirochetes, Fibrobacteres, Fusobacteria, strains of Staphylococcus epidermidis, respectively. Staphylococcus and Euryarchaeota phyla using Resfinder and ARG-ANNOT, but epidermidis together with Citrobacter sp. NLAE-zl-C269 were the it should be noted that the number of genomes representing these only species harboring resistance genes to quinolones. Escherichia taxa were much lower compared with the other phyla analyzed in coli PA-3 was the single ruminal genome showing the highest this study. Although Resfams detected some potential ARGs in number of ARGs (n = 5), including tetracycline resistance, these microbial taxa, the majority were ABC efflux pumps. The beta-lactam resistance, aminoglycoside resistance, trimethoprim ABC efflux pumps represent one of the largest protein families in resistance, and sulfonamide resistance. Bacteroides ovatus NLAE- microorganisms, contributing not only to reduce the intracellular zl-C500 was the only species in Bacteroidetes-possessing genes for concentrations of toxic compounds but also to the influx of sulfonamide resistance. Beta-lactam-resistance genes were con- substrates and other nutrients, and are not necessarily related to centrated in the family Enterobacteriaceae, in the genus antibiotic resistance16. Bacteroides, and in two clades harboring the genera Selenomonas, The distribution of the ARGs in ruminal microbial genomes Staphylococcus, Succiniclasticum, and . Proteus mirabilis was represented by the antibiotic class and according to the 16S NLAE-zl-G534 and Proteus mirabilis NLAE-zl-G285 shared the

NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0

Tree scale: 0.01

0.987

0.998 0.89

1 0.937

0.976

0.984

1

0.997

1

0.938

1

1 1

1

0.999

0.875

1

0.739 0.997

0.9

1

0.987

0.912

0.989 1 1

0.996

1

0.864 0.961

1

0.927 0.985

0.999

0.866

0.94

0.989 0.985 0.993

0.964

0.883

1

0.973

0.959

0.999

0.918

0.984

1 0.912 0.876

0.995 0.976

1

0.888

0.965 0.986 1

1 1

0.999

0.986

1

0.853

0.977

0.947

Fig. 2 Distribution of ARGs in species of ruminal bacteria. The distribution of the resistance genes detected by at least two bioinformatic tools used in this study is presented in front of the tree as filled colored boxes. The 16S rRNA gene sequences used to construct the tree were obtained from the Hungate1000 collection and aligned using RDP Aligner. The tree was generated with the Maximum-Likelihood method using FastTree (1000 replicates) and visualized and annotated using iTOL. Only bootstrap values >0.7 are shown

4 NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0 ARTICLE

a Blautia schinkii DSM 10518

5000 10,000 15,000 20,000 25,000

Soj SpoJ VirD4 TetW VirD1 RepA Dnal VirB6 TraE ATPase TnpW Sigma 70 protein Mobilization Recombinase Recombinase Phage proteins Endonuclease Methyltransferase Methyltransferase

Transcriptional regulator Acessory genes Regulation

Mobilization Tetracycline resistance

Recombination Unknown function

b 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10,000 10,500 11,000 11,500 12,000 12,500 13,000 13,500 14,000 14,500 15,000 15,500 16,000 16,500 17,000 17,500 18,000 18,500 19,000 19,500 20,000 20,500 21,000

Blautia schinkii DSM10518.fasta –2000 –1500 –1000 –500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10,000 10,500 11,000 11,500 12,000 12,500 13,000 13,500 14,000 14,500 15,000 15,500 16,000

Butyrivibrio hungatei DSM14810.fasta –1000 –500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10,000 10,500 11,000 11,500 12,000 12,500 13,000 13,500 14,000 14,500 15,000 15,500 16,000

Lachnoclostridium aminophilum DSM10710.fasta –3000 –2500 –2000 –1500 –1000 –500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10,000 10,500 11,000 11,500 12,000 12,500 13,000 13,500 14,000 14,500 15,000 15,500 16,000 16,500 17,000 17,500 18,000

Lachnoclostridium citroniae NLAE-zl-G70.fasta –4500 –4000 –3500 –3000 –2500 –2000 –1500 –1000 –500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10,000 10,500

Lachnoclostridium lavalense NLAE-zl-G277.fasta –3500 –3000 –2500 –2000 –1500 –1000 –500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10,000 10,500 11,000 11,500 12,000 12,500 13,000 13,500 14,000 14,500 15,000 15,500 16,000

Pseudobutyrivibrio sp. NOR37.fasta –2500 –2000 –1500 –1000 –500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10,000 10,500 11,000 11,500 12,000 12,500 13,000 13,500 14,000 14,500 15,000 15,500 16,000

Pseudobutyrivibrio sp. OR37.fasta

Fig. 3 ICE_RbtetW_07 structure and conservation. a ICE_RbtetW_07 structure in Blautia schinkii DSM 10518. The genes of the ICE are represented by arrows classified and colored in modules according to their function. The essential modules of the ICE machinery are represented in orange, blue, and pink, for mobilization, recombination, and regulation genes, respectively. The ICE is located in scaffold 17 (nucleotide positions from 88079 to 111915) of the Blautia schinkii DSM 10518 genome. b Alignment of the ICE_RbtetW_07 from ruminal bacterial genomes Blautia schinkii DSM 10518, Butyrivibrio hungatei DSM 14810, Lachnoclostridium aminophilum DSM 10710, Lachnoclostridium citroniae NLAE-zl-G70, Lachnoclostridium lavalense NLAE-zl-G277, Pseudobutyrivibrio sp. NOR37, and Pseudobutyrivibrio sp. OR37. Lines connecting blocks with identical colors represent the aligned regions, and show synteny or gene rearrangements. The tet(W) gene is represented in blue in the center of the figure same pattern of resistance to tetracycline and phenicol, while involved in bacterial conjugation, such as relaxases, plasmid Staphylococus epidermidis AG42 and Staphylococus epidermidis mobilization proteins (e.g., MobC), and conjugation proteins NLAE-zl-G239 harbored highly homologous resistance genes for (e.g., TraE, TraG/TraD). In addition, genes enconding phage beta-lactams, quinolones, and fosfomycin. proteins (e.g., PhiRv2) were detected in the flanking regions of the tet(W) genes in ten ruminal microbial genomes (Supplementary Data 1). Nonetheless, the scaffolds carrying the tet(W) genes ARGs genetic context. The genetic locus of the tet(W) genes as lacked complete mobile elements showing the structural organi- well as their flanking regions were analyzed to investigate their zation commonly found in plasmids, phages, and transposons. mobility potential. For this, the 28 genomes harboring tet(W) A more detailed analysis of the sequences flanking the tet(W) genes detected by ResFinder were evaluated in sequences of up to gene revealed that at least seven genomes contained the complete 4000 bp within the scaffold where tet(W) was located. Genes machinery required for a functional ICE, despite the fact that no encoding a putative methyltransferase protein were detected such genetic elements were found when the bacterial genomes adjacent to or located immediately upstream or downstrem of the were screened for using ICEberg17. The genetic organization of tet(W) gene in 18 genomes. From these, the gene encoding the the ICEs identified in the genomes of ruminal bacteria, as protein under the accession number D1PK82 [https://www. exemplified for Blautia schinkii DSM 10518 (Fig. 3a), contained uniprot.org/uniprot/D1PK82] was the most prevalent, and putative genes encoding for conjugation proteins (e.g., TraE), showed 98.9–100% sequence identity across the genomes of secretion proteins (e.g., VirD4), proteins controlling the cellular ruminal bacteria. In seven of these genomes, a gene encoding the cycle (e.g., RepA, DnaI, σ70), and recombinases, in addition to Maff-2 protein, which is known to be involved in bacterial other accessory proteins commonly found associated with these resistance to tetracycline, was also found flanking the tet(W) gene. mobile elements (e.g., ATPases, methyltransferases, transposons, Genes encoding transposon proteins (e.g., TnpW, TnpV, Tnp, and phage-related proteins). The hypothesis that a novel ICE TnpX, and ISPsy9) were identified flanking the tet(W) gene in 10 (named here as ICE_RbtetW_07) could be mediating tetracycline ruminal microbial genomes, and 22 out of the 28 genomes har- resistance in the genomes of rumen bacteria was reinforced by the boring tet(W) genes also had genes encoding proteins potentially presence of conserved elements (reversed organization in three

NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0

LNB a 5

4

3

10,037 2 8000 6000 5000 1 4000 3000 2500 0 0 50 100 150 200 250 300 350 400 450 500 550 600 650 2000 b 5 1500

1000 4

800 3

600 dN/dS 2

400 1

200 0 0 50 100 150 200 250 300 350 400 450 c Fig. 4 PCR detection of the ICE_RbtetW_07 in Blautia schinkii DSM 10518. 5 The expected amplicon size of the region from the virD4 gene to the tet(W) gene was 4794 pb. L: 1 kb HyperLadder, N: negative sample, and B: Blautia 4 schinkii DSM 10518 3 genomes) in the ICE carrying the tet(W) genes on ruminal bacterial chromosomes (Fig. 3b). Moreover, phylogenetic analyses 2 indicated that the ICE_RbtetW_07 had low sequence homology to other integrative conjugative elements previously reported 1 (Supplementary Fig. 6). 0 Detection of the ICE_RbtetW_07 in B. schinkii DSM 10518.To 0 50 100 150 200 250 300 350 400 450 500 550 600 650 confirmthepresenceoftheICEcarryingthetet(W) genes in rumen Site bacteria, Blautia schinkii DSM 10518 was grown in the batch cul- Fig. 5 Selection pressure analysis in tetracycline-resistance genes. The ture, and a fragment of 4794 bp covering the region from the virD4 d /d ratio was calculated for the tet(W) (a), tet(Q) (b), and tet(O) genes fi N S gene to the tet(W) gene was PCR ampli ed from the genomic DNA (c) using JCoDA. The lines in each panel represent dN/dS scores by site. (Fig. 4). The amplicon was subjected to shotgun sequencing per- The phylogenetic tree generated by JCoDA for dN/dS analysis was formed on the MiSeq with 300 cycles using the next-Flex Illumina reconstructed using the aligned protein sequences of the ARGs and workflow, and the assembled reads confirmed the genetic organi- Maximum-Likelihood approach with 100 replicates. Source data are zation of the ICE (type IV secretion system gene adjacent to the provided as a Source Data file tetracycline-resistance gene) predicted in the Blautia schinkii DSM – – – 10518 genome (accession number PRJNA223460, [https://www. 141 265, 310 341, and 600 639, with an average dN/dS ratio per ncbi.nlm.nih.gov/bioproject/223460]) with 100% of nucleotide site of 1.9, 2.6, and 3.1, respectively. For the tet(Q) and tet(O) identity and 99.7% of coverage (Supplementary Data 2). genes, our analysis indicated 25 and 17 amino acid sites with positive selection pressure, respectively (Fig. 5b, c). Nonetheless, tet Selection pressure analysis of resistance genes. Because the the overall dN/dS ratio calculated for these genes was <1, and the tetracycline-resistance genes tet(W), tet(Q), and tet(O) were likelihood ratio test data (LRT) for the substitution models did highly abundant and broadly distributed in the genomes of not present statistically significant difference at 95% confidence ruminal bacteria, we investigated if these genes are evolvable and level, indicating absence of positive selection for tet(Q) and tet if genetic variants are being selected for in the genomes of (O) genes. To confirm that the tet(W) gene is under positive ruminal bacteria. For this purpose, the number of non- selection, neutrality tests were applied in these ARG sequences. ’ = − = synonymous per synonymous substitutions (dN/dS ratio) was The results of Tajima s D (D 1.65)andFuandLitest(D* calculated to determine the codon substitution rate, which could −2.22. p <0.05; F* = −2.39, p < 0.05) corroborate with our indicate the evolution of the gene in ruminal microbial genomes. previous findings. Ouranalysisconfirmedthepresenceofpositiveselectionpres- sure in the tet(W) gene, with a dN/dS ratio > 1 for 36 out of the 639 amino acid residues in the protein (Fig. 5a). Sites showing Confirmation of resistance phenotypes. To validate the resis- positive selection were mainly clustered in three distinct regions tance phenotype that was computationally predicted in the gen- of the TetW protein, located between amino acids positions omes of ruminal bacteria, 26 pure cultures matching the genomes

6 NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0 ARTICLE

Bacteria BLA AMG TET VAN CHL MAC CLD ruminis DSM 5522 Bifidobacterium adolescentis DSM 20087 Bifidobacterium boum DSM 20432 Bifidobacterium merycicum DSM 6492 Bifidobacterium pseudolongum DSM 20092 Bifidobacterium thermophilum DSM 20212 Blautia schinkii DSM 10518 Butyrivibrio fibrisolvens DSM 3071 Butyrivibrio proteoclasticum B316 Butyrivibrio sp. M55 Lachnoclostridium aerotolerans DSM 5434 Clostridium intestinale DSM 6191 ResFinder Clostridium lundense DSM 17049 ARG-ANNOT Lachnoclostridium aminophilum DSM 10710 In vitro Lachnoclostridium polysaccharolyticum DSM 1081 Resfams Lachnospira multiparus D15d Lactobacillus ruminis DSM 20403 Megasphaera elsdenii T81 Mitsuokella jalalundinii DSM 13811 Olsenella umbonata DSM 22619 Prauserella rugosa DSM 43194 Proteiniclasticum ruminis DSM 24773 Ruminococcus flavefaciens FD-1 Selenomonas ruminantium DSM 2872 Sharpea azabuenensis DSM 18934 Sharpea azabuenensis DSM 20406

Fig. 6 Concordance between antibiotic resistance detected in silico and in vitro. Twenty six bacteria were analyzed both in silico and in vitro in this study. Colors represent the approach used to predict the ARG or to detect the resistance phenotype of each bacteria. BLA (beta-lactam); AMG (aminoglycoside); TET (tetracycline); VAN (vancomycin); CHL (Chloramphenicol); MAC (macrolide); CLD (clindamycin) analyzed in silico were selected for further in vitro characteriza- lactam, macrolide, tetracycline, and vancomycin) detected in tion. Our in silico analysis predicted 57 resistance determinants in rumen bacterial genomes to 15 rumen metranscriptome data sets the ruminal bacteria that were selected for in vitro testing. of dairy and beef cattle and . At least one resistance gene Minimum inhibitory concentration (MIC values) were deter- from each antibiotic class was expressed in one or more meta- mined using the Epsilometer test (E-test) method18, and 18 transcriptomes (Fig. 7). In general, tetracycline-resistance genes resistance phenotypes were confirmed in vitro (Supplementary presented the highest level of expression, mainly the tet(O), tet Table 1). From this, eight ARGs (~44.0%) were detected/con- (Q), tet(W), and tet(37) gene. Moreover, individual genes varied firmed both in silico and in vitro by all approaches used in this in their expression levels between different data sets, being more study (Supplementary Fig. 7). ARG-ANNOT predicted 14 ARGs prevalent in sheep and beef cattle. Some of the ARGs identified in in the 26 cultures of ruminal bacteria tested in vitro, and 10 the ruminal microbial genomes, such as strA, cepA, ermB, tet(B), resistance phenotypes (71.4%) were confirmed in the E-test and vanC, among others, were not expressed in any of the assays. ResFinder and Resfams predicted 14 and 53 ARGs in the metatranscriptome data sets investigated in this study (Fig. 7). bacterial cultures, and resistance to 9 (64.3%) and 17 (32.1%) of these antibiotics were confirmed phenotypically (Supplementary Fig. 7). Therefore, resistance phenotypes predicted by ResFinder Discussion and ARG-ANNOT agreed, for the most part, with the in vitro The hypothesis that livestock animals represent a reservoir of results, while Resfams predicted several resistance mechanisms ARGs is not new, and previous studies have demonstrated that that could not be confirmed by other computational or in vitro bacteria isolated from fecal samples of monogastrics and rumi- methodologies used in this study. nants harbor genetic elements that can confer resistance to In addition, tetracycline resistance was detected simultaneously clinically used antibiotics19. However, less attention has been by at least three approaches in 69% of the genomes and bacterial given to the occurrence of ARGs in the rumen ecosystem and cultures, providing further evidence that tetracycline resistance is their potential to be transferred to commensal and pathogenic widespread in the rumen ecosystem. Moreover, some resistance bacteria. Microorganisms colonizing the rumen also disseminate phenotypes were confirmed by our in vitro assays even when to the environment through animal saliva during rumination and the presence of resistance determinants were predicted by only one due to the flow of rumen microbial biomass to the omasum, bioinformatic tool. This was observed for vancomycin resistance in abomasum and to the distal parts of the GIT, being released Clostridium and Lachnoclostridium strains and in Lactobacillus through fecal discharge into the soil20. ruminis DSM 20403, for beta-lactams in Mitsuokella jalalundinii In this work, we investigated the occurrence and distribution of DSM 13811 and Selenomonas ruminantium DSM 2872, for ARGs in hundreds of ruminal microbial genomes made available aminoglycosides in Bifidobacterium pseudolongum DSM 20092 through the Hungate1000 project12. Our results indicate that and for macrolides in Prauserella rugosa DSM 43194 (Fig. 6). ARGs are widely distributed among ruminal bacteria, with genes conferring resistance to tetracycline being frequently detected in the genomes of several species. The regions flanking the Expression of ARGs in metatranscriptomes. To confirm if the tetracycline-resistance genes showed a conserved pattern in an ARGs detected in ruminal microbial genomes are expressed in the apparently novel integrative and conjugative element (ICE), rumen ecosystem, we aligned the genes conferring resistance to suggesting dissemination of antibiotic resistance through hor- the five most abundant antibiotic classes (aminoglycoside, beta- izontal gene transfer in the rumen microbiome. In addition, we

NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0

excisionases, and recombinases, while genes encoding proteins involved in DNA transcription are found in the regulation module. Moreover, several accessory genes, including methyltransferases, transposons, phages, and plasmid related proteins, in addition to 27 SRR8050517 SRR8050516 SRR8050514 SRR3257011 SRR3256999 SRR5805554 SRR5805529 SRR594654 SRR3169847 SRR3169851 SRR6201277 SRR6201276 SRR873462 SRR873454 SRR1138697 the ARGs, can be present in an ICE structure . At least one aadE essential gene from each module and several accessory genes were detected in the flanking region of the tet(W) gene in seven ruminal blaACl-1 rpkm

mef(A) 0 microbial genomes analyzed in this study, suggesting that ICEs could be mediating the horizontal transfer of tetracycline-resistance tet(32) 0.607 1.214 genes between ruminal bacteria. Some essential genes of integrative tet(40) 1.82 and conjugative elements were also found in other genomes of tet(J) 2.427 rumen bacteria harboring the tet(W) gene, but the genetic structure tet(O) 3.034 of these elements appeared incomplete. The structure of ICE_Rb- tet(Q) 3.641 tetW_07 was conserved in the genomes of ruminal bacteria, and the 4.248 fi tet(W) 4.854 presence of the tet(W) gene on the ICE was con rmed in the genome of Blautia schinkii DSM 10518 by PCR amplification and tet(37) 5.461 6.068 shotgun sequencing. ICEs are known to be frequently transferred vanR-C between distant bacterial taxa28, and our analysis suggest that these vanR-D genetic elements might have great mobility in the rumen ecosystem. If this is indeed the case, there is a potential threat of these genetic Beef cattle Dairy cattle Sheep elements to be transferred to bacterial pathogens in the GIT of Fig. 7 Expression levels (RPKM) of ARGs in ruminal metatranscriptomes. ruminants and into the environment. Beef and dairy cattle and sheep data sets were analyzed, and only ARGs Some groups of microorganisms, such as the Enterobacter- that were expressed in at least one metatranscriptome are represented. iaceae, are frequently associated with the horizontal transfer of Source data are provided as a Source Data file ARGs29,30. The enrichment of ARGs appears to be a common feature in the mobile resistome of Proteobacteria from both the animal and the human gut microbiome31. This may be asso- show evidence of positive selective pressure on the tet(W) gene, ciated with the frequent use of antibiotics targeting Enter- which may contribute to the selection of genetic variants of this obacteriaceae to control infections in the respiratory, urinary, gene, conferring ribosome protection against tetracycline anti- and GIT of humans and livestock, thus selecting for resistant biotics in ruminal bacteria. strains5. Our in silico results demonstrated that members of the Tetracyclines are broad-spectrum antibiotics that inhibit pro- Proteobacteria phylum that inhabit the rumen can harbor tein synthesis in susceptible bacteria21 and have been used in multiple ARGs, as exemplified by the genome of E. coli PA-3, a human and veterinary medicine to fight bacterial infections and ruminal isolate that contains five genes conferring resistance to to promote the growth of food-producing animals, improving four distinct antibiotics. Although generic Escherichia coli feed efficiency and animal performance22. The growth-promoting populations are often not so abundant in the rumen (<106 cells/ activity of tetracyclines in livestock was first reported in 1949 ml), cattle are considered a natural reservoir for pathogenic when feeds supplemented with dried Streptomyces aureofaciens strains of E. coli32,33. In addition, management and nutritional mash were administrated to poultry23. However, the long-term practices, such as feeding the animals a high-grain diet, can use of subtherapeutic doses of tetracyclines has been associated cause significant increases in ruminal and fecal populations of E. with increased levels of bacterial resistance in the GIT of food- coli and influence the abundance and diversity of ARGs in the producing animals24, which led countries such as the USA and ruminal resistome33,34. members of the European Union to ban the use of tetracyclines as Furthermore, a complete operon for vancomycin resistance growth promoters since 197525. The prevalence of tetracycline- (vanC) was found in the genomes of Enterococcus casseliflavus resistance genes among ruminal bacteria has a potential negative and in Enterococcus gallinarum SKF1, which makes these species impact in livestock health given that tetracyclines account for up intrinsically resistant to this antibiotic35. Vancomycin is one of to 24% of the antibiotics used to treat respiratory, uterine, or the last-option antibiotic used to treat infections caused by Gram- locomotion diseases in Europe26. positive bacteria36, but little is known about vancomycin resis- In our study, tet(W), tet(Q), and tet(O) were the main tance in ruminal bacteria. Our analyses predicted resistance to tetracycline-resistance genes identified by ResFinder, but tet(W) vancomycin in Clostridium, Lachnoclostridium, and Blautia, and was the most abundant gene, being distributed in 28 genomes of in vitro experiments confirmed these phenotypes in Blautia ruminal bacteria with high similarity of nucleotide sequences. schinkii DSM 10518, Lachnoclostridium aerotolerans DSM 5434, Since ResFinder is an online platform that uses whole-genome Clostridium intestinale DSM 6191, and Clostridium lundense sequencing data to identify acquired AMR genes in bacteria13, the DSM 17049. However, our transcriptomic analysis indicated tet(W) gene was selected for further analysis with an aim to that the expression of vancomycin-resistance genes in the investigate if it could be horizontally transferred among ruminal rumen microbiome is low. In addition, vancomycin has limited bacteria. use for therapeutic purposes in animals, and vancomycin analogs Ouranalysessuggestthatthetet(W) gene of several species of have been restricted as growth promoters in livestock36. ruminal bacteria are located on an ICE with a unique modular Overall, our in vitro assays confirmed 31.6% of the resistance pattern, not previously reported for tetracycline resistance phenotypes predicted by computational approaches. Nonetheless, (ICE_RbtetW_07). The organization of the ICE structure is char- 64.3% and 71.4% of the resistance phenotypes based on the ARGs acterized by a core region containing essential modules for con- predicted by ResFinder and ARG-ANNOT, respectively, were jugation, recombination, and regulation27. The conjugation confirmed in vitro. These results show concordance between machinery often contains relaxases and secretion systems to med- these different approaches, despite the fact that some predicted iate the transfer of DNA between donor and recipient cells. The ARGs might not be functional in vitro. In addition, our tran- recombination module usually contains genes encoding integrases, scriptomic analyses showed higher expression of tetracycline-

8 NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0 ARTICLE resistance genes and low expression of other classes or ARGs Phylogenetic analysis and distribution of ARGs. To evaluate if the ARGs (such as beta-lactams and aminoglycosides) in the rumen data grouped according to the evolutionary history of the microbial species, phyloge- sets, which also agreed with the resistance phenotype observed netic trees were reconstructed using the 16S rRNA gene sequences from the gen- omes analyzed in this study. The 16S rRNA gene sequences were obtained from the in vitro for the pure cultures of ruminal bacteria. Therefore, the Hungate1000 collection12. Next, all the sequences were organized in a single txt file results presented here, based on genomic, transcriptomic and and aligned using the RDP Release 11.5 aligner in the Ribosomal Database Project phenotypic data, increase the confidence that ARGs are dis- website (RDP, https://rdp.cme.msu.edu/)39. FastTree v2.1 (http://www. 40 tributed among different species of ruminal bacteria, and that microbesonline.org/fasttree/) was used to infer Approximately-Maximum- Likelihood phylogenetic trees using default settings. The generated output file these genes are expressed by members of the microbial commu- (.tree) was visualized and annotated on the Interactive Tree of Life (iTOL) interface nity in vivo. Nonetheless, additional studies will be needed to v4 (https://iTOL.embl.de/)41. evaluate the effects of animal ageing, production systems, and To evaluate sequence conservation of the ARGs that were identified in silico as nutritional practices, including antibiotic feeding, on the expres- described above, the sequences of the most frequent genes conferring resistance were extracted from the ruminal microbial genomes, and aligned as described sion, and potential transit of the ARGs detected in this study below. To extract the sequences of the resistance genes, all genomes were annotated across the GI tract of cattle. by Prokka v1.1242 using the Galaxy platform (minimum contig size of 200 and E- Taken together, our results provide insights that characterize value < 10–6). To obtain the location of the resistance genes in the bacterial the antibiotic resistance in species of ruminal bacteria. Our genomes, the ResFinder v2.1 database was aligned against the annotated genomes fi using the BLASTn tool on the Galaxy platform. Minimum query coverage and ndings demonstrate that ARGs, in particular the genes for tet- sequence identity were 60 and 70%, respectively, with a threshold E-value < 10–6. racycline resistance, are prevalent among ruminal bacteria and The sequence of the resistance genes identified in the previous alignment were subjected to positive selection pressure. Moreover, analysis of the extracted using the SAMtools software v1.9 (http://SAMtools.sourceforge.net/)43, flanking regions of the tet(W) genes provided evidence that this and the most abundant ARGs were aligned using MUSCLE version 3.8.3144. The gene can be disseminated horizontally through ICE transfer. phylogenetic reconstruction was based on the Maximum-Likelihood method using FastTree v2.1, and iTOL v4 graphical interface was used to represent the Given the fact that the rumen ecosystem also harbor a highly phylogenetic trees, as described above. dense and diverse microbiota that maintain complex ecological associations, and the capacity of microorganisms to exchange Genetic context of ARGs. To assess potential mechanisms of ARG mobility, DNA through direct (conjugation) and indirect mechanisms the genetic elements flanking the resistance genes in a scaffold were investigated (transformation, transduction), the GIT of ruminants should be using the software Artemis45. When these flanking elements were annotated as considered as a relevant source of antibiotic-resistance genes. The hypothetical protein, sequence function was predicted on the UNIPROT website evidence of a new ICE (ICE_RbtetW_07) carrying the tet(W) gene (http://www.uniprot.org/)46, using BLASTn, UniProtKB as the target database and default settings. Putative mobile genetic elements flanking the resistance genes were emphasizes this idea and presents a risk for both animal and searched using different computational tools. ISfinder (https://www-is.biotoul.fr/)47 human health, due to the capacity of these genetic elements to was used to screen for transposons and/or integrons, and phage sequences were disseminate into the environment. This hypothesis is also sup- confirmed using PHASTER (http://phaster.ca/)48. The genomes that harbored ported by the observation of non-phylogenetically related bacteria proteins with predicted plasmid functions near the resistance genes were further analyzed against the Enterobacteriaceae and Gram-positive database of Plas- harboring the same resistance genes, and the expression of the midFinder 2.0 (https://cge.cbs.dtu.dk/services/PlasmidFinder/)49, using a threshold same ARGs in ruminal metatranscriptomes data sets from beef for minimum sequence identity of 70% and minimum coverage of 60%. Finally, and dairy cattle and sheep. In addition, our results highlight the ICEberg 2.0 (http://db-mml.sjtu.edu.cn/ICEberg/)17 was used to identify ARGs importance of using multiple computational tools to search for carried by integrative and conjugative elements (ICE). Scaffolds harboring selected ARGs in genomic data and demonstrate the relevance to validate ARGs were used as query sequences (FASTA format) in the WU-BLAST2 search tool in ICEberg, using BLASTn as the search program and ICE nucleotide sequence these results in vitro and/or in vivo. as the sequence database. All the other parameters were kept as default settings.

Conservation and in vitro confirmation of a novel ICE. To evaluate the genetic Methods conservation of the integrative and conjugative elements identified in the genomes fi fi Collection of genomic data and identi cation of ARGs. The sequence les of 435 of ruminal bacteria, the nucleotide sequences of these ICEs were aligned using ruminal microbial genomes (425 bacteria and 10 archaea), from the Hungate1000 progressiveMAUVE50. In addition, a phylogenetic tree was reconstructed in MEGA project, were downloaded in FASTA format from the NCBI (National Center for X software using Muscle to compare the sequences of the novel ICE with other Biotechnology Information, http://www.ncbi.nlm.nih.gov/genome) and the JGI ICEs available in the ICEberg database. Phylogenetic reconstruction was performed (Joint Genome Institute, http://genome.jgi.doe.gov) websites (Supplementary using the Maximum-Likelihood method with 100 replicates. Data 3). Putative ARGs in the bacterial and archaeal genomes were predicted using To confirm the presence of a novel ICE in the genome of ruminal bacteria, PCR three bioinformatics tools (ResFinder v2.1, Resfams v1.2, and ARG-ANNOT Nt amplifications were performed using Blautia schinkii DSM 10518. Primers were fi V3, see below) that vary in sensitivity and speci city and apply distinct databases designed to amplify a 4794 pb DNA fragment that included the region from gene and search methods to detect putative ARGs in nucleotide sequences. virD4 to gene tet(W). Briefly, the bacteria was grown in anaerobic media at 39 °C We initially prospected the bacterial and archaeal genomes using the ResFinder for 72 h. Genomic DNA was extracted using the Qiagen DNA extraction kit 13 v2.1 database (https://cge.cbs.dtu.dk/services/ResFinder/) against all antibiotic (Manchester, UK). PCR amplification of the region of interest was carried out in classes available for analysis of acquired ARGs, which included aminoglycosides, 25 µl volumes as follows: 12.5 µl 2x MyTaq HS Red Mix, 1 µl each of 20 µM primers fl beta-lactams, colistin, uoroquinolones, fosfomycin, fusidic acid, glycopeptides, drawn in this study (RbtetW07_F: ATGAAAAAGCAGCTTGACATCAAAAAGC MLS (macrolides, lincosamides, and streptogramins), nitroimidazole, and RbtetW07_R: TTACATTACCTTCTGAAACATATGGCGC), 8 µl of nuclease- oxazolidinone, phenicols, rifampicin, sulfonamides, tetracyclines, and free water, and 2.5 µl of 50 ng template DNA. The cycling conditions were as trimethoprim. For homology-based screening, the minimum gene identity and follows: initial denaturation step at 95 °C for 60 s, 35 cycles of denaturation at 95 °C sequence length were set to 70 and 60%, respectively, in relation to the reference- for 15 s, annealing at 58 °C for 15 s, and extension at 72 °C for 3 min, and a final resistance gene. extension step at 72 °C for 5 min. The size and integrity of PCR products were The software Resfams v1.2 was used to search for conserved structural domains visualized and assessed by agarose gel electrophoresis using a 1% (w/v) agarose in 14 of antibiotic-resistance function in protein sequences of ruminal bacteria . The TAE buffer (40 mM Tris, pH 8.0, 20 mM acetic acid, 1 mM EDTA, BioRad Ltd., protein sequences of the analyzed genomes were downloaded in the FASTA format Hemel Hempstead, UK). The PCR product was subjected to shotgun sequencing from the NCBI or JGI websites and when these sequences were not found, the performed on the MiSeq with 300 cycles using the next-Flex Illumina workflow at online interface Prodigal v1.2 (http://compbio.ornl.gov/prodigal/server.html) was the Queen’s University of Belfast Genomics Core Technology Unit (Belfast, UK) to used to convert nucleotide to protein sequences using the Standard Bacteria/ confirm the presence of the predicted integrative and conjugative element. Quality Archaea translation table code. filtering of the reads was performed using Trimmomatic software v0.2751, and Genomes of ruminal microorganisms were also analyzed using the ARG- sequences were assembled using SPAdes 3.10.152. The sequence was aligned to the ANNOT database (http://en.mediterranee-infection.com/article.php? ICE coding sequence in the genome of Blautia schinkii DSM 10518, using Clustal = = 15 laref 283&titre arg-annot-) . Sequences of the ARGs in the ARG-ANNOT Omega53. database (ARG-ANNOT Nt V3) were used for similarity analysis against the ruminal microbial genomes using BLASTn37 on the Galaxy platform38. The alignment parameters were minimum sequence identity of 70%, sequence length Selective pressure on identified ARGs. Sequences of the most prevalent ARGs cutoff of 60%, and E-value < 10–6. detected in ruminal microbial genomes were selected to evaluate if these genes are

NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0 under selection pressure. Selective pressure on the ARGs was estimated by the 2. Ventola, C. L. The antibiotic resistance crisis: part 2: management strategies number of non-synonymous and synonymous substitutions (dN/dS) using the and new agents. P. T. 40, 344–352 (2015). software JCoDA54. JCoDA uses ClustalW to align the nucleotide sequences and 3. Elliott, K. A., Kenny, C. & Madan, J. A global treaty to reduce antimicrobial Phylogenetic Analysis using Maximum Likelihood (PALM) to calculate the dN/dS use in livestock. https://www.cgdev.org/sites/default/files/global-treaty-reduce- ratio. Analyses were performed using nucleotide sequences in the FASTA format as antimicrobial-use-livestock.pdf (2017). input sequences and the universal (default) genetic translation code. The alignment 4. Aarestrup, F. M. The livestock reservoir for antimicrobial resistance: a option was set to ClustalW, and phylogenetic trees were reconstructed by the personal view on changing patterns of risks, effects of interventions and the Maximum-Likelihood method with 100 replicates using protein sequences in the way forward. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 370,1–13 (2015). Phylip-sequential format. The dN/dS was calculated by site (amino acid) using the 5. Hitch, T. C. A., Thomas, B. J., Friedersdorff, J. C. A., Ougham, H. & Creevey, Bayes Empirical Bayes (BEB) model, which takes into account the uncertainties of C. J. Deep sequence analysis reveals the ovine rumen as a reservoir of 54 model parameters, avoiding the generation of false positives . The selective antibiotic resistance genes. Environ. Pollut. 235, 571–575 (2018). fi pressure on the antibiotic-resistance genes were con rmed by two neutrality tests, 6. Huws, S. A. et al. Addressing global ruminant agricultural challenges through 55 56 the Tajima test (performed in MEGAX ) and Fu and Li test (DnaSP v6 ). understanding the rumen microbiome: past, present, and future. Front. Microbiol. 9, 2161 (2018). In vitro analysis of the predicted resistance phenotypes. To evaluate if the 7. Rubino, F., Carberry, C., Kenny, D., McCabe, M. S. & Creevey, C. J. Divergent ARGs detected in the ruminal microbial genomes conferred the predicted resis- functional isoforms drive niche specialisation for nutrient acquisition and use tance phenotype in vivo, 26 cultures of ruminal bacteria that had their genomes in rumen microbiome. ISME J. 11, 932–944 (2017). analyzed for the presence of ARGs were tested in vitro to confirm these pheno- 8. Shoemaker, N. B., Wang, G. R. & Salyers, A. A. Evidence for natural transfer types. Cultures stored at −80 °C were activated twice in brain heart infusion (BHI) of a tetracycline resistance gene between bacteria from the human colon and broth at 39 °C for 24 h under anaerobic conditions before each experiment. For the bacteria from the bovine rumen. Appl. Environ. Microbiol. 58, 1313–1320 antibiotic susceptibility test, the optical density (OD600nm) of the cultures was (1992). − adjusted to obtain a cell count of ~108 CFU ml 1 (equivalent to 0.5 McFarland 9. Shterzer, N. & Mizrahi, I. The animal gut as a melting pot for horizontal gene scale). Subsequently, cultures were spread on the surface of solid BHI media using transfer. Can. J. Microbiol. 61, 603–605 (2015). swabs. E-test strips (Biomérieux, Basingstoke, UK; Fannin, Dublin, Ireland) of the 10. Toomey, N., Monaghan, A., Fanning, S. & Bolton, D. Transfer of antibiotic antibiotics to be tested were distributed on the surface of the inoculated medium, resistance marker genes between in model rumen and plant and the plates were incubated anaerobically at 39 °C for another 24 h. Only anti- environments. Appl. Environ. Microbiol. 75, 3146–3152 (2009). biotics to which resistance was predicted in silico were evaluated in vitro using pure 11. Cameron, A. & McAllister, T. A. Antimicrobial usage and resistance in beef cultures of ruminal bacteria. The minimum inhibitory concentration (MIC) of each production. J. Anim. Sci. Biotechnol. 7,1–22 (2016). antibiotic was determined at the intersection of the inhibition zone with the E-test 57 12. Seshadri, R. et al. Cultivation and sequencing of rumen microbiome members strips . All experiments were performed with three biological replicates. Reference from the Hungate1000 collection. Nat. Biotechnol. 36, 359–367 (2018). values from the European Committee on Antimicrobial Susceptibility Testing 13. Zankari, E. et al. Identification of acquired antimicrobial resistance genes. J. (EUCAST) breakpoint table were used to interpret the MIC results58. In the case of Antimicrob. Chemother. 67, 2640–2644 (2012). a few antibiotics where breakpoint values were not available for the tested species, 14. Gibson, M. K., Forsberg, K. J. & Dantas, G. Improved annotation of antibiotic MIC interpretations were based on breakpoint values published for the most clo- resistance determinants reveals microbial resistomes cluster by ecology. ISME sely related organism. J. 9, 207–216 (2015). 15. Gupta, S. K. et al. ARG-ANNOT, a new bioinformatic tool to discover Metatranscriptomic analyses. The metatranscriptome data sets were downloaded antibiotic resistance genes in bacterial genomes. Antimicrob. Agents in the FASTQ format from the NCBI Sequence Read Archive SRA (https://www. Chemother. 58, 212–220 (2014). 59 ncbi.nlm.nih.gov/sra) . Fifteen ruminal data sets representing cattle from distinct 16. El-Awady, R. et al. The role of eukaryotic and prokaryotic ABC transporter production systems (dairy and beef) and from sheep were selected for the analysis family in failure of chemotherapy. Front. Pharm. 7,1–15 (2016). (Supplementary Table 2). FastQC software v0.11.5 (http://www.bioinformatics. 17. Bi, D. et al. ICEberg: a web-based resource for integrative and conjugative babraham.ac.uk/projects/fastqc) was initially applied for quality control checks on elements found in Bacteria. Nucleic Acids Res. 40, D621–D626 (2012). fi raw sequence data using a Phred quality score of 20 to lter high-quality bases. 18. Sader, H. S. & Pignatari, A. C. E test: a novel technique for antimicrobial However, because most reads were discarded using this criteria, gene expression susceptibility testing. Sao Paulo Med. J. 112, 635–638 (1994). was evaluated using raw sequence data. 19. Karikari, A. B., Obiri-Danso, K., Frimpong, E. H. & Krogfelt, K. A. Antibiotic In this study, the most abundant group of ARGs identified in the ruminal resistance of Campylobacter recovered from faeces and carcasses of healthy microbial genomes was selected to verify their expression in the livestock. Biomed. Res. Int. 2017,1–9 (2017). metatranscriptomic data sets. These genes were predicted to encode resistance to 20. Muurinen, J. et al. Influence of manure application on the environmental aminoglycosides (aadE, ant(6)-Ia, aph(3’)-III, strA, strB), beta-lactams (cepA, resistome under finnish agricultural practice with restricted antibiotic use. blaACl−1, blaSED1, blaDHA-2, blaZ), macrolides (ermB, erm(G), mef(A), Isa(A), Environ. Sci. Technol. 51, 5989–5999 (2017). IsaC), tetracycline (tet(32), tet(40), tet(A), tetA(P), tet(B), tetB(P), tet(D), tet(J), tet (M), tet(O), tet(Q), tet(W), tet(37), and vancomycin (vanC, vanD, vanH-D, vanR- 21. Chopra, I. & Roberts, M. Tetracycline antibiotics: mode of action, applications, molecular biology, and epidemiology of bacterial resistance. C, vanR-D, vanS-C, vanS-D, vanT-C, vanX-D, vanXY-C, vanY-D, vanZ-F). The – Bowtie2-build tool was used to index the DNA sequences of the resistance genes, Microbiol. Mol. Biol. Rev. 65, 232 260 (2001). while Bowtie260 was employed to align these sequences to the metatranscriptomic 22. Daghrir, R. & Drogui, P. Tetracycline antibiotics in the environment: a review. – data sets. The expression level of each gene was calculated by the number of Environ. Chem. Lett. 11, 209 227 (2013). uniquely mapped reads per kilobase in a gene per million mappable reads 23. Stokstad, E. L. et al. The multiple nature of the animal protein factor. J. Biol. – (RPKM)61 using a cutoff value of 0.362. Chem. 180, 647 654 (1949). 24. Michalova, E., Novotna, P. & Schlegelova, J. Tetracyclines in veterinary medicine and bacterial resistance to them. Vet. Med. 49,79–100 (2004). Reporting summary . Further information on research design is available in 25. Schwarz, S. & Chaslus-Dancla, E. Use of antimicrobials in veterinary medicine the Nature Research Reporting Summary linked to this article. and mechanisms of resistance. Vet. Res. 32, 201–225 (2001). 26. De Briyne, N., Atkinson, J., Pokludova, L. & Borriello, S. P. Antibiotics used Data availability most commonly to treat animals in Europe. Vet. Rec. 175, 325 (2014). The data that support the findings of this study are included in the paper and/or 27. Cury, J., Touchon, M. & Rocha, E. P. C. Integrative and conjugative elements its Supplementary Information files. Data underlying Figs. 5, 7 and Supplementary and their hosts: composition, distribution and organization. Nucleic Acids Res. Figs. 1, 2, and 7 are provided as a Source Data file. 45, 8943–8956 (2017). 28. Cury, J., Oliveira, P. H., de la Cruz, F. & Rocha, E. P. C. Host range and genetic plasticity explain the co-existence of integrative and extrachromosomal mobile Received: 5 January 2019; Accepted: 21 October 2019; genetic elements. Mol. Biol. Evol. 35, 2230–2239 (2018). 29. Colavecchio, A., Cadieux, B., Lo, A. & Goodridge, L. D. Bacteriophages contribute to the spread of antibiotic resistance genes among foodborne pathogens of the Enterobacteriaceae family—a review. Front. Microbiol. 8, 1–13 (2017). 30. Rozwandowicz, M. et al. Plasmids carrying antimicrobial resistance genes in References Enterobacteriaceae. J. Antimicrob. Chemother. 73, 1121–1137 (2018). 1. O’Neill, J. Tackling drug-resistant infections globally: final report and 31. Hu, Y. et al. The bacterial mobile resistome transfer network connecting the recommendations. https://amr-review.org/sites/default/files/160518_Final% animal and human microbiomes. Appl. Environ. Microbiol. 82, 6672–6681 20paper_with%20cover.pdf (2016). (2016).

10 NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-13118-0 ARTICLE

32. Callaway, T. R., Elder, R. O., Keen, J. E., Anderson, R. C. & Nisbet, D. J. Forage 58. European Committee of Antimicrobial Susceptibility Testing (EUCAST). feeding to reduce preharvest Escherichia coli populations in cattle, a review. J. Breakpoint tables for interpretation of MICs and zone diameters. http://www. Dairy Sci. 86, 852–860 (2003). eucast.org (2017). 33. Diez-Gonzalez, F., Callaway, T. R., Kizoulis, M. G. & Russell, J. B. Grain 59. Leinonen, R., Sugawara, H. & Shumway, M. & International Nucleotide feeding and the dissemination of acid-resistant Escherichia coli from cattle. Sequence Database, C. The sequence read archive. Nucleic Acids Res. 39, Science 281, 1666–1668 (1998). D19–D21 (2011). 34. Auffret, M. D. et al. The rumen microbiome as a reservoir of antimicrobial 60. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. resistance and pathogenicity genes is directly affected by diet in beef cattle. Methods 9, 357–359 (2012). Microbiome 5,1–11 (2017). 61. Mortazavi, A., Williams, B. A., McCue, K., Schaeffer, L. & Wold, B. Mapping 35. Gold, H. S. Vancomycin-resistant enterococci: mechanisms and clinical and quantifying mammalian transcriptomes by RNA-Seq. Nat. Methods 5, observations. Clin. Infect. Dis. 33, 210–219 (2001). 621–628 (2008). 36. Wijesekara, P. N. K., Kumbukgolla, W. W., Jayaweera, J. & Rawat, D. Review 62. Warden, C. D., Yuan, Y.-C. & Wu, X. Optimal calculation of RNA-Seq fold- on usage of vancomycin in livestock and humans: maintaining its efficacy, change values. Int. J. Comput. Bioinfo. Silico Model 2, 285–292 (2013). prevention of resistance and alternative therapy. Vet. Sci. 4,1–11 (2017). 37. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local alignment search tool. J. Mol. Biol. 215, 403–410 (1990). Acknowledgements 38. Afgan, E. et al. The Galaxy platform for accessible, reproducible and collaborative This work has been supported by the Coordenação de Aperfeiçoamento de Pessoal de biomedical analyses: 2016 update. Nucleic Acids Res. 44,W3–W10 (2016). Nível Superior (CAPES; Brasília, Brazil), Fundação de Amparo a Pesquisa do Estado de 39. Cole, J. R. et al. Ribosomal database project: data and tools for high Minas Gerais (FAPEMIG; Belo Horizonte, Brazil), and the Conselho Nacional de throughput rRNA analysis. Nucleic Acids Res. 42, D633–D642 (2014). Desenvolvimento Científico e Tecnológico (CNPq; Brasília, Brazil). We are also grateful 40. Price, M. N., Dehal, P. S. & Arkin, A. P. FastTree 2 – approximately for RCUK Newton Institutional Link Funding (172629373). maximum-likelihood trees for large alignments. PloS ONE 5,1–10 (2010). 41. Letunic, I. & Bork, P. Interactive tree of life (iTOL) v3: an online tool for the Author contributions display and annotation of phylogenetic and other trees. Nucleic Acids Res. 44, Y.N.V.S., H.C.M., and S.A.H. conceived the project. Y.N.V.S. and A.J.S.M. completed the W242–W245 (2016). laboratory work under the supervision of H.C.M. and S.A.H. Y.N.V.S., M.F.S., and F.G.S. 42. Seemann, T. Prokka: rapid prokaryotic genome annotation. Bioinformatics 30, analyzed the data and generated figures. S.A.H. provided the anaerobic cultures analyzed 2068–2069 (2014). in the study. L.B.O. assisted Y.N.V.S. with the in vitro experiments. L.B.O., M.F.S., 43. Li, H. et al. The sequence alignment/Map format and SAMtools. and S.A.H. provided valuable ideas into the project from the time of conception. Y.N.V.S. Bioinformatics 25, 2078–2079 (2009). and H.C.M. wrote the paper with imput from all co-authors. 44. Edgar, R. C. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 32, 1792–1797 (2004). 45. Carver, T., Harris, S. R., Berriman, M., Parkhill, J. & McQuillan, J. A. Artemis: Competing interests an integrated platform for visualization and analysis of high-throughput The authors declare no competing interests. sequence-based experimental data. Bioinformatics 28, 464–469 (2012). 46. UniProt Consortium, T. UniProt: the universal protein knowledgebase. Additional information Nucleic Acids Res. 45, D158–D169 (2018). Supplementary information 47. Siguier, P., Perochon, J., Lestrade, L., Mahillon, J. & Chandler, M. ISfinder: the is available for this paper at https://doi.org/10.1038/s41467- reference centre for bacterial insertion sequences. Nucleic Acids Res. 34, 019-13118-0. D32–D36 (2006). Correspondence 48. Arndt, D. et al. PHASTER: a better, faster version of the PHAST phage search and requests for materials should be addressed to S.A.H. or H.C.M. tool. Nucleic Acids Res. 44, W16–W21 (2016). Peer review information 49. Carattoli, A. et al. In silico detection and typing of plasmids using Nature Communications thanks the anonymous reviewers for PlasmidFinder and plasmid multilocus sequence typing. Antimicrob. Agents their contribution to the peer review of this work. Peer reviewer reports are available. Chemother. 58, 3895–3903 (2014). Reprints and permission information 50. Darling, A. C. E., Mau, B., Blattner, F. R. & Perna, N. T. Mauve: multiple is available at http://www.nature.com/reprints alignment of conserved genomic sequence with rearrangements. Genome Res. – Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 14, 1394 1403 (2004). fi 51. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for published maps and institutional af liations. Illumina sequence data. Bioinformatics 30, 2114–2120 (2014). 52. Bankevich, A. et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012). Open Access This article is licensed under a Creative Commons 53. Sievers, F. & Higgins, D. G. Clustal Omega, accurate alignment of very large Attribution 4.0 International License, which permits use, sharing, numbers of sequences. Methods Mol. Biol. 1079, 105–116 (2014). adaptation, distribution and reproduction in any medium or format, as long as you give 54. Steinway, S. N., Dannenfelser, R., Laucius, C. D., Hayes, J. E. & Nayak, S. appropriate credit to the original author(s) and the source, provide a link to the Creative JCoDA: a tool for detecting evolutionary selection. BMC Bioinforma. 11,1–9 Commons license, and indicate if changes were made. The images or other third party (2010). material in this article are included in the article’s Creative Commons license, unless 55. Kumar, S., Stecher, G., Li, M., Knyaz, C. & Tamura, K. MEGA X: molecular indicated otherwise in a credit line to the material. If material is not included in the evolutionary genetics analysis across computing platforms. Mol. Biol. Evol. 35, article’s Creative Commons license and your intended use is not permitted by statutory 1547–1549 (2018). regulation or exceeds the permitted use, you will need to obtain permission directly from 56. Rozas, J. et al. DnaSP 6: DNA sequence polymorphism analysis of large data the copyright holder. To view a copy of this license, visit http://creativecommons.org/ – sets. Mol. Biol. Evol. 34, 3299 3302 (2017). licenses/by/4.0/. 57. Nagy, E., Justesen, U. S., Eitel, Z., Urban, E. & Infection, E. S. G. O. A. Development of EUCAST disk diffusion method for susceptibility testing of the Bacteroides fragilis group isolates. Anaerobe 31,65–71 (2015). © The Author(s) 2019

NATURE COMMUNICATIONS | (2019) 10:5252 | https://doi.org/10.1038/s41467-019-13118-0 | www.nature.com/naturecommunications 11